Perception

PERCEPTION Philosophical Issues, 7, 2 996 PHILOSOPHICAL ISSUES Edited by Enrique Villanueva (Universidad Nacional...

Author: Enrique Villanueva

152 downloads 2533 Views 6MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

PERCEPTION

Philosophical Issues, 7, 2 996

PHILOSOPHICAL ISSUES Edited by Enrique Villanueva

(Universidad Nacional Autonoma de Mdxico)

EDITORIAL ADVISORY BOARD Ned Block (Massachusetts Institute of Technology)

Paul Boghossian (New York University)

Jerry Fodor (Rutgers University)

Richard Foley (Rutgers University)

James Higginbotham (University of Oxford)

Jaegwon Kim (Brown University)

Brian Loar (Rutgers University)

Christopher Peacocke (University of Oxford)

Sydney Shoemaker (Cornell University)

Ernest Sosa (Brown University)

James Tomberlin (California State University, Northridge)

Previously published volumes:

CONSCIOUSNESS

(Philosophical Issues, 1, 1991)

RATIONALITY IN EPISTEMOLOGY


SCIENCE AND KNOWLEDGE


NATURALISM AND NORMATIVITY


TRUTH AND RATIONALITY


CONTENT


Forthcoming volumes: TRUTH


Philosophical Issues, 7, 1996

PERCEPTION

edited by

Enrique Villanueva SOCIEDADFILOSOFICA IBEROAMERICANA

Ridgeview Publishing Company

Atascadero, California

Copyright O 1996 by Enrique Villanueva All rights reserved. No part of this book may be reproduced or utilized in any form or by any means, electrical or mechanical, including photocopying, recording or by any informational storage or retrieval system, without written permission from the copyright owner.

Paper text: ISBN 0-924922-25-7 Cloth (library edition): ISBN 0-924922-75-3

The typesetting was done by Jose Luis Olivares.

Published in the United States of America by Ridgeview Publishing Company Box 686 Atascadero, California 93423 Printed in the United States of America by Thomson-Shore, Inc.

Contents

1

2 3

4

5 6

Preface Enrique Villanueva Explaining Objective Color in Terms of Subjective

Experience

Gilbert H a n a n Mental Paint and Mental Latex Ned Block Orgasms Again Michael Tye Colors, Subjective Relations and Qualia Sydney Shoemaker Is Color Psychological or Biological? Or Both? Ernest Sosa Qualia and Color Concepts Gilbert H a n a n

7 Layered Perceptual Representation William Lycan 8 On a Defense of the Hegemony of Representation Robert Stalnaker 9 Perception and Possibilia James Tomberlin 10 Perceptual Experience Is a Many-Layered Thing Michael Tye 11 Replies to Tomberlin, Tye, Stalnaker and Block William L ycan 12 Phenomenal Externalism

Fred Dretske 13 Dretske's Qualia Externalism

Jaegwon K i m

vii

14 Comnient on Dretske Paul Horwich 15 Dretske on Phenomenal Externalism John Biro 16 Reply to Commentators Fred Dretske

17 Is the External World Invisible? Mark Johnston 185 18 Visible Properties of Human Interest Only Allan Gibbard 199 19 Getting Acquainted with Perception David Sosa 209 20 Would more acquaintance with the external world relieve epistemic anxiety? Enrique Vil lanueva 215 21 A Mind-Body Problem at the Surface of Objects Mark Johnston 219 22 Precis of Mind and World John McDowell 23 Perception and Rational Constraint: McDowell's Mind and World Robert Brandom 24 Spin Control. Comment on John McDowell's Mind and World Alex Byrne 25 McDowell's Direct Realism and Platonic Naturalism Roger Gibson 26 Reply to Gibson, Byrne, and Brandom John McDowell 27 Molineux's Question John Campbell 28 Comments on John Campbell, "Molineux's Question" Brian Loar 29 Shape Properties and Perception Kirk Ludwig 30 Shape Properties, Experience of Shape and Shape Concepts John Campbell

Contributors

231 241 261 2 75 283 301 319 325 351 365

Preface The papers in this volume are concerned with issues on perceptual representation, functionalism, qualia, phenomenal externalism, acquaintance, realism, naturalism, rationality, shape and perceptual properties and other subtopics. Those issues spin on the frontiers of Metaphysics, Philosophy of Mind and Epistemology. The papers in this volume were presented at the Eighth Annual Philosophy Conference of our Sociedad Filosofica Ibero Americana, SOFIA, held in Cancun, Mkxico, June 19-21. The volume reflects the order and structure of that Conference. It is composed of six symposia each with a speaker followed by a number of comments and ending with the speaker's replies. This new structure suits better the purpose of getting a closer dialogue, one that brings about warm memories of Classical Greece. We will keep tailoring a better structure to fit adequately the excellence of those participating. In our Society's name I want to express our gratitude to Lic Javier Barros Valero form Subsecretaria de Educaci6n Superior e Investigaci6n Cientifica (through anexo 94-01-09-160-212) for continuous support that made possible to hold our annual conference even in the face of financial stress. I want to express again my gratitude both to Ernie Sosa and Jim Tomberlin by their generous assistance through the organisation of this conference. Lourdes Valdivia again offered invaluable help before and through the celebration of the conference. Lic Gerardo Osuna offered kind support while in Cancun City. Once more I want to express my gratitude to my university the Universidad Nacional Aut6noma de Mkxico who provided a sabbatical leave and support through Direcci6n de Asuntos del Personal Acadkmico, Dr Josk Luis Boldu, Director. For a second year I want to thank the P h D Program in Philosophy at the Graduate Center, at CUNY, Chair Mr Richard Mendelsohn

who generously provided a room from where I kept organising this Conference. Thanks also to Mr Peter Klein Chair at the Department of Philosophy, at Rv.tgers, for kindly renewing my status as a visiting fellow. Thanks to Josk Luis Olivares who did the typesetting again. Enrique Vzllanueva Tepoztlan, January 1996

PHILOSOPHICAL ISSUES,7 Perception, 1995

Explaining Objective Color in Terms of Subjective Reactions Gilbert Harman

I am concerned with attempts to explain objective color in terms of subjective reactions so I had better begin by saying what I mean by "objective color" and what I mean by "subjective reactions". By "objective color" I mean the color of an object, in a very broad sense of "object" that includes not only apples and tables, but also the sky, a flame, a shadow, and anything else that has color. So, objective color in this sense includes the red of an apple, the blue of the sky, the yellow of a flame, the purple cast of a shadow, and so forth. By "subjective reactions" I mean a normally sighted perceiver's subjective impressions of color: how color looks. A person blind from birth might learn that objects have colors and might in some sense have subjective reactions to color, but not the sort of subjective reactions I mean. I will be concerned both with the sort of explanation of color and the relevant subjective reactions to color that is available to normally sighted perceivers and with the sort of explanation that is available to others, including those who cannot have the relevant subjective reactions themselves.

1 Why Objective Color Should Be Explained in Terms of Subjective Reactions hIany salient facts about color cannot be explained purely in terms of properties of the surfaces of colored objects. Lie need also to appeal to the biology and psychology of color perception. These facts include red's being closer in color to blue than to green, even though the frequency of pure red light is farther from that of pure blue than from pure green. Related to that fact is the circular structure of hues, as opposed to the linear structure of relevant light frequencies. There is also the way in which colors can be organized in terms of three polar contrasts, white-black, red-green, and blue-yellow. These aspects of color are due to facts about the biology and psychology of color perception rather than to facts about the structures of surfaces. Shepard (1992, 1993) offers an evolutionary explanation of the biological and psychological facts by noting that natural illumination from the sun varies in three independent respects: (1) in amount of total overall illumination, (2) in relative amount of longer red wavelengths (depending on the sun's angle), and (3) in relative amount of shorter blue wavelengths (depending on whether illumination is directly from the sun or indirect from light scattered by the atmosphere). (If the red wavelengths are removed from sunlight, the remaining wavelengths center on green. If the blue wavelengths are removed instead, the remaining wavelengths center on yellow.) Given these sorts of variation in natural illumination, a visual system structured like ours, in which light is analyzed in terms of white versus black, red versus green, and yellow versus blue, will be able to achieve a kind of constancy in colors attributed to objects, a color constancy that Shepard sees as having evolutionary benefits. Shepard's evolutionary account contrasts with the suggestion in Dennett (1991) that human color perception and the colors of natural objects have evolved toget her. But Shepard's proposal explains facts of color perception not accounted for by Dennett's suggestion. In what follows, I will assume Shepard is basically right. In any event, it would seem that we have to explain facts about objective color in terms of facts about perceivers (Gold, 1993). One simple way to do so identifies a n object's being a particular color C with its tendency to be perceived as C by normal observers viewing it under standard lighting conditions. Various complications arise here, e.g., concerning chameleon's that change color when looked at (Johnston, 1992). I want to disregard those (significant)

issues in order to try to say more about the relevant subjective reaction, "perceiving something as C".

2

Color Sensations

Some authors call the relevant subjective reactions "color sensations". A normal viewer's perception of a red object in adequate lighting provides the viewer with "red color sensations". Some authors (Shoemaker, 1981; Peacocke, 1983) use the term "color qualia" in much the same way that other writers use the term "color sensations". I will argue below (section 4) that talk of color sensations is misleading in important respects, but let me use that terminology for the time being. In these terms, a blind person does not in the normal way obtain color sensations from the perception of objects. Someone blind from birth may never have experienced color sensations. Such a person would normally not know what it is like to have such sensations. Red color sensations are not red in the same sense in which red apples are red. We might say that red apples are red in the sense that they tend to produce certain reactions when viewed by perceivers. But red color sensations are not red in that sense. Red sensations cannot be viewed and they are (supposed to be) the relevant reactions, not the causes of the reactions. To avoid possible confusion, some authors use a symbolism that distinguishes these senses of the word red, for example, distinguishing the word red from the word red' and saying we have red' color sensations rather than red color sensations (Peacocke, 1983). One problem is to say how these senses are related.

It may seem that the most obvious way to relate the two senses of color terms along the lines of the suggested reduction of objective color to subjective reactions is to try to define the colors of objects in terms of the color sensations they produce in observers: An object is red if and only if perception of it would give normal perceivers red' sensations under standard viewing conditions. A number of issues arise here. What makes a perceiver a normal perceiver? What determines standard viewing conditions? Circularity must be avoided. A normal perceiver cannot be defined as one who gets the right sensations from colored objects, nor can

standard viewing conditions be defined as those in which normal perceivers get the right sensations. One possible approach simply asserts that there are objective criteria of normalcy N and objective criteria of standardness S such that an object has color C if and only if perception of it by perceivers who are N in conditions that are S would produce C color sensations. Let us suppose for the sake of argument that such criteria exist. (A full account would have to investigate the criteria S and N.1

A further issue concerns the nature of color sensations. What are they and what makes a sensation that sort of color sensation that it is? Many authors (e.g., Nagel, 1974) believe that no purely scientific account of color sensations is possible. In their view, the essence of such sensations is precisely their subjective "qualitative" character. They believe that there is no way to describe this qualitative character in purely scientific terms so that it would be fully understood by someone who had never experienced the sensation first hand; someone who has never experienced a red' sensation cannot know what it is to have such a sensation. In their view, the notion of a red' sensation is the notion of a sensation "like this", where "this" refers to a sensation that one is actually experiencing or imagining. Since in this view the redness of an object is its power to produce red' sensations in perceivers, it follows that someone who has never experienced a red' sensation cannot fully understand what it is for an object to be red. A red object is an object with the power to produce sensations "like this" in perceivers. Someone who never experiences a red' sensation is never in a position to be able to identify red objects as objects with the power to produce sensations "like this". I postpone discussion of what conception of color and color sensations might be available to someone who has never had such experiences.

It is interesting to consider the possibility that different people might have relevantly different sorts of sensations when perceiving objects that they call "red".

This possibility may seem quite likely, (1) if the relevant aspect of a sensation is its intrinsic qualitative character, (2) if the intrinsic qualitative character of a sensation depends on the exact nature of the underlying physical events in the brain giving rise to the sensation, and (3) if there are differences in these underlying physical events from one brain to the next (Block, 1990, pp. 56-57). The suggested analysis of objective color in terms of sensations implies that, if there are relevant differences in people's color sensations, then different people have difference concepts of the colors of objects and do not mean the same thing by their color terms. They do not mean the same thing by "red", "green", etc. even though they use the terms in exactly the same way of exactly the same objects, at least as far as their outer usage is concerned. For George, an object is red if and only if it has the relevant power to produce sensations "like this" in normal perceivers, where George refers to the kind of sensation that he gets from viewing red objects. For Mary, an object is red if and only if it has the relevant power to produce sensations "like this" in normal perceivers, where Mary refers to the kind of sensation that she gets from viewing red objects. If George and Mary get different sorts of sensations from viewing red objects, they mean different things when they say that an apple is "red", in this view. This is actually a pretty strange consequence: that people might mean different things by their words even though they use them in the same way with respect to objects in the world. But a further consequence is even stranger. Given the hypothesis that people do get different kinds of color sensations from objects they call "red", the suggested analysis implies that no objects have any colors (Block, 1990, p. 56)! According to the analysis, an object is red if and only if it has the power to produce sensations "like this" in normal perceivers viewing the object in standard viewing conditions. But, by hypothesis, no object has that power. A ripe tomato may have the power to produce sensations '(like this" in me under those conditions, but it does not have the power t o produce sensations of that sort in all other normal perceivers viewing the object in standard conditions. By hypothesis, viewing a ripe tomato produces different kinds of sensations in different otherwise normal perceivers. So, the analysis we are considering implies that a ripe tomato has no color, given that hypothesis. So we have two absurd results. First, different people mean different things by their color terminology even when they use the terminology in the same way of external objects. Second, when people

use color terminology to say that eternal objects are "red", "blue", or whatever, what they say is always false. To avoid such absurd results, we must either abandon the suggested analysis of objective color or rule out the possibility that different perceivers get different sorts of color sensations from viewing objects called "red".

3

Functional Definitions of Color Sensations

The discussion so far has assumed that it is possible to fix on a kind of sensation by attending to it and intending to include in that kind of sensation anything "like this". But a sensation that occurs on a particular occasion can be classified in infinitely many ways and is therefore an instance of many different kinds of sensation. The sensation itself does not determine a single kind or type of sensation. Saying that red objects are objects with the power to produce sensations "like this" is not yet to say what type of sensations red objects have the power to produce, since there are infinitely many different ways in which sensations can be "like this". It is necessary also to say in what respect sensations have to be "like this" in order to count as "red' sensations". It would be wrong to say that a sensation has to share every aspect of "this sensation" if it is to be "like this", for then no other occurrences would count. There are always some differences among sensations: they occur at different times, have different causes, different effects, occur to different people, and so forth. Not all these differences are important. For example, the fact that "this sensation" is a sensation of mine distinguishes it from all sensations of other people, but that had better be irrelevant to its being a red' sensation if anyone else is to be able to have a red' sensation. What is needed is a way of classifying sensations so that the sensations normal observers have on viewing a given color normally fall under the same classification, even if, according to some other way of classifying sensations, people have different sorts of sensations from viewing a given color. One approach to solving the problem of type specification appeals to a "functional definition" of the relevant type of sensation in terms of typical causes and effects (Armstrong, 1968; Lewis, 1966). For example, the sensations classified as pains are those that are typically caused by tissue damage or extremes of pressure or heat applied to some location in one's body and that typically have as effects the

belief that something undesirable is occurring at the relevant bodily location and the desire to be free of the occurrence. A first stab at a functional definition of color sensations might suppose that they are typically caused by the perception of appropriately colored objects and that their typical effects are beliefs that perceived objects have the appropriate colors. So, a red' sensation would be a sensation that is typically caused by the perception of red things and that typically leads to the belief that one is perceiving something red. Various worries can be raised about this account and more needs to be said. If it could be made to work, the suggested account would avoid some of the problems raised about the previous account. The account would not have to suppose that differences in the brain events that underlie color sensations mean that different otherwise normal perceivers have relevantly different sensations from the perception of the same objects. If you and I are both normal color perceivers, we will both receive red' sensations from the perception of ripe red tomatoes. What we mean by "red object" (namely, "object with the power to produce red' sensations.. . ") will be the same if we use the word in the same way of external objects, even if our red' color sensations differ in their detailed neurophysiological realizations. However, when a functional account of color sensations is combined with an explanation of objective color in terms of color sensations, the resulting account of objective color is circular. It reduces to the claim that red objects are those that produce the sort of sensation that red objects produce. This is not only to explain the notion of objective color in terms of itself but to do so in a way that is almost completely empty.

4

Complication: There Are No Color Sensations

Before addressing the problem of circularity, it is necessary to clear up a point we have been so far ignoring, namely, that it is wrong to describe color impressions as color "sensations". Normally, we use the term "sensation" for bodily feelings. Usually sensations have a more or less definite location in one's body -a headache, a pain in one's foot, butterflies in the stomach, etc. There are also other cases, such as a sensation of dizziness. But the perception of color does not normally involve sensations in any ordinary sense of the term "sensation". When someone literally has visual sensations, they are pains or other feelings in the eye,

resulting from overly bright scenes, perhaps, or itching from allergies or minor eye injuries. Color perception does not normally involve such sensations. On seeing what appears to be a ripe tomato, one does not feel a sensation of red in one's eye, nor is there literally a sensation or feeling at the location at which the tomato looks red. How then should we think of perceptual experience, if not as involving visual sensations?

One important point is that perceptual experience has a certain presentational or representational character, presenting or representing the environment in a certain way. When it looks to you as if you are seeing a ripe tomato, your perceptual experience presents or represents the environment as containing a red and roughly spherical object located at a certain distance and orientation "from here". When you think about visual representation, it is very important to distinguish (A) qualities that experience represents the environment as having from (B) qualities of experience by virtue of which it serves as a representation of the environment. When you see a ripe tomato your visual experience represents something as red. The redness is represented as a feature of the tomato, not a feature of your experience. Does your experience represent this redness by being itself red at a relevant place, in the way that a painting of a ripe tomato might represent the redness of the tomato with some red paint on the appropriate place on the canvas? No. That is not how visual representation works. Does your experience represent this redness by having a t some place some quality other than redness, a quality of red'ness, which serves to represent the redness of the tomato in some other way, different from the way in which a painting might use red paint to represent a tomato? Well, who knows? You have no conscious access to the qualities of your experience by which it represents the redness of the tomato. You are aware of the redness of the represented tomato. You are not and cannot become consciously aware of the mental "paint" by virtue of which your experience represents the red tomato. It follows that your concept of a red object cannot be analyzed into your concept of a red' experience, meaning the specific quality that your perceptual experience has in order to represent objective redness, because you have no such concept of a red' experience. You

have no idea what specific quality of your perceptual experience is used to represent objective redness. You only have the concept of objective redness!

In fact, your color concepts are almost certainly basic and not analyzable in causal terms. You perceive colors as simple primitive features of the world, not as dispositions or complexes of other causal features. (Maybe some color concepts like orange can be analyzed in terms of concepts of primary colors, like red and yellow. And maybe some color concepts like brown can be analyzed in terms of hue, brightness, and saturation. I am not concerned with such internal analyses. I am concerned only with analyses of color in external terms, especially causal terms.) Now, a scientific explanation may involve an analysis of something without claiming to be analyzing an ordinary concept. For example, when a scientific explanation of facts about the circle of hues treats color as a tendency to produce certain responses in perceivers, it is not offering that analysis of color as an account of the perceptual concept of color. The perceptual concept of color can be quite simple even if color itself is a complex phenomenon. Of course, it would be useful to give an account of what it is to have a basic perceptual concept like the concept of redness, an account that might even be understood by someone lacking that basic perceptual concept. A congenitally blind person can understand that a normal color perceiver might have a basic perceptual concept of redness, for example. And normal color perceivers can understand that there may be animals or other alien creatures with basic perceptual concepts that humans do not have. To this end, we might try to provide a functionalist account of what it is for perceptual experience to have a given perceptual concept. Now, in general, perceptual experience represents the environment in ways that enable a perceiver to negotiate paths among objects, to locate desired things and to avoid undesired things. Normally and for the most part, a perceiver accepts his or her (or its) perceptual representation, believing that things are as they appear, although the strong disposition to acceptance can be inhibited on special occasions. The perceptual concept of red figures as part of the perceptual experience of red objects, enabling a perceiver to identify and reidentify objects as red. In other words, if a perceptual concept is a

concept Q such that one has perceptual experiences of something being Q, then (roughly speaking) we can say that the concept Q is the concept of redness if perception of red things tends to produce experiences of something being Q. But we must be careful to avoid circularity. Recall that we have been supposing that red is a tendency to produce perceptual reactions of a certain sort. We have seen that it is incorrect to describe the relevant reactions as sensations. Suppose then that we take the relevant reactions to be experiences with a certain representational or presentational content. If the relevant representational or presentational content is then identified functionally, we seem to be identifying rediness as a property R, where R is a tendency to produce experiences that represent something as Q, where Q is the concept produced by perception of R things. That characterization is circular and does not distinguish red from green, for example. Sosa (1990) points out that we can avoid circularity if we use the normally sighted person's primitive perceptual concept of red objects in our account. Then we can say that something is red if and only if it has a property R, where R is a tendency to produce experiences that represent something as Q, where Q is the concept produced by perception of red things, where here we are using the primitive perceptual concept red. Of course, a person lacking that perceptual concept of red could not avoid circularity in the way Sosa suggests. Nor could normal human perceivers use that approach to provide noncircular accounts of animal or alien perceptual concepts that do not correspond to human perceptual concepts. One noncircular account that might be useful to those lacking the relevant perceptual concepts would identify color in terms of biological mechanisms of color perception, perhaps via the evolutionary reasons for those mechanisms. In this view, for something to be red is for it to have a tendency to have a certain specific complex affect on a normal perceiver's sensory apparatus, in ways described by the scientific theory of color.

5 The Inverted Spectrum Supposing this last account can be made to work up to a point, one might still worry that it seems to leave out an important aspect of color perception. As Block (1990) puts the objection, the functional account of what it is to have a concept of red captures the "intentional content" of the concept, but not its "qualitative content".

Qualitative content is what we imagine to be different when we imagine that one person perceives colors in a way that differs from the way in which another person perceives them. We seem even to be able to imagine the possibility of an inverted spectrum in the sense that the way things look to one of two otherwise normal color perceivers, George, might be qualitatively hue inverted with respect to the way things look to the other, Mary (Shoemaker, 1981). It is not that (we imagine that) what looks red to George looks green to Mary. A given object looks green to both or red to both. That is, the colors their experiences represent the environment as having are the same. The imagined difficulty is that "what it is like" for George to see something as red is different from "what it is like" for Mary to see something as red. The "what is it like" terminology comes from Nagel (1974). I am not convinced that this particular appeal to "what is it like" for a particular person to see something as green is in the end really intelligible. (I discuss what this might mean in section 7, below.) But let us assume that it is intelligible in order to explore the idea. As I have argued, the difference in "what it is like" for George and Mary to see something as red cannot be a difference in visual sensations, so it has to be a difference in how George and Mary perceive objects to be. And it cannot be a difference in what colors they perceive objects to have, because they both count as correctly perceiving the colors of objects. The difference between them cannot be at that level. So, the difference must be a difference in other qualities that objects are presented or represented as having. What we are imagining, then, seems to be something like this (Shoemaker, 1994). When George sees a red apple, his perceptual experience represents it as being Q. His experience also represents the apple as being red. Furthermore, the fact that his experience represents the apple as being Q makes it true in present circumstances that his experience represents the apple as being red. It seems we can imagine other circumstances in which neuronal connections leading from George's retina to his visual cortex were switched before birth in such a way that later, when his experience represents something as being Q, that constitutes representing it as being green, because, in these imagined circumstances, the perception of green things normally leads George to have perceptual experiences of those things as Q. When we imagine normal perceivers like George and Mary with inverted spectra, we are then imagining something like this: A red object looks Q to George and T to Mary. A green object looks T to George and Q to Mary. An object's looking Q to George counts as

its looking red to George. An object's looking Q to Mary counts as its looking green to Mary. This may seem odd, so let me briefly review what led to this seemingly strange idea. We want to describe a case in which two people have inverted spectra with respect to each other. The difference between them has to be a difference in what they experience, but it cannot be a difference in properties they perceive their experience to have, because the relevant properties are perceived as properties of objects in the environment. The difference cannot be a difference in the colors they perceive these objects to be, because we are assuming that, as normal color perceivers, they attribute the same colors to external objects. So, it has to be a difference in other properties objects are experienced as having, properties we can identify as Q and T.

6 Worries about Inverted Spectra I am not sure that the imagined possibility of inverted spectra is really coherent. (Please note: I a m not sure.) A red object supposedly looks Q t o George and T to Mary. It would seem that an object cannot be both Q and T in the same place at the same time in the same way. That would be like an object's being both red and green at the same place and time in the same way. But then, either George's experience or Mary's experience or both of their experiences must be in some respect nonveridical, incorrectly representing the object seen. Shoemaker (1994) observes that the best way to avoid this result is to suppose that the properties Q and T are radically relational, so that something can be Q to one person without being Q to another. Q and T would be incompatible only in the sense that an object cannot be both Q and T to the same person at the same place at the same time in the same way. On the other hand, an object could be Q to George and T to Mary a t the same place at the same time in the same way. What is it for an object to be Q to a given person? Shoemaker (1994) mentions two possibilities. First, it might be that an object is Q to a given person S if and only if S's perceptual experience currently represents the object as Q. Second, it might be that an object is Q to a given person S if and only if the object has a tendency to provide S with perceptual experiences representing that object as Q. In the first case, objects are Q to S if and only if S is experiencing them as Q. In the second case, objects can be Q to S even if S is not

currently experiencing them if they are such as to produce relevant experiences under the right conditions.' But either of these possibilities involves a serious circularity. In order to understand what the concept Q is, we need to understand what objects are Q to someone, but in order to understand what objects are Q to someone, we need to understand what the concept Q is.2 Recall Sosa's point, noted on page 10 above, that we might use the perceptual concept of color possessed by normal color perceivers to give a key part of a functional account of what it is to have such a concept: the perceptual concept of red is activated in perceptual experiences produced by the perception of red objects. Could we use the same idea here --experiences involving the concept Q are produced by perception of Q objects? That explanation would be satisfactory only if one had a firm grasp of the concept Q. But I do not find that I have a firm grasp of that concept. If it is suggested that we try to break out of this circle as I suggested we might break out of our earlier circle with respect to actual color terms, like "red" and "green", by appeal to some tendency objects have to affect S's perceptual mechanisms, I find myself at a loss to know what aspects of perceptual mechanisms would be relevant. Settling on something, for example, certain events in the visual cortex would seem simply to let the same problem arise all over again. For surely we can imagine even molecule for molecule identical people with the same events occurring in their visual cortex having inverted spectra with respect to each other. That seems to be just as imaginable as the previous case. But this suggests the "problem" is really a pseudo-problem. This makes me doubt that there is a concept Q of the claimed sort and so doubt that I have the relevant grasp of "what it is like for so and so to see red" that would allow me to suppose that red things might look different to different, otherwise normal, color perceivers. 'I have oversimplied. Shoemaker's actual account supposes t h a t a n experience has a certain intrinsic phenomal feature x t h a t is responsible for its representing something as Q. For something t o be Q is for it t o be such as t o produce experiences with feature x. 'Shoemaker's attempted way out of this circle is t o say, for example, t h a t experiences of something as Q (i.e., experiences with feature x) are those experiences t h a t "are phenomally like those I have when I see a ripe tomato". T h a t would help only if we already had t h e sort of account of interpersonal phenomenal similarity t h a t would enable us t o make sense of interpersonal inverted spectrums. B u t we are in the process of trying t o develop such a n account. So t h a t account would be circular if we adopted Shoemaker's suggestion.

But let me try to say more about "what it is like" to have an experience of a certain sort.

7 What Is It Like To See Red? Terniinology can become confusing here because different people seem to use similar terminology in different ways and people often use a variety of terminology. For example, philosophers talk about "what it is like" to have a given experience, about a n experience's "phenomenological charecter" , and about "qualia" , as if these are different ways of getting at the same thing. In fact, at least two different issues are involved. Let me explain by citing two different ways in which the term "qualia" has been used. First, qualia are sometimes taken to be experienced qualities of a mental experience, those qualities by virtue of which one's experiences represent what they represent (when they represent things), the mental paint of one's picture of the environment, one's mental sense-data. Philosophers who use the term "qualia" in this sense tend to hold that not all mental experiences involve qualia. They take qualia to be involved in perception and sensation but not always in relatively abstract beliefs and thoughts. In this sense of "qualia", it is at least a matter of controversy whether all experiences involve qualia. O n the other hand, qualia are sometimes identified with what it is like to have a given experience and it is supposed to be relatively obvious that all mental experiences involve qualia. Even with respect to a relatively abstract judgment, there is something that it is like to have that judgment -some qualitative character in this second sense. Now, as I have indicated already, I am strongly inclined to deny that there are qualia ill the first sense, the mental paint or sense datum sense. In perception, all qualities of which we are aware seem to be presented to us as qualities of perceived things, external objects for the most part. Introspection does not support the claim that we are aware of mental paint. And, although arguments can be given for supposing that despite appearances we are aware of mental paint, these arguments seem to be uniformly fallacious through confusions over intentionality (Harman, 1990a). So I see no reason to suppose that we are aware of mental paint. With respect to qualia in sense (2), what it is like to have a given experience, I agree with Nagel (1974) that there is a distinctive kind of understanding that consists in finding a n equivalent in one's own

case. That is "knowing what it is like to have that experience". I have compared that sort of understanding with knowing what an expression used by someone else means. One understands it to the extend that one finds an equivalent expression in one's own language or by learning how to use the expression oneself. Even if use determines meaning, an external objective description of use need not provide the sort of understanding that comes from knowing the translation into one's own terms. (Harman, 1990b, 1993a). Translation is a holistic enterprise. I map as much as I can of your language into mine in a way that tries to preserve certain constraints as much as possible (Harman, 1993b). Similarly, in trying to understand what it is like for you to have certain experiences, I map as much as I can of your total experiential system into mine in a way that tries to preserve certain constraints as much as possible. In either case, there is the possibility of "indeterminacy of translation" (Quine, 1960). We can imagine that there are two different ways to map your color vocabulary into mine, or your color experiences into mine, preserving relevant constraints as much as possible. That is to imagine a genuine indeterminacy as to what it is like for you (or me) to see red, just as there is a genuine indeterminacy as to the best interpretation of numbers in set theory. With respect to the color experiences of normal perceivers who are normal speakers of English, there is no such indeterminacy, because the relevant constraints on mapping one person's experiences into another's include taking into account what objects in the world give rise to those experiences. So, what it is like for one normal color perceiver to see red is quite similar to what it is like for any other normal color perceiver to see red.

8 Conclusion My tentative conclusion is that objective color is plausibly identified with a tendency to produce a certain reaction in normal perceivers, where the relevant reaction is identified in part with reference to the mechanisms of color perception. The subjective response to color is constituted by perceptual experience presenting or representing the environment as relevantly colored. The concept of color as it figures in this representation is simple and unanalyzable in causal terms, because color is experienced as a simple basic quality, rather than a disposition or complex of causal properties. Possession of a perceptual concept of color is

to be understood functionally: objective color leads to experiences in which the perceptual concept of color is manifested. These causal accounts do not capture everything we seem to be able to imagine about color. In particular, they do not allow for possible inverted spectra in otherwise normal color observers. But it is far from clear that what we seem to be able to imagine is actually a coherent possibility.

Armstrong, D., (1968). A materialist theory of the mind. London: Routledge & Kegan Paul. Block, N., (1990). Inverted earth. Philosophical Perspectives 4 pp. 53-79. Dennett, D., (1991). Explaining consciousness. Boston: Little, Brown. Gold, I., (1993). Color and other illusions: A philosophical theory of vision. Princeton University Ph. D. Dissertation. Harman, G., (1982). Conceptual role semantics. Notre Dame Journal of Formal Logic 23. Harman, G., (1990a). "The Intrinsic Quality of Experience", Philosophical Perspectives 4 pp. 31-52. Harman, G., (1990b). "Immanent and transcendent approaches to the theory of meaning". In Roger Gibson and Robert B. Barrett, eds., Perspectives on Quine. Oxford: Blackwell, pp. 144157. Harman, G., (1993a). "Can Science Understand the Mind?" in Gilbert Harman, ed., Conceptions of the Human Mind: Essays i n Honor of George A. Miller, Hillside, New Jersey; Lawrence Erlbaum, pp. 111-121. Harman, G., (1993b). "Meaning holism defended", Grazer Philosophische Studien 46, pp. 163-171. Johnston, M. (1992). How to speak of the colors. Philosophical Studies 6 8 pp. 221-263. Lewis, D., (1966). An argument for the identity theory. Journal of Philosophy 6 3 pp. 17-25. Nagel, T., (1974). What is it like t o be a bat? Philosophical Review 83 pp. 435-450. Peacocke, C., (1983). Sense and Content. Oxford: Oxford University Press. Shepard, R.N. (1992). The perceptual organization of colors: an adaptation to regularities of the terrestrial world? The Adapted Mind: Evolutionary Psychology and the Generation of Culture, edited by J.H. Barkow, L. Cosmides, & J. Tooby. New York: Oxford University Press. Shepard, R.N. (1993). On the physical basis, linguistic representation, and conscious experience of colors. Conceptions of the Human Mind: Essays i n Honor of George A. Miller, edited by G. Harman. Hillsdale, N.J.: Erlbaum.

Shoemaker, S., (1981). The inverted spectrum. Journal of Philosophy 74 pp. 357-381. Shoemaker, S., (1994). Phenomenal character. Nous 28 pp. 21-38. Sosa, E., (1990). Perception and reality. Information, Semantics and Epistemology, edited by E . Villaneueva. Oxford: Basil Blackwell. Quine, W.V., (1960). Word and Object. Cambridge, Massachusetts: M.I.T. Press.


Menta1 Paint and Ment a1 Latex Ned Block

The greatest chasm in the philosophy of mind -maybe even all of philosophy- divides two perspectives on consciousness. The two perspectives differ on whether there is anything in the phenomenal character of conscious experience that goes beyond the intentional, the cognitive and the functional. A convenient terminological handle on the dispute is whether there are "qualia". Those who think that the phenomenal character of conscious experience goes beyond the intentional, the cognitive and the functional are said to believe in qualitative properties of conscious experience, or qualia for short. The debates about qualia have recently focused on the notion of representation, with issues about functionalism always in the background. All can agree that there are representational contents of thoughts, for example the representational content that virtue is its own reward. And friends of qualia can agree that experiences at least sometimes have representational content too, e.g. that something red and round occludes something blue and square. The recent focus of disagreement is on whether the phenomenal character of experience Copyright @ Ned Block 1995.

is exhausted by such representational contents. I say no. Don't get me wrong. I think that sensations -almost always -perhaps even always- have representational content in addition to their phenomenal character. What's more, I think that it is often the phenomenal character itself that has the representational content. What I deny is that representational content is all there is to phenomenal character. I insist that phenomenal character outruns representational content. We can call this view "phenomenism". Phenomenists believe in qualia, as defined above. The opposed view, which we can call "representationism", holds that the phenomenal character of an experience does not go beyond its representational content. (Terminological note: everyone believes that experience has phenomenal character -the disagreement is about whether that phenomenal character is exhausted by its representational content.) In its most extreme form, this view holds (e.g. Dennett, 1991, following Armstrong) that sensation is just a form of judgement. This version is close to eliminativism about phenomenal character. But other representationists recognize that sensations and other experiences are not judgements. Nonetheless, according to them, all experiences have representational content, and these representational contents completely capture the phenomenal character of the experiences. This paper is a critique of representationism. First I will briefly discuss an internalist form of representationism, then I will go on to the main topic of the paper, externalist forms of the view.

1 Internalism One form of representationism holds that the phenomenal character of experience is its "narrow intentional content", intentional content that is "in the head" in Putnam's phrase. That is, heads that are the same in aspects of their physico-chemical configuration that don't involve relations to the environment share all narrow intentional contents. In holding that phenomenal contents supervene on what is in the head, this view is "internalist". A full dress discussion of this view would discuss various ideas of what narrow intentional content is supposed to be. But this isn't a full dress discussion. I will simply say that all versions of this view that I can think of that have even the slightest plausibility (and that aren't committed to qualia) are in some way functionalist. They are functionalist in that they involve the idea that narrow intentional content supervenes on internal functional organization as well as physico-chemical configuration. That is, there can be no differences in narrow intentional contents without

2. MENTALPAINTA N D MENTALLATEX

21

corresponding differences at the level of causal interactions of mental states within the head. The view comes in different flavors: functional organization can be understood in terms of the interactions of common-sense mental states or in terms of the causal network of computational states. In both cases, there is a level of "grain" below which brain differences make no differences. One functional organization is multiply realizable physico-chemically in ways that make no difference in narrow intentional content. In other words, there is a level of organization above the level of physiology ("mental" or "computational") that determines narrow intentional content. (Tye, 1994 takes this view and I understand Rey, 1992a, 1992b and White, 1995 as evincing some sympathy for it.) Of course phenomenists can (and should) be internalists about phenomenal character too. But phenomenists can allow that phenomenal character depends on the details of the physiology or physico-chemical makeup of the brain. Of course, there are also dualist forms forms of phenomenism, but both the physicalist and dualist forms of phenomenism agree that there is no need to suppose that qualia supervene on functional organization.' There is a very simple thought experiment that raises a serious (maybe fatal) difficulty for any such (functionalist) internalist form of representationism. Suppose that we raise a child (call her Eliza) in a room in which all colored surfaces change color every few minutes. Further, Eliza is allowed no information about grass being green or the sky being blue. The result is that Eliza ends up with no standing beliefs that distinguish one color from another. We can allow that Eliza learns the color names in English, but nothing else that distinguishes the colors. Now it may be that the result is that there are no functional differences among her color experiences (except the attachment to different names). For example, it may be that the result of this process is that Eliza has no associations or behavioral inclinations or dispositions towards red that are any different from her associations or inclinations or dispositions towards blue. Nonetheless, she can name the colors and describes vividly different color experiences, just as we do. 'Shoemaker 1994a, 1994b combines phenomenism with internalist representationism. He holds t h a t when one looks a t a red tomato one's experience has a phenomenal character ( t h a t is not exhausted by its functional characterization) t h a t represents t h e tomato as having a phenomenal property and also as being red, t h e latter via t h e former. T h e view is representationist because t h e phenomenal character of t h e experience consists in representing phenomenal properties of objects. (The objects have those phenomenal properties in virtue of standing in causal relations t o those experiences).

Suppose this is so. The challenge to the internalist representationist, then, is to say what the difference is in intentional content between the experience of red and the experience of blue? There is a difference in phenomenal character, so the internalist representationist is committed to finding a difference in function. But the example is designed to remove all differences in function. Of course, there may be innate behavioral differences between the experience of red and the experience of blue. Perhaps we are genetically programmed so that red makes us nervous and blue makes us calm. But despite claims of this sort (Dennett, 1991), I think it is fair to say that such assertions are empirical speculations. (See Dennett's only cited source, Humphrey, 1992, which emphasizes the poor quality of the empirical evidence.) Perhaps the internalist will say that there would be no differences among her experiences. Red would of necessity look just the same to her as yellow. But this is surely an extraordinary thing for the internalist to insist on. It could be right of course, but again it is surely an unsupported empricial speculation. In short, the internalists are caught in a dilemma. Either they claim that there are innate functional differences between the experiences of red, blue, etc. In that case, they depend on an empirical speculation. Or else they claim that there would be no differences between Eliza's experience of red, of blue, etc. And in that case too, they depend on an empirical speculation. Let me be clear about what I am n o t claiming. I admit that I do think it immensely plausible that Eliza would have color experience much like ours. However, I am not depending on that, but rather on something much weaker, namely that this is an o p e n e m pirical q u e s t i o n . Further, I am certainly not claiming that there are no genetically programmed behavioral differences among color experiences, but only that this also is an open empirical question. Representationists usually see the belief in qualia as confused or otherwise ruled out on purely logical or conceptual grounds. I am maintaining that the dispute between lovers of qualia and internalist representationism depends on an open empirical question. But physicalism is an empirical thesis, so why should the representationist be embarrassed about making an empirical claim? Answer: physicalism is a very general empirical thesis having to do with the long history of successes of what can be regarded as the physicalist research program. Internalist representationism, by contrast, depends on highly specific experimental claims. Fol example, the issue would be settled if newborn babies showed differences in their reactions to red and green before having had any differential experience

2. MENTALPAINT A N D MENTALLATEX

23

with these colors. I doubt that very many opponents of qualia would wish their point of view to rest on speculations as to the results of such experiments. Of course the defender of qualia does not rest everything on a prediction about the results of such experiment. Our view is that even if such experiments do show some asymmetries, there are possible creatures -maybe genetically engineered versions of humans, in whom the asymmeteries are ironed out. (See Shoemaker, 1982.) And those genetically engineered humans could nonetheless have color experience much like ours.

2

Externalism

That is all I will have to say about internalist representationism. Now I will move to the main topic of this paper, externalist r e p resentationism. In this section, I will try to motivate externalist representationalism. The consideration I will advance in motivating it will be of use to me later. Then I will advance various considerations that cause one or another sort of difficulty for the view. I won't be making any crushing objections here, though I will be discussing some issues having to do with one objection that, when properly spelled out (something I won't try to do here), does crush one form of representationism. The purpose of this paper is more exploratory. I will be discussing a number of considerations that are relevant to the issue even if inconclusive. The structure of what follows is that of a list. I list considerations, discuss them, then go on to the next one. Often, when I see water I see it as water; that is, my visual experience represents it as water. (Sometimes my visual experience represents milk as water, and then when I drink the milk, I get a dreadful shock, the shock of clashing representations.) Most of you have seen water as water all your lives, but I'm different, I'm a foreigner. I was born on Twin Earth and emigrated to Earth at age 18. When I was 15 and looked at the sea on Twin Earth, my visual experience represented the twin-water as twin-water (though of course we didn't call it that). Perhaps you are skeptical about whether visual experience represents such properties, but please bear with me. when I first got here, I saw water as twin-water, just as you, if you went to Twin Earth would see the liquid in the oceans as water. Now, many years later, my practices of applying concepts are relevantly the same as yours: my practices show that I am committed to the concepts of my adopted home, Earth. Now when I look at

the sea of my adopted home, my visual experience represents the water as water just as yours does. (Again, let's suppose.) So the representational content of my experience of looking at the sea has changed. But my visual experience is nonetheless indistinguishable from what it was. If you took me up in a space ship and put me down by some sea-side, I wouldn't know whether I was on Earth or Twin Earth. (See Stalnaker's (1996) commentary on Lycan (1996) on Block (1990)) The representational content of my experience has changed, but the phenomenal character has stayed the same. And that shows, someone (not me) might argue, that there is some sort of gap between representational content and phenomenal character. One way for the representationist to answer would be to note that though all phenomenal character is (according to representationism) representational content, the converse is not true. Not all representational contents are phenomenal characters. For example, the thought content that mu-mesons have odd spin is not a phenomenal character. But how is this point supposed to apply to the issue at hand? The putative gap between representational content and phenomenal character has to do with the representational content of visual experience, not thought. Here is one way of extending the representationist response: the representationist should hold that visual experience has two kinds of representational content: one kind of representational content can be identified with phenomenal character and another kind of representational content is distinct from phenomenal character. The (relevant) phenomenal character of experience that has remained the same throughout my life is a matter of the observable or appearance properties of the liquids in oceans, rivers and streams, their color, sheen, motion, taste, smell, etc. It is these properties that my experience has represented liquids as having that have been shared by my experiences of both twater and water. Those are the representational contents that make my visual experience of the ocean at age 15 indistinguishable from my visual experience of the ocean now. The reply just considered is a success for representationism. We started with an experiential continuity despite representational change that challenged the representationist. But then it turned out that the representationist can explain the difference representationally. The main burden of this paper is to explore some reasons for thinking that the representationist cannot always repeat this success. Two preliminary issues: First, one might object to my claim that visual experience can ever represent anything as water. Perhaps, you might say, visual experience can represent something as round


25

or at least red, but not as water. Nothing that is really important to my argument will depend on this issue. Nonetheless, I will note that disallowing visually representing water as water has the effect of limiting the resources available to the representationist. Next, a terminological reminder. I take a qualitative character or quale as a phenomenal property of an experience that eludes the intentional, the functional and the purely cognitive. 'Phenomenal character' is a more neutral term that carries no commitment to qualia. Both the representationist and the phenomenist can agree that there are phenomenal characters, even though the former but not the latter thinks phenomenal characters are wholly representational.

3 Supervenience If phenomenal character supervenes on the brain, there is a straightforward argument against representationism. For arguably, there are brains that are the same as yours in internal respects (molecular constitution) but whose states represent nothing at all. I will produce an example that supports this idea by combining two somewhat less successful attempts. Consider first the poor swampman, the molecular duplicate of you who comes together by chance from particles in the swamp. He is intended as an example of total lack of representational content. How can he refer to Newt Gingrich when he has never had any causal connection with him; he hasn't ever seen Newt or anyone else on TV, never seen a newspaper, etc. But although he has no past, he does have a future. If at the instant of creation, his brain is in the same configuration that yours is in when you see a ripe tomato, perhaps the swampman's state has the same representational content as yours in virtue of that state's role in allowing him to "track" things in the future that are the same in various respects (e.g. color) as the tomato. He will track appearances of Bill Clinton in the future just as well as you will. So let's try another familiar case. The brain in a vat is also often intended as a case of no representational content. But there is something in the idea that one can apply causal theories of reference and come up with the conclusion that its states represent features of the vat computer. Now we can do better at avoiding these difficulties by combining the two thought experiments. A moleculefor molecule duplicate of your brain comes into existence in my bathtub just now. It has no life support system so it won't last long. Unlike the swampman, it has no future, and unlike the brain in a vat it has no past. So we should allow that its states represent nothing.

Objection: the bathtub-brain has the capacity to track things in the future in virtue of counterfactuals about what it would do if it were embodied. Well, yes, but it is not clear how damaging this point is. (Note: not clear.) To say the bathtub-brain has the capacity to track is like saying that men have the capacity to become pregnant if only they were to be given wombs (Block, 1981). Or to use Strawson's (1994) example, it is like saying that a heap of unassembled piano parts has the capacity to play music. The bathtub-brain has exactly the same internal configuration as your brain when you are looking at a ripe tomato. If we can assume supervenience of phenomenal character on the brain, we can refute the representationist (depending on the success of the reply just considered). The phenomenal character of the bathtub-brain's experience is the same as yours but its experiences have no representational content at all. So how can phenomenal character be captured representationally? Maybe some differences in the representational contents of experience don't make a phenomenal difference; but if representationism is right, some do. There is more to experience than representational content. But if this argument is successful, won't the representationist just deny supervenience? And in fact, that's what representationists often do (Lycan, 1996a, 1996b; McDowell, 1994). I don't think that this is a no-cost move for the representationist, but I won't labor the point.

4 What is the Issue? One of the ways that I have been framing the issue is this: Is there more to experience than its representational content? But if experiences are brain states, there will be more to experiences than their representational contents just as there is more to sentences in English than their representational contents. For example, the size of the font of this sentence is something more than (or anyway, other than) its representational content. So the question is better taken as: is there anything mental in experience over and above its representational content? I say yes, the representationist says no. Harman (1990, 1996) expresses his version of representationism about experience by claiming that in experience we only are aware of properties of what is represented, not the vehicle of representation. When we look at a red tomato, no matter how hard we try to introspect the aspect of the experience that r e p r e s e n t s redness, all we succeed in doing is focusing our attention on the redness of the tomato itself. Harman relies on the famous diaphanousness of per-


27

ception (Moore, 1922), that is, when one looks at a red tomato, the effect of concentrating on the experience of red is simply to attend to the redness of the tomato itself. As a point about introspection, this seems to me to be straightforwardly wrong. For example, close your eyes in daylight and you will find that the diaphanousness of perception -you will have experiences that are not clearly and uncontroversially as of anything. (More on this below.) Of course, Harman can allow that we know when an experience happens and how long it lasted, but that will no doubt be glossed by him as a matter of when we looked at the tomato and how long we looked. Harman concludes that introspection gives us no access to anything non-representational about the experience, no access to "mental paint". He argues that the contrary view confuses properties of what is represented, the intentional object of perception, with properties of the vehicle of representation, what is doing the representing. I don't agree with the imputation of fallacy. But my point right now is more preliminary: that if Harman means to define representationism in this way, his definition is too narrow. I also think his view depends on an unreliable approach to introspection. He writes as if the issue is whether we can introspect the representational features of the experience, the mental paint that represents the redness of the tomato. But there are two deeper issues. 1. The first is whether there is mental paint, even if Harman is right that we cannot become aware of it when we are seeing a tomato. One way -not the only way- of seeing why there is a real issue here is to consider the idea that the possibility of an inverted spectrum shows that there is more to experience than its representational content. According to this argument (Shoemaker, 1982; Block, 1990; see also Block and Fodor, 1972), your experience and my experience could have exactly the same representational content, say as of red, but your experience could have the same phenomenal character as my experience as of green. Shoemaker (1994a, 1994b) agrees with Harman's views about introspection. He agrees that we cannot be aware of mental paint, that when we try to introspect the experience of the redness of the tomato, all we succeed in doing is attending to the color of the tomato itself. But according to Shoemaker, we have a kind of indirect introspective access to the phenomenal character of the experience via our intuitions about the inverted spectrum. By imagining that things we both call red look to you the same way that things we both call green look to me, we succeed in gaining indirect introspective access to men-

tal paint. Thus there is mental paint. So Shoemaker's view gives us a n example of how on certain assumptions it would be reasonable to think that there is mental paint even if we can't be directly aware of it, and that is one way of illustrating the distinction. 2. A second deeper issue is whether there are phenomenal features of experience that are not even vehicles of representation. For example, according to me, the phenomenal character of the experience of orgasm is partly non-representational. and further, when we have both auditory and visual representions as of something moving, there are experiential but non-representational differences between these representations in different sensory modalities. A light flashes in one modality, a horn blows in another. Even in the case of mental paint, phenomenal character that is representational, perhaps that same phenomenal character might sometimes not represent anything, just as paint can represent color in one painting but not in another. So the even more controversial issue is whether there are phenomenal features of experience that are not representations at all. Such a mental feature would be (in this respect) like the font of a line of print or like the latex in latex paint. Paint has pigment t h a t differs from color to color- and also a base that can be common to many colors. One such base is latex. The pigment represents the color, the latex represents nothing. So the two issues are whether there is mental paint and whether there is mental latex.

T h e analogy is in some ways misleading. Paint always has a base, latex in one paint, oil in another. But I do not want to claim that there are non-representational phenomenal features of every experience or that when there are, these non-representational features form support the representational features in the manner of a "base". I don't even believe this for color experience. The quale that represents red fur me might represent green for you, and that same quale might represent nothing at all for the bathtub-brain. But that doesn't suggest that there is anything phenomenal in common to all color experience. And even if there is something in common, that something may be fully representational in you and me (though not in the bathtub-brain). There is an ambiguity in the phrase 'representational feature of experience' that interacts with the mental paintlmental latex distinction. As I mentioned earlier, Harman (1990,1996) argues that we are only aware of representational features of our experience and


29

we cannot be aware of any non-representational features. But 'representational feature of our experience' as used by Harman does not encompass mental paint -that's what he holds we are never aware of. In Harman's lingo, a representational feature of experience is a feature of what is represented, not what is doing the representing. On his way of making the representational/non-representational distinction, both mental paint and mental latex are non-representational. But as I have just been noting, there is another way of making the representational/non-representational distinction that draws the line between paint and latex. Paint is representational (It represents); latex is not (It does not represent). To sum up then, we can distinguish three things: 1. The intentional content of an experience. I am currently looking at a tomato and my experience represents the tomato as red.

2. Mental properties of the experience that represent the redness of the tomato. This is mental paint. According to me, the phenomenal character of the experience is such a mental property: it represents the tomato as red. According to me, one can have introspective access to this phenomenal character. Harman denies both claims. 3. Mental properties of the experience that don't represent anything. This is mental latex. I don't know whether there are any such properties in the case of a normal experience of a red tomato, but I do claim that such properties are involved in orgasm-experience,

As I mentioned, Harman says that we are only aware of what is represented by our experience, the intentional object of the experience, not what is doing the representing, not the vehicle of representation. But what will Harman say about illusions, cases where the intentional object does not exist? Surely, there can be something in common to a veridical experience of a red tomato and a hallucination of a red tomato, and what is in common can be introspectible. This introspectible commonality cannot be constituted by or explained by the resemblance between something and nothing. It seems that the representationalist ought to appeal to the intentional content that is shared by the two experiences, the content that there is a red tomato in front of me. As I understand Harman, this is very much not his view; he seems to insist rather that one is aware of the intentional object, not the intentional content. But suppose Harman were to hold that we are aware of the shared intentional content;

what would that come to? What is it to be aware of the intentional content that I am seeing a red tomato or that two experiences have that intentional content? I don't see what awareness of a n intentional content could come to if not awareness that some state has that intentional content. And surely, awareness that two experiences have the same intentional content requires awareness that each has that intentional content. But if Harman were to give this account of the introspectible similarity, he would have to concede that we have introspective awareness of some mental properties of experience, not just of the intentional objects of experience. The view I've just mentioned is taken by Lycan (1995) -that one can be aware of a family of mental properties of an experience, namely that the experience represents something, that it represents a tomato, that it represents the tomato as red, etc. So Lycan can deal with illusions by saying that when one hallucinates a red tomato, there is something introspective in common with a normal veridical perception of a tomato, namely in both cases one is aware that the experience represents a red tomato. These mental properties aren't mental paint because they don't represent. However, my definition of mental latex does, unfortunately, count them as mental latex because they are mental properties of experience that do not represent. So let's understand the definition of mental latex as containing a qualification: the property of having certain intentional contents does not count as mental latex. I can't resist interjecting the observation that from the point of view of a phenomenist (like me), the view that I have just described sounds like a bizarre religion. The conception according to which the content of experience is the same as the content of belief sounds even stranger when we think of what makes two experiences experientially similar. Some representationists combine externalism and internalism. For example, Rey (1992a, 199213) individuates color experience partly in terms of what colors it represents and partly in terms of what he sees as syntactic properties of the vehicle of representation.(There is a similar view in Lycan, 1996b.) I won't try to consider such mixed views here. Those who deny both mental paint and latex are representationists; those who countenance one or the other are phenomenists. The representationists include Dretske (1995, 1996), Harman (1990, 1996), Lycan (1995, 1996), McDowell (1994), Rey (1992a,b) and Tye (1995, 1996) (See also White (1995) for a representationist view of color experience). The phenomenists include Block (1990, 1994a), Loar (1990), McGinn (1991), Peacocke (1983) and Shoemaker (1982,


31

1994a, 1994b). Shoemaker and I hold that the inverted spectrum argument and the inverted earth argument make strong cases for phenomenism. (Loar, McGinn and Peacocke have declared doubts about the inverted spectrum arguments; I don't know what they think of the inverted earth argument.) I won't go into the inverted spectrum argument here, and I will only be able to mention inverted earth briefly. I will mention some other considerations that are less effective but which I hope put some pressure on the representationist.

5 Intentional Inexistents The phenomenist can allow (and I do) that pain often has some representational content. For example, pain-experiences often represent a location. Of course the representationist also says that the representational content of a pain is all there is (that is mental) to it. (Well, this isn't quite right, since as just noted, there is good reason for the representationist to count the property of having a certain intentional content as an introspectible and therefor mental property of an experience.) Since both representatists and phenomenists should agree that there is more to pain than location, the representationist should say more about what this extra intentional content of pain is. Some representationists say that the intentional object of a pain experience is simply this: a pain. But what are these intentional objects? And do they exist? One line of thought is that they are sense data, mental individuals. Lycan (1995) defines a "quale" as an "introspectible monadic qualitative property of what seems to be a phenomenal individual." And he goes on to say "I am among those philosophers who think that qualia are merely the representational contents of the sensations that feature them." So he believes in qualia in what he calls the loose-Lewisian (C.1, not David) sense, namely introspectible monadic qualitative properties of what seem to be phenomenal individuals and are referred to and individuated as such. Lycan's view is that phenomenal individuals don't exist (I agree). He says they are merely seeming phenomenal individuals. So qualia, in Lycan's view are never really instantiated. (However, he also says that we can take them to be colors of sub-regions of the visual field -and what is the visual field if not a phenomenal individual? It looks like Lycan himself is buying into the doctrine that he officially rejects.) So the theory is that the representational content of a red afterimage-experience is that there is a phenomenal indidividual that is red -really red, like a tomato!! But as Boghossian and Velleman (1989)

point out, there is something suspect about the phenomenology that this claim is based on. Afterimages -at least the ones that I have tried- don't look as if they are really objects or as if they are really red. They look.. . illusory. Try it out yourself. Don't get me wrong. I agree that a n image experience and a tomato experience share something that one might call a color property. My point is that when one has a n afterimage one has no tendency to think thereby that anything is really red, and so the introspective foundation for the theoretical claim that the afterimage represents something in the world as really red is weak. The fact that the afterimage-experience and the tomato-viewing experience share something that one might call a color property might be better glossed, I think, as sharing a phenomenal property that represents red in one case but not the other. More importantly, this whole discussion gets us off the track. If we are interested in the phenomenism/representationalism issue, we would do well to ignore the question of whether there really are phenomenal individuals. For even if there are no such things (as I believe), we can put all the important issues in terms of whether there are any mental properties of experience that go beyond the functional, the intentional and the cognitive. Lycan uses the word 'quale' to mean a property (a color) which he thinks our experience ascribes to phenomenal individuals (which don't actually exist). But to the extent that the word 'quale' is the terminological locus of the representationism/phenomenism issue, it ought to be used to mean a property of experience that goes beyond the intentional, the functional and the cognitive. Further, even if Lycan were right about the intentional content of afterimage experiences -that they ascribe colors to seeming phenomenal objects- that would cast no light on the main issue of representationism w h e t h e r there are properties of those experiences that go beyond these intentional contents. I could decide to go along with Lycan on his claim about the representational content of imageeexperiences, but insist that in addition to these representational contents, image experiences have phenomenal character that goes beyond the representational content (as shown, e.g. by the inverted earth thought experiment). Even if we were to accept Lycan's view of the representational content of after-image experiences, it would not be clear how it applied to other experiences, e.g. pain. If the intentional object is a pain, what are the properties experience ascribes to it? And do we really want to claim that pain experiences ascribe phenomenal properties to non-existent objects? A less counter-intuitive account has it that the intentional object of a pain-experience is not the pain itself (a

2 . MENTAL PAINTAND MENTAL LATEX

33

non-existent phenomenal individual), but, say in the case of a pain in the leg, the leg itself (which normally does exist). This more plausible view would say that one's pain-experience ascribes a property to the leg, the property of hurting. But the plausibility of this view is purchased at the price of a difficulty from the representationist point of view -how to give a representational account of hurting. Instead of discussing this approach to pain, I will switch to another bodily sensation, orgasm.

6 Latex and Orgasm Is the experience of orgasm completely captured by a representational content that there is an orgasm? Orgasm is phenomenally impressive and there is nothing very impressive about the representational content that there is an orgasm. I just expressed it and you just understood it, and nothing phenomenally impressive happened (at least not on my end). I can have an experience whose content is that my partner is having an orgasm without my experience being phenomenally impressive. In response to my invocation of the non-representational content of orgasm (Block, 1995a, 1995b), Tye (1995a) says that the representational content of orgasm "in part, is that something very pleasing is happening down there. One also experiences the pleasingness alternately increasing and diminishing in its intensity." But once again, I can have an experience whose representational content is that my partner is having a very pleasing experience down there that changes in intensity, and although that may be pleasurable for me, it is not pleasurable in the phenomenally impressive way that that graces my own orgasms. I vastly prefer my own orgasms to those of others, and this preference is based on a major league phenomenal difference. The location of "down there" differs slightly between my perception of your orgasms and my own orgasms, but why should that matter so much? Of course, which subject the orgasm is ascribed to is itself a representational matter. But is that the difference between my having the experience and my perceiving yours? Is the difference just that my experience ascribes the pleasure to you rather than to me (or to part of me)? Representational content can go awry in the heat of the moment. What if in the heat of the moment I mistakenly ascribe your orgasm to me or mine to you? Would this difference in ascription really constitute the difference between the presence of absence of the phenomenally impressive quality? Perhaps your answer is that there is a way in which my orgasm-experience ascribes the orgasm to me that is immune to

the intrusion of thought, so there is no possibility of a confused attribution to you in that way. But now I begin to wonder whether this talk of 'way' is closet phenomenism. No doubt there are functional differences between my having an orgasm-experience and merely ascribing it to you. Whether this fact will help to defend representationism depends on whether and how representationism goes beyond functionalism, a matter to be discussed in the section after next. According to me, there are features of the experience of orgasm that don't represent anything; so mental latex exists. I don't expect this example to force representationists to concede that there is mental latex. Appeals to intuitions about relatively unstructured cases are rarely successful. That is why the complex thought experiments such as those involving the inverted spectrum and inverted earth are useful. The disagreement focuses not on the intuitions themselves (I believe that they reflect our concepts so directly that everyone can be brought to agreement on them) but on what the intuitions show. Of course, we should not demand that a representationist be able to capture his contents in words. But if we are to try to believe that the experience of orgasm is nothing over and above its representational content, we need to be told something fairly concrete about what that representational content is. Suppose the representational content is specified in terms of recognitional dispositions or capacities. One problem with this suggestion is that the experience of orgasm seems on the face of it to have little to do with recognizing orgasms. Perhaps when I say to myself "There's that orgasm experience again" I have a somewhat different experience from the cases where no recognition goes on. But there is no plausibility in the insistence that the experience must involve some sort of categorization. And if you are inclined to be very intellectual about human experience, think of animals. Perhaps animals have the experience without any recognition. The representationists should put up or shut up. The burden of proof is on them to say what the representational content of experiences such as orgasm and pain are.

Representationists often emphasize the diaphanousness of visual experience. For example, Harman says "Look at a tree and try to turn your attention to intrinsic features of your visual experience. I predict you will find that the only features there to turn your attention


35

to will be features of the represented tree." (1990). But the diaphanousness of perception is much less pronounced in a number of visual phenomena, notably phosphene-experiences. (I use the cumbersome 'phosphene-experience' instead of the simpler 'phosphene' by way of emphasizing that the phenomenist need no have any commitment to phenomenal individuals.) Phosphene-experiences are visual sensations "of" color and light stimulated by pressure on the eye or by electrical or magnetic fields. (I once saw an ad for goggles that you could put on your eyes that generated phosphenes via a magnetic field.) Phospheneexperiences have been extensively studied, originally in the 19th Century by Purkinje and Helmholz. Close your eyes and place the heels of your hands over your eyes. Push your eyeballs lightly for about a minute. You will have color sensations. Are they representational? It is hard to say. What would the world have to be like for a phosphene experience to be veridical? "The world would have to be such that there are colored moving expanses somewhere," it might be said. (Mark Johnston urged this.) Perhaps phospheneexperiences are representational in a way that leaves many features (e.g. location) indeterminate. Nonetheless, no one should conclude that introspecting a phosphene experience is purely a matter of attending to an intentional object of perception. I don't take phospheneexperiences to compel anyone to believe that there is mental paint or latex. I don't think one can show that via such a simple introspection. But phosphenes do serve to remind us that not all of visual experience is clearly and obviously diaphanous. Lycan (1987) says: ". . .given any visual experience, it seems to me, there is some technological means of producing a veridical qualitative equivalent -e.g. a psychedelic movie shown to the subject in a small theater" (p. 90). But there is no guarantee that phosphene experiences produced by pressure or electromagnetic stimulation could be produced by light. (Note: I don't say there is a guarantee that phosphene-experiences could not be produced by light, but only that there is no guarantee that they could.) I do wonder if Lycan's unwarranted assumption plays much of a role in leading philosophers to suppose that phosphene-experiences, afterimage-experiences and the like are entirely representational. According to me, in normal perception one can be aware of the mental paint -the sensory quality that does the representing. This idea can be illustrated (this is an illustration, not an argument) by Bach-y-Rita's famous experiment in which he gave blind people a kind of vision by hooking up a TV camera to their backs which produced tactual sensations on their backs. Bach-y-Rita says that

the subjects would normally attend to what they were "seeing". He says "unless specifically asked, experienced subjects are not attending to the sensation of stimulation on the skin of their back, although this can be recalled and experienced in retrospect." (Quoted in Humphrey, 1992, p. 80.) When I a m looking at a tomato and attending to the color of the tomato, I a m not attending to the visual sensation, but I believe that we can attend to the visual sensation just as Bach-y-Rita's subjects could attend to their tactual sensations. I think that the Bach-y-Rita experiment is useful in thinking about whether Lycan's version of representationism is better than Harman's. Let me remind you about the difference. Harman insists that all we can introspect in experience are the intentional objects of experience. Lycan, however, allows that we can actually introspect certain properties of the experiences themselves. At first glance, reflection on the the Bach-y-Rita experiment provides support for Lycan over Harman. For the ability of Bach-y-Rita's subjects to introspect their tactual sensations helps to remind us that we really can notice features of our own visual sensations. But Lycan has a very limited idea of introspecting the phenomenal properties of experiences. His concession, you will recall, was to allow introspection of the property of having certain intentional properties. But is that what Bach-y-Rita's subjects were doing? Were they introspecting that the sensations on their backs represented, say a couch? Perhaps occasionally, but I doubt it that that's what Bach-y-Rita was talking about. I think he meant that they were attending to the experiential quality of the feelings on their backs. And I think that this case helps to remind us that at least sometimes when we introspect visual experience, we are attending to the phenomenal properties of experience, not the fact that they have certain intentional properties. So although Lycan's version of representationism is superior to Harman's in allowing the existence of introspection of something other than intentional properties of experience, it does not seem true to what that introspection is often like. Moving now to the orgasm and pain cases, there is a challenge here for the representationist. Just what is the representational content of these states? In vision, it often is plausible to appeal to recognitional dispositions in cases where we lack the relevant words. What's the difference between the representational contents of the experience of color A and color B, neither of which have names? As I mentioned earlier, one representationist answer is this: T h e recognitional dispositions themselves provide or are the basis of these contents. My experience represents A as that color, and I can misrepresent some

2 . MENTALPAINT AND MENTALLATEX

37

other color as that color. But note that this model can't straightforwardly be applied to pain. That color is a feature of objects, but that pain is a feature of the mind. Suppose I have two pains that are the same in intensity, location, and anything else that language can get a handle on -but they still feel different. Say they are both twinges in the belly that I have had before, but they aren't burning or sharp or throbbing. "There's that one again; and there's that other one." is the best I can do. If we rely on my ability to pick out that pain, (arguably) we are demonstrating a phenomenal character, not specifying a representational content. (Note the difference between Loar's (1990) proposal of a recognitional view of phenomenal concepts and the current suggestion that a recognitional disposition can specify phenomenal character itself. Phenomenal character is what a phenomenal concept specifies or refers to.) The appeal to recognitional dispositions to fill in representational contents that can't be specified in words has some plausibility, so long as the recognitional dispositions are directed outward. But once we direct them inward, one begins to wonder whether the resulting view is an articulation of representationism or a capitulation to phenomenism. I will return to this point.2

8 Is Representationism Just a Form of Functionalism? Consider what I shall call "quasi-representationism" . Quasi-representationists agree with phenomenists that there are differences between sensory modalities that cannot be cashed out representationally. One modality is flashing lights, another is tooting horns. But quasi-representationists agree with representationists that within a single modality, all phenomenal differences are representational differences. (I think that this is the view that Peacocke, 1983, Ch. 1 argues against .) I think the first thing to note about quasi-representationism is that it is something of a contrived view. Some philosophers are attracted to representationism but can't bring themselves to treat the experiential differences between say vision and touch as entirely representational. So they treat this difference functionally. They plug a *AS I mentioned earlier, there is also a problem for the recognitional view in the plausibility of the idea that we (or animals) can have an experience without any sort of categorization or recognition.

gap that they see in representationism with functionalism. But they should tell us why they don't chuck representationalism altogether in favor of functionalism. Do they think that functionalism is in some way inadequate? Is there some phenomenon (e.g. afterimages) that requires the functionalist to add a non-functional element to supplement the a ~ c o u n t ? ~ Many philosophers in the representationist ballpark are rather vague about whether they are pure representationists or quasi-representationists, but Tye (1995b) makes it clear that he is a pure representationist . How can we decide whether representationism needs supplementation by functionalism? Suppose I both touch and see a dog. Both experiences represent the dog as a dog, but they are different phenomenally. Representationists are quick to note that the two experiences also differ in all sorts of other representational ways. (See Tye, 1995a, for example.) The visual experience represents the dog as having a certain color, whereas the tactual experience represents it as having a certain texture and temperature. In Block (1995a, b) I tried to avoid this type of rejoinder by picking experiences with very limited representational content. If you wave your hand in the vicinity of your ear, you experience movement without size, shape or color. You have a visual experience that plausibly represents something moving over there and nothing else. And I imagined that there were auditory experiences with the same content. But my expert consultants tell me that I was wrong. There is no auditory analog of peripheral vision. For example, any auditory experience will represent a sound as having a certain loudness. But that does not ruin the point. It just makes it slightly harder to see. Imagine the experience of hearing something and seeing it in your peripheral vision. It is true that you experience the sound as having a certain loudness, but can't we abstract away from that, concentrating on the perceived location? And isn't there an obvious difference between the auditory experience as of that location and the visual experience as of that location? If so, then there is either mental paint or mental latex. (The ways in which representationally identical experiences might be phenomenally different could involve differences in either paint or late^.)^

%y impression is t h a t Lycan (1995, 199613) regards functionalism as inadequate t o capture t h e experiential content of images and t h a t motivates his form of representationism. See also White (1995). 4See the papers in Crane (1992) for more on this issue.


39

9 Seeing Red for the First Time Marvin is raised in a black and white room, never seeing anything of any other color. He learns that fire engines and ripe tomatoes are red, but he never actually sees anything red. Then he is taken outside and shown something red without being told that it is red. (I've changed 'Mary' to 'Marvin' so as to emphasize the small differences between this and Jackson's (1982) argument.) He learns what it is like to see red, even though he does not know what that color is called. He might say: "So that's what it is like to see that color." Lewis (1990) (following Nemirow) says that Marvin aquires an ability, some sort of recognitional know-how. But as Loar (1990) notes, this idea can't account for embedded judgements. Here's an example that fits the Marvin case: "If that's what it is like to see red, then I will be surprised." What does the representationist say about what Marvin has learned? If Marvin is told that what he sees is red, the representationist might say that he has acquired a visual representational concept, the concept of red. But can the representationist say this if Marvin doesn't know that it is red? Perhaps the representationist will say "Sure, he acquires the concept of red without the name 'red'." Perhaps the representationist will say that what Marvin acquires is a recognitional concept. After all, he can say "There's that color again". He has a recognitional concept that he applies on the basis of vision, even though it doesn't link up to his linguistic color concepts. But there is a trap for the representationist in this reply. For what, according to the representationist, is the difference between Marvin's concept of red and Marvin's concept of blue? He recognizes both. When he sees a red patch he says "There's that color again" and when then sees a blue patch he says "There's that other color again", each time collating his outer ostension with an inner ostension. But what, according to the representationist, is the difference between Marvin's concept of red and Marvin's concept of blue? The phenomenist will link the difference to an internal difference, the difference in the phenomenal quality of the experience of blue and the experience of red. But the representationist can't appeal to that without changing sides. And there is nothing else that is internal for him to appeal to. (Remember, we are supposing, as with Eliza raised in the room of changing colors, that he knows nothing that distinguishes the unnamed colors.) The representationist can't distinguish the concepts simply on the basis of the colors

themselves, for we can have different concepts of the same thing. The representationist owes us an answer. Here is a closely related point. If the representationist is willing to recognize a color concept that has been cut loose from everything but recognition, why shouldn't he also recognize such concepts turned inward? Why can't he have a recognitional concept of his own phenomenal state -"There's that experience again." (As I mentioned, Loar 1990 argues that a recognitional concept of one's own experiences just is a phenomenal concept.) Rey (1992a, 19928) postulates that color experiences involve the tokenings of special restricted predicates. So he would say that Marvin tokens 'R' when he sees red and 'B' when he sees blue. Is that a suitable representationist answer? Recall that we are now discussing externalist representationalism, and Rey's view would deal with this problem by bringing in an internalist element. Recall my objection to internalism in terms of Eliza, the girl raised in the room in which everything changes colors. Eliza perhaps has more or less normal color experience but may have no asymmetrical associations in her color experience. So what's the representational difference between her 'R' and ' B y ?Suppose Rey says: who needs a representational difference; the syntactic difference is enough. Then it becomes difficult to distinguish his position from the phenomenist position. I am a physicalist phenomenist. I believe that the difference between Marvin's experience of red and of blue is a physical difference. I suspect it is a difference in brain events, not naturally capturable in terms of :"syntaxn, but this is not a n important feature of the phenomenist position. I allow that it is an empirical question (of course) what the physical natures of phenomenal qualities are. Perhaps these natures are even functional.

10 Externalist Memory At the beginning of the discussion of externalism, I discussed the thought experiment in which I, a native of Putnam's Twin Earth, emigrated to Earth. When I first looked at water, I thought it was twater and, let's suppose, my visual experience represented it as twater. Much later, after learning everything that I've just told you, I decided to become a member of the Earth language community, to speak English, not Twenglish. Now, when I look at the sea, I take what I am seeing to be water and my visual experience represents it as water (let's suppose), not twater. But in some very obvious sense, water looks the same to me as it did the first time I saw it

2 . MENTALPAINTAND MENTALLATEX

41

even though my representational content has changed. If you blindfolded me and put me down a t the sea side, I wouldn't know from looking at the liquid in the ocean whether it was water or twater. My phenomenal character has stayed the same even though the representational content of my visual experience changed. But this doesn't show that there is anything non-representational about phenomenal character. For the shared representational contents are sufficient t o explain why there's no difference in what its like to see water. The key to the success of the representationist move here is that the representational change (from of twater to of water) is a change in a "non-appearance" property. The representationist move involves the following picture: experiences have representational properties of two types: the phenomenal character of an experience can be identified with one of those two types. The non-phenomenal type includes the representation of water as water, the phenomenal type includes such properties as color. (See Block, 1995a,b and Tye, 1995a.) What I will argue is that there is a twin earth case that turns on a property that does not allow a move corresponding the one just made. The property is color, which is an "appearance property" if anything is. The upshot, I will argue, is that there is mental paint (but there is no argument here for mental latex). I will illustrate this point with a n argument from Block (1990, 1994a) about color. I won't go into this argument in full detail.5 Inverted Earth is a place that differs from Earth in two important ways. First, everything is the complementary color of the corresponding earth object. The sky is yellow, the grass-like stuff is red, etc. Second, people on Inverted Earth speak an inverted language. They use 'red' to mean green, 'blue' t o mean yellow, and so forth. If you order paint from Inverted Earth and you want yellow paint, you FAX an order for "blue" paint. The two inversions have the 5~ will make use of Harman's (1983) Inverted Earth example. Block (1980) uses a cruder example along the same lines. (P. 302-303 of Block (1980) reprinted on p. 466 of Lycan (1990) and p. 227 of Rosenthal (1991).) Instead of a place where things have the opposite from the normal colors, I envisioned a remote arctic village in which almost everything was black and white, and the subject of the thought experiment was said to have no standing color beliefs of the sort of "Grass is green". Two things happen to him: he confuses color words, and a color inverter is placed in his visual system. Everything looks to have the complementary of its real color, but he doesn't notice it because he lacks standing color beliefs. In the 1980 version of the paper, I mistakenly attributed the suggestion to Sylvain Bromberger who told me when he later read it that he had no idea why I had attributed any such idea to him. Harman used the Inverted Earth e x a m ~ l eto make a different ~ o i n tfrom that made here: that representational content does not supervene on the brain.

effect that if inverters are inserted behind your eyes (and your body pigments are changed), you will notice n o difference when you go to Inverted Earth. After you step off the space-ship, you see some Twin-grass. You point at it, saying it is a nice shade of "green", but you are wrong. You are wrong for the same reason that you are wrong if you call the liquid in a Twin-earth lake 'water' just after you arrive there. The grass is red (of course I am speaking English not Twenglish here). But after you have decided to adopt the concepts and language of the Inverted Earth language community and you have been there for 50 years, your word 'red' and the representational content of your experience as of red things (things that are really red) will shift so that you represent them correctly. Then, your words will mean the same as those of the members of your adopted language community. There is a small difference between this form of the thought experiment and that of my 1990 and 1994.~In the old version, you are kidnapped and inserted in a niche in Inverted Earth without your noticing it (your twin having been removed to make the niche). In the new version, you are aware of the move and consciously decide to adopt the concepts and language of the Inverted Earth language community. The change has two advantages: first, it makes it clearer that you become a member of the new community. On the old version, one might wonder what you would say if you found out about the change. Perhaps you would insist on your membership in the old language community and defer to them rather than to the new one. The new version also makes it easier to deal with issues of remembering your past of the sort brought up in connection with the inverted spectrum in Dennett, 1991. The upshot is: 1. The phenomenal character of your color experience stays the same. That's what you say, and why shouldn't we believe you? 2. But the representational content of your experience, being ex-

ternalist, shifts with external conditions in the environment and the language community. (Recall that I am now discussing representationists who are externalists; I discussed internalist representationalism at the beginning of the paper.) Your phenomenal character stays the same but what it represents changes. This provides the basis of an argument for mental paint, -

61 am indebted to Bob Stalnaker for suggesting the change.

not mental latex. Mental paint is what stays the same; its representational content is what changes. What exactly is the argument for mental paint? Imagine that on the birthday just before you leave for Inverted Earth, you are looking at the clear blue sky. Your visual experience represents it as blue. Years later, you have a birthday party in Inverted Earth and you look at the Inverted Earth sky. Your visual experience represents it as yellow (since that's what color it is and your visual experience by that time is veridical let us suppose -1'11 deal with an objection to this supposition later). But the phenomenal character stays the same, as indicated by the fact that you can't tell the difference. So there is a gap between the representational content of experience and its phenomenal character. Further, the gap shows that phenomenal character outruns representational content. Why? How could the representationist explain what it is about the visual experience that stays the same? What representational content can the representationist appeal to in order to explain what stays the same? This is the challenge to the representationist, and I think it is a challenge that the representationist cannot meet. The comparison with the water case is instructive. There, you will recall, we also had phenomenal continuity combined with representational change. But this created no problem for the representationist because the phenomenal continuity itself could be given a representational interpretation. The phenomenal character of my visual experiences of twater and water were the same, but their representational contents differed. No problem because the common phenomenal character could be said by the representationist to be a matter of the representation of color, sheen, flow pattern and the like. Put what will the representationist appeal to in the Inverted Earth case that corresponds to color, sheen, flow pattern, etc.? That's the challenge that the thought experiment raises for representationism. There are many obvious objections to this argument, some of which I have considered elsewhere. I will confine myself here to two basic lines of objection. Bill Lycan has recently objected (1996a, 199613) that the testimony of the subject can only show that the phenomenal character of color experience is indistinguishable moment to moment, and that allows the representationist to claim that it shifts gradually, in synch with the shift in the representational content of color experience. (I briefly considered this objection in my 1990, p. 68.) If that were right, the refutation of representationism would evaporate. But this objection ignores the longer term memories. The idea is that you remember the color of the sky on your birthday last year, the year before that,

ten years before that, and so on, and your long term memory gives you good reason to think that the phenomenal character of the experience has not changed gradually. You don't notice any difference between your experience now and your experience 5 years ago or 10 years ago or 60 years ago. Has the color of the American flag changed gradually over the years? The stars used to be yellow and now they are white? No, I remember the stars from my childhood! They were always white. Of course, memory can go wrong, but why should we suppose that it must go wrong here? Surely, the scenario just described (without memory failure) is both conceptually and empirically possible. (As to the empirical possiblity, note that the thought experiment can be changed so as to involve a person raised in a room who is then moved to a different room where all the colors are changed. No need for a yellow sky, a yellow ceiling will do.) Now a different objection may be mounted on Lycan's behalf that the externalist representationist should be externalist about memory. According to my story, the representational contents of the subject's color experience have shifted without his knowing about it. So if my story is right, Lycan (if he is to be an externalist about memory) should say that the subject's color experience has shifted gradually without the subject's knowing it. And that shows that the subject's memory is d e f e ~ t i v e . ~ One possible justification is simply that the nature of phenomenal character is representational (and externalist), so the phenomenal character of experience shifts with its representational content. Since memory is powerless to reveal this shift, memory is by its nature defective. But this reply is weak, smacking of begging the question. The Inverted Earth argument challenges externalist representationism about phenomenal character, so trotting in externalist representationism about memory of phenomenal character to defend it seems a bit pathetic. The idea of the Inverted Earth argument is to exploit the first person judgement that in the example as framed the subject notices no diflerence. The subject's experience and memories of that experience reveal no sign of the change in environment. Yet his representational contents shift. Since the contents in question are color contents, the move that was available earlier about a set of representational contents that capture what stays the same is not available here. And that suggests for reasons that I just gave that '1 took this point t o be raised by some of t h e discussion of Lycan's paper a t t h e meeting of Sociedad Filos6fica IberwAmericana in Canclin in June, 1995. If I had t o credit it t o anyone, it would be Alan Gibbard.

2. MENTAL PAINTAND MENTAL LATEX

45

there is more to experience than its representational content. The defender of the view that memory is defective must blunt or evade the intuitive appeal of the first person point of view to be successful. It is no good to simply invoke the doctrine that experience is entirely representational. But the reply to the Inverted Earth argument as I presented it above does just that. It says that the memories of the representational contents are wrong, so the memories of the phenomenal characters are wrong too. But that is just to assume that as far as memory goes, phenomenal character is representational content. For the argument to have any force, there would need to be some independent reason for taking externalism about phenomenal memory seriously. We could dramatize what is question-begging about the argument by augmenting the thought experiment so that the subject understands the philosophical theories that dictate that his representational contents shift. Then, being careful, he will acknowledge that the thought he used to think with the words "The sky is blue" is not the same as the thought he now thinks with those words. And he will acknowledge something similar about the representational content of his perception of the sky. So in the new version of the thought experiment, he knows that the representational contents of his experiences have shifted. But that gives him no reason to back down from his insistence that there is no difference in the way the sky looked to him (in one sense of that phrase), that if he could have both experiences side by side he would not be able to discern a difference. Plainly, he is justified in saying that there is no difference in something, something we could call the phenomenal character of the experience of looking a t the sky. Another way to dramatize the point I've just been making about the externalist view of defective memory is by appeal to a different thought experiment. Different languages cut up the color spectrum differently (Berlin and Kay, 1969). Suppose I move to a different language community in a different country, knowing full well that they cut up the colors differently. I immediately resolve that I will reject my old language community and be loyal to the new one. I commit myself to deferring to their experts, not the experts of my old language community. If according to the standards of my new community, I have misapplied a color concept, I commit myself to agreeing that I have misapplied it (Burge, 1979). Perhaps visually experiencing something as red involves a comparison to a memory image. If so, we can suppose that the memory images I use are derived from my first few experiences of red things in my new country. Now on one way of thinking of color concepts, my color concepts

have changed in a flash. And thus the representational content of my color experience has changed. But there is no time here for "gradual" but unnoticed change in what it is like to see red. T h e externalist view of memory would dictate that my memory of what it was like to see ripe tomatoes just a few moments ago before my decision is wrong. Five minutes ago, before my decision to change loyalties, I was looking at a ripe tomato. Now I have made the decision and I am looking at the same ripe tomato. It looks just the same -as far as I can tell. Yet the externalist about memory has to say that I mis-remember. Now you may want to reject the view of color concepts I a m using in this paragraph (but not earlier) on the ground that it ties color concepts too closely to language and conscious decision-making. Perhaps you are right. My point is that if anything is wrong with the claim about this thought experiment that the representational contents of my experiences changed in a flash while their phenomenal characters stayed the same, it is this theory of color concepts, not taking memory at face value. It will be useful to briefly consider a related objection to the Inverted Earth argument. Suppose it is said that the subject's (that is, your) representational contents don't ever switch. No matter how long you spend on Inverted Earth, the sky still looks blue t o you. After all, I have insisted that you notice no difference. This line of objection has more than a little force, but it can easily be seen not to lead away from my conclusion. For it is hard to see how anyone could accept this objection without also thinking that the subject (viz, you) on Inverted Earth has an "inverted spectrum" relative to the other denizens of Inverted Earth. The sky looks yellow to them but blue to you. And you are as functionally similar to them as you like. (We could even imagine that your monozygotic twin brother is one of them.). But once such easy generation of an inverted spectrum is allowed, we can imagine it happening on earth. Twins are raised, one with, the other without, inverting lenses. They end up functionally and representationally identical, but phenomenally different. And that provides an even clearer counter to representationism than the Inverted Earth argument.' 'clearer because it is a case of same representational content, different phenomenal character, and this yields a more direct argument t h a t phenomenal character goes beyond representational content. Inverted earth provides t h e converse, same phenomenal character, different representational content. And so I have had t o resort t o a burden of proof argument. I have challenged t h e representationist with a question: what kind of representational content of experience stays t h e same?

2. MENTALPAINT A ND MENTALLATEX

47

It may be objected to this last point that even if raised on earth, the twins are not representationally identical. It may be said that the sky looks blue to one and yellow to the other, so they are representationally different. But if inverted spectra are rife (and given the huge variation in genes for cone pigments, there is reason to expect substantial differences in color perception among normal perceivers) then why should we suppose that the sky looks blue to some of us, but looks green to others and looks yellow to others? Why should it be supposed that only some of us perceive the sky veridically? Better to suppose that phenomenal character is not expressible in English. Phrases like 'looks blue' should be used to characterize all normal perceivers who are seeing something as blue (that is, whose intentional content of perception is as of blue), even if their spectra are shifted or inverted with respect to one another. Now we are in a position to counter another of Lycan's (1996a, 199613) arguments. He notes that I concede that our Inverted Earth subject has experiences whose representational contents (on looking at the Inverted Earth sky) shift from looking blue to looking yellow. And he concludes that the subject's claim that he notices no difference is thereby undermined. Not so, for "looks blue" does not express a phenomenal character but rather a representational content! Phenomenal content is what is relevant to noticing a difference; representational content can change purely externally. I began the discussion of externalism by discussing a thought experiment involving Putnam's Twin Earth. The idea was that I had emigrated from Twin Earth to Earth and that after many years on Earth the representational contents of my visual experiences of the liquid in the oceans shifted even though the phenomenal character of the experiences stayed the same. I noted that there is no problem for the representationist here, since the constant phenomenal character can be understood in representational terms. However, there is no corresponding move available to the representationistin the case of an emigration to or from Inverted Earth. And this puts the burden of proof on the representationist.g

BIBLIOGRAPHY Berlin, Brent & Kay, Paul, 1969: Basic Color Terms. Berkeley: University of California Press. '1 am grateful for discussions with Tyler Burge, Brian Loar, Paul Horwich, Pierre Jacob and Georges Rey and t o Bill Lycan and his NEH Summer Seminar, 1995. I am grateful to Burge, Lycan, Rey and to Sydney Shoemaker for helpful comments on an earlier draft.

Block, Ned, 1980: "Troubles with Functionalism", in Block (ed.), Readings i n Philosophy of Psychology Vol. 1, Cambridge: Harvard University Press. Block, Ned, 1981: "Psychologism and Behaviorism", The Philosophical Review LXXXX, 1, January, 1981, 5-43. Block, Ned, 1990: "Inverted Earth", in James Tomberlin (ed.), Philosophical Perspectives 4, Action Theory and Philosophy of Mind, 53-79. Atascadero: Ridgeview. Block, Ned, 1994a: "Qualia". In S. Guttenplan (ed.), A Companion to Philosophy of Mind. Oxford: Blackwell, 514-520. Block, Ned, 1994b: "Consciousness", in S. Guttenplan (ed.), A Companion to Philosophy of Mind. Oxford: Blackwell, 209-218. Block, Ned, 1995a: "On a Confusion about a Function of Consciousness", Behavioral and Brain Sciences 18, 227-247. To be reprinted in Block, Flanagan and Guzeldere, 1996. Block, Ned, 1995b: "How Many Concepts of Consciousness?", Behavioral and Brain Sciences 18. Block, Ned & Fodor, Jerry, 1972: "What Psychological States are Not", The Philosophical Review LXXXI, 2, 159-181. Block, Ned, Flanagan, Owen & Guzeldere, Guven, 1996: The Nature of Consciousness: Philosophical and Scientific Debates. Cambridge: MIT Press. Boghossian, Paul & Velleman, David, 1989: "Color as a Secondary Quality", Mind 98, 81-103. Burge, Tyler, 1979: "Individualism and the Mental", Midwest Studies i n Philosophy IV, 73-121. Crane, Tim, 1992: The Contents of Experience. Dennett, Daniel, 1991: Consciousness Explained. Boston: Little Brown. Dretske, Fred, 1995: Book. Cambridge: MIT Press. Dretske, Fred, 1996a: "Phenomenal Externalism or If Meanings Ain't in the Head, Where are Qualia?", in this volume. Harman, Gilbert, 1982: "Conceptual Role Semantics", Notre Dame Journal of Formal Logic 23. Harman, Gilbert, 1990: "The Intrinsic Quality of Experience", in James Tomberlin (ed.), Philosophical Perspectives 4, Action Theory and Philosophy of Mind, 31-52 Atascadero: Ridgeview. Harman, Gilbert, 1996: "Explaining Objective Color in Terms of Subjective Reactions", in this volume. Humphrey, N. 1992: A History of the Mind. New York: Simon & Schuster. Jackson, Frank, 1982: "Epiphenomena1 Qualia" , Philosophical Studies 32, 127-36. Lewis, David, 1990: "What Experience Teaches", in Lycan, 1990.


49

Loar, Brian, 1990: "Phenomenal Properties", in James Tomberlin (ed.), Philosophical Perspectives 4, Action Theory and Philosophy of Mind. Atascadero: Ridgeview. Lycan, William G., 1987: Consciousness. Cambridge: MIT. Lycan, William G., 1990: Mind and Cognition, A Reader. Oxford: Blackwell. Lycan, William G., 1995: "We've Only Just Begun", Behavioral and Brain Sciences 18, 262-263. Lycan, William G. 1996a: "Layered Perceptual Representation", in this volume. Lycan, William G. 1996b: Consciousness and Experience. Cambridge: MIT Press. McDowell, John, 1994: "The Content of Perceptual Experience", Philosophical Quarterly April. McGinn, Colin 1991: The Problem of Consciousness. Oxford: Blackwell. Peacocke, Christopher, 1983: Sense and Content. Oxford: Oxford University Press. Rey, Georges, 1992a: "Sensational Sentences", in Consciousness: Psychological and Philosophical Essays, ed by M. Davies and G. Humphreys. Oxford: Blackwell. Rey, Georges, 1992b: "Sensational Sentences Switched", Philosophical Studies 68, 289-331. Rosenthal, David, 1991: The Nature of Mind. Oxford: Oxford University Press. Shoemaker, Sydney, 1982: "The Inverted Spectrum", The Journal of Philosophy 79, 7 , 357-81. Shoemaker, Sydney 1994a: "Self-Knowledge and Inner Sense; Lecture 111: The Phenomenal Character of Experience", Philosophy and Phenomenological Research VIV, 2, June 1994, 291-314. Shoemaker, Sydney, 1994b: "Phenomenal Character", Nous 28, 21-38. Stalnaker, Robert, 1996: "On a Defense of the Hegemony of Representation", in this volume. Strawson, Galen: Mental Reality. Cambridge: MIT Press. Tye, Michael, 1994: "Qualia, Content and the Inverted Spectrum", Nous 28: 159-183. Tye, Michael, 1995a: "Blindsight, Orgasm and Representational Overlap", Behavioral and Brain Sciences 18, 268-269. Tye, Michael, 1995b: Ten Problems of Consciousness. Cambridge: MIT Press. White, Stephen, 1995: "Color and the Narrow Contents of Experience", in Philosophical Topics.

PHILOSOPHICAL ISSUES, 7 Perception, 1995

Orgasms Again Michael Tye

Ned Block, in his reply to Gil Harman, raises some interesting issues for representationists to consider.' I do not believe that my own version of representationism (see here Tye 1995 for the full view) is presented with any insuperable difficulties by the various objections and examples Block adduces. But I lack the space to argue for this general claim here. Instead, let me focus upon a specific demand that Block makes of representationists. Block says (at the end of the section entitled "Latex and Orgasm"): The representationist should put up or shut up. The burden of proof is on them to say what the representational content of experiences such as orgasm and pain are. On one way of understanding this demand, it is reasonable enough. On another, it is too strong. This needs a little sorting out. Representationism, in my view, is a theory that is justified by its capacity to explain a variety of pieces of data about phenomenal consciousness in a simple, straightforward way, and to solve a number of perplexing philosophical problems (I distinguish ten in Tye 'The term "representationist" is Block's. See his "Mental Paint and Mental Latex" in this volume.

1995). On my account, what it is exactly that a given experience or feeling represents need not be accessible to the subject's cognitive centers, including his or her powers of introspection, except in the most general and uniformative way (for example, as an experience of this sort). Nonetheless no two states that differ in the relevant representational contents can differ phenomenally, I claim; moreover, any two states that are alike in the relevant contents must be alike phenomenally. Why? Because phenomenal character is one and the same as representational content of the appropriate sort. What is the relevant sort of content? Here I agree with Block that the representationist should indeed put up or shut up. For without a further story at this level, there is no clearly defined theory.2 But to concede this is not to grant that the representationist needs to specify what the pertinent content is in particular cases. T h a t , I argue next, is a matter for empirical investigation. Sensory experience is often contrasted with judgement or belief or thought. States of the latter sort demand concepts on the part of their possessors. I cannot think that the violin is being played or judge that the man most likely to succeed is N.N., for example, unless I possess the concepts, violin, man, and success, among others. These concepts are then exercised in the judgement or thought or belief. But sensory experience, in its most basic form, doesn't seem to be like this. Our experiences are imbued with a wealth of detail that far outstrips our conceptual resources. For example, the experience or sensation of the determinate shade of color, red2g, is phenomenally different from that of the shade, reds2. But I have no such concept as red2g. So, I cannot see something as red29 or recognize that specific shade as such. My ordinary color judgements are, of necessity, far less discriminating than my experiences of color. Human memory simply isn't up to the task of capturing the wealth of detail found in the experiences. Beliefs or judgements abstract from the details and impose more general conceptual categories. Sensory experience is the basis for many beliefs or judgements, but it is far, far richer. This point is not restricted to color, of course. Experiences of shapes can likewise go beyond our concepts. Presented with an inkblot, for example, I will likely have an experience of a shape for which I have no matching concept. %n my view, t h e relevant sort of content is PANIC ( s u i t a b l y ) poised, abstract, nonconceptual, intentional content. See here Tye 1995 and Tye 1996. My conception of what it is for a state t o be poised, I might add, is not t h e same as Block's in his 1995. In what follows, I focus on nonconceptual content without further qualification. But I am certainly not suggesting t h a t any state with nonconceptual content is phenomenally conscious.

Of course, the ways in which things are conceived sometimes have a n effect upon how they phenomenally appear. Conceiving of a scene in one way rather than another may sometimes influence how I break it up cognitively into spatial parts, for example, and the shapes I then experience may not be the ones I would have experienced under a different conceptualization. But the sensory experiences of shapes (at the most basic level) do not require concepts. In this way, they, and other sensory experiences, have nonconceptual contents. Sensory experiences, broadly understood, often have conceptual contents as well as nonconceptual ones.3 I cannot see that the dog is wagging its tail or hear that the soprano is off-key unless I possess the concepts, dog, soprano, etc. One issue representationists sometimes disagree upon is whether these conceptual contents are themselves directly phenomenally relevant. My own view is that they are not. In cases like the two just mentioned, the states are compound, having both thoughts and sensations as components, and thoughts, I claim, have no intrinsic phenomenology.4 I do not deny, of course, that often when we think, we are subject to linguistic (or verbal) images, which have the phonological and syntactic structure of items in our native language. These images frequently even come complete with details of stress and intonation. It is as if we are speaking to ourselves. We 'hear' an inner voice. Images of this sort clothe our conscious thoughts; and it is, I maintain, in features of the nonconceptual contents of these images that the phenomenology sometimes attributed t o thoughts really resides. T h e images are not always present in thought, however. They are not essential to thought. Consider, for example, the thoughts of creatures without a natural language.5 Even if you disagree with me here, it does not undermine my main point. Sensory experiences have nonconceptual contents. Representationists are free to appeal to these contents as phenomenally relevant (and, in my view, they should make such an appeal). But representationists need not be in a position to say just what enters into these contents in specific cases. I do not know what it is exactly that my orgasms nonconceptually represent, for example. I can 3 ~ I snoted in my reply t o Bill Lycan (this volume), perceptual experience has many Some. but not all, of these layers are nonconce~tual. - lavers. " 4Harman seems t o take t h e opposing view of thoughts. 'Block (1995) mentions two cases which may appear problematic for nonconceptual representationism (one involving visual experience as of a building versus visual e x ~ e r i e n c eas of a facade and another involving ~ h e n o m e n a ldifferences in hearing a language one understands versus hearing t h e same language without understanding it). I respond t o these in Tye 1996.

".

make some very general remarks, and I did so in the passage Block quotes in his paper, but I never claimed that they told the whole story. Clearly, my orgasms represent something much more specific than just that something very pleasing is going on down there. They also nonconceptually represent a certain sort of physical change in my body. This is crucially relevant to their phenomenal character. Orgasms, at heart, are bodily sensation^.^ Like other bodily sensations, under optimal conditions (conditions of well functioning), they 'track' bodily states, just as perceptual sensations 'track' environmental state^.^ This, in my view, is how they get their nonconceptual contents. Which states are tracked is not t o be settled a priori. Still, it seems to me plausible t o suppose that orgasms track certain physical changes in the genital region (just as the feeling of thirst tracks dryness in the throat; hunger pangs track contractions in the stomache walls when the stomache is empty; pains track tissue damage; and so on). Moreover, my orgasms track certain bodily changes in me. This, in part, is why I do not experience an orgasm when I see that my partner is having one. Here I represent something about her; moreover, my representation is conceptual. I form the belief that she is having an orgasm on the basis of associated visual sensations. Feeling an orgasm, however, requires the right sort of nonconceptual representation of the pertinent bodily changes, not conceptual representation of the generic state. No belief about myself or my partner is necessary. Nor is visual experience -I might be blind. That need not impoverish my bodily experience. Let us all enjoy our orgasms, then. There is nothing to trouble the representationist here.

Block, N. 1995 "On A Confusion about a Function of Consciousness", Behavioral and Brain Sciences, 18, 227-247. Tye, M. 1995 Ten Problems of Consciousness: A Representational Theory of the Phenomenal Mind, Cambridge, Mass: MIT Press, Bradford Books. Tye, M. 1996 "The Function of Consciousness", Nous forthcoming. Tye, M. forthcoming "Is Consciousness in the Head?".

re or mall^, when an orgasm is experienced, strong desires are also present. These have a characteristic felt quality t o them. One feels "pulled" or "tugged". This general feeling (which is also to be handled representationally) contributes t o t h e overall phenomenal character of one's state. needs t o be said here, of course. See Tye 1995 and Tye forthcoming.

re ore


Colors, Subjective Reactions, and Qualia Sydney Shoemaker

Let me begin by indicating where I think Harman and I are in agreement. We both think that "subjective reactions" must come into an account of color, although we have different views about how they do. We both think that perceptual experience has a "presentational or representational character", and that color is represented by our visual experiences as a feature of external objects, not as a feature of our experience. Moreover, we agree that, as Harman puts it, "color is experienced as a simple basic quality, rather than a disposition or complex of causal properties". As Harman emphasized in an earlier paper,' what we are introspectively aware of in our experience is its presentational or representational content, not any "mental paint" which bestows this content. I shall refer to all of this as Harman's "phenomenological point". Because we agree on this, we also agree that if his characters George and Mary were spectrum inverted relative to each other, supposing that to be possible, this would have to '"The Intrinsic Quality of Experience", in James Tomberlin (ed.), Philosophical Perspectives, 4 , Action Theory and Philosophy of Mind. Ascadera, CA: Ridgeview Publishing Co., 1990. 31-52.

involve their perceiving the same objects as having different properties, this despite the fact that as normal perceivers they would perceive these objects as having the same colors. And I think we agree that in this case the properties would have to be relational ones, defined or constituted by their relations to the experiences of the subject perceiving them. On all these points I think we are in substantive agreement. There is another point on which our agreement is, I think, merely verbal. We agree that there are no "color sensations". But for me this is just the point that the current meaning of "sensation" makes it an inappropriate term for our color experiences, while for Harman the rejection of color sensations seems to be linked with the rejection of "qualia". Now in agreeing with Harman's phenomenological point, I agree with him in rejecting one conception of qualia -the "mental paint" conception, according to which we are aware of the representational content of our color experiences by being aware in a quasi-perceptual way of their qualia, in much the way we are aware of what colors are represented in a painting by perceiving the pigments on the canvas. But I think that there is another conception of qualia on which they are required by a satisfactory account of color experience. Finally, there is one point of difference between us that is not quite, or not yet, a point of disagreement. Harman says that he is "not sure" whether the imagined possibility of spectrum inversion is coherent, whereas I am quite sure that it is coherent. One of the main things I want to do in what follows is to show that if Harman were to climb off the fence on this issue, and climb down in my direction, then given our other points of agreement he could not stop short of accepting the rest of my view, qualia and all. Along the way I will show that the circularities Harman claims to find in my view are simply not there; they seem to be there only because, perhaps out of misguided charity, he states my view in a way that omits its commitment to qualia. That is truly a case of Hamlet without the Prince of Denmark. So let me start laying out the argument. Since my object is to show what Harman is committed to on the assumption that spectrum inversion is possible, I assume that it is possible. Assuming Harman's phenomenological point, we have 1) If George and Mary are normal perceivers who are spectrum inverted relative to each other, then when George sees a red object it looks to him to have a property, call it Q, which green objects look to Mary to have, and when Mary sees a red object

4. COLORS, SUBJECTIVE REACTIONS, A N D QUALIA

57

it looks to her to have a property, call it T, which green objects look to George to have. But Harman thinks, as I do, that 2) If George and Mary are normal perceivers, then (under normal conditions) red objects look red to both and green objects look green to both.

It follows that 3) The Q and T referred to in (1) are distinct from redness and greenness.

It is compatible with (3) that Q and T are in fact not properties that visible objects ever have -that while George and Mary always perceive red and green objects as having one or the other of these properties, their color experiences are to that extent illusory. There are two versions of this view. One, which I call "literal projectivism", says that Q and T and the like are in fact properties of experiences, qualia in fact, which are projected onto external objects. So on this view we do perceive "mental paint", but perceive it (misperceive it) as belonging to external things. I take this to be the view that Paul Boghossian and David Velleman have advanced in several papers. On their version of the view it is colors that are falsely but unavoidably projected onto objects by our experience, and this conflicts with (1) and (2), which presuppose that objects literally do have colors. But there could be a version of literal projectivism which allows that external objects can be colored but holds that our visual experience projects onto objects properties -Q and T, and their ilk- which in fact are instantiated only in our minds. A different version of projectivism is what I have called "figurative projectivism"; according to this, Q and T would be properties that, are in fact instantiated neither in external objects nor in our minds, but which certain experiences falsely (but unavoidably) represent external objects as instantiating. This seems to be the view that J.L. Mackie endorses and ascribes to Locke in his book Problems From Locke. But I take it that Harman agrees with me in rejecting all such projective error views. So, still assuming for the sake of discussion the possibility of spectrum inversion, he would agree that 4) Under normal circumstances, when George or Mary perceives an object to be Q or T, that object is Q or T.

This means that when both George and Mary are looking at a red object under normal circumstances, that object is both Q and T .

Now one could perhaps tell a story according to which in addition to their colors objects have other intrinsic properties -or other properties with as good a claim to being intrinsic as colors have- and that Q and T are among these, and that for some reason George can perceive something to have Q only when the thing is red while Mary can perceive something to be Q only when it is green, and so on. But such a story would have no plausibility at all as applied to actual objects. I take Harman to agree that given the nature of the actual world, the only plausible view about the nature of Q and T is

5) Q and T are relational properties that objects have in virtue of their relations to the visual experiences of subjects viewing them. It should be emphasized that (5) does not mean that Q and T are perceived as relational properties. Harman says that it is compatible with colors having a complex nature that "the concept of color as it figures in [perceptual representation] is simple and unanalysable in causal terms, because color is experienced as a simple basic property"; and the same could be said about the concepts of Q and T, as they occur in our perceptual representations.' And certainly one can perceive what is in fact a relational property without perceiving it as relational. When a n object feels heavy to one, one does not perceive it as having a property that is relative to people of one's own build and strength, yet it is true, even conceptually true, that the heaviness one feels is such a property. The question now becomes, what relation is involved in such properties as Q-ness and T-ness, and what it is a relation to? Elsewhere I have used the term "phenomenal property" for properties of the sort Q and T are supposed to be, and I have suggested that a phenomenal property is a property an object has in virtue of actually causing, in a certain sort of way (the way involved in normal cases of vision), an experience having a certain qualitative character, i.e., having a certain quale. The suggestion Harrnan attributes to me is different. The relevant difference is that instead of specifying the mental term in the relationship in terms of qualitative character, it specifies it in terms of representational content. So, in the case of 2 ~ a t h e trhan say t h a t our experience represents colors, and properties such as Q and T, as simple, it would be better (if we want t o avoid a projective error theory) t o say t h a t it does not represent them as complex. Likewise, rather than

say t h a t our experience represents such properties as mo~ladic,it is better t o say t h a t it does not represent them as relational. Here our experience is best thought of as noncommittal about such matters.

the phenomenal property of being Q, this would be constituted by a relation to an experience that represents the object as Q. But of course this makes the account circular. As Harman puts it, "In order to understand what the concept Q is, we need to understand what objects are Q to someone, but in order to understand what objects are Q to someone, we need to understand what the concept Q is" But that circularity has nothing to do with me, because the view that is afflicted with it is not mine. In a footnote Harman says that he has "oversimplified", and goes on to say "Shoemaker's actual account supposes that an experience has a certain intrinsic phenomenal feature x that is responsible for its representing something as Q. For something to be Q is for it to be such as to produce experiences with feature x". But the view ascribed to me in the footnote, unlike the view ascribed to me in the text of Harman's paper, is not circular. So "oversimplified" isn't quite the right word. Let me take stock. We have seen that Harman seems committed to holding that if spectrum inversion is possible then color percep tion involves the perception of relational properties of objects that are defined, or constituted, by relations to visual experiences. And we have seen that it is unacceptable, because circular, to specify the mental terms of these relationships, i.e., the experiences, in terms of what properties of this sort they represent. It seems obvious that there is no prospect of our being able to specify these experiences in terms of any other representational properties they have. So it seems that it must be the case that having such a property consists in being appropriately related to experiences having a certain nonrepresentational property. I.e., 6) Q and T are properties objects have in virtue of being related in certain ways to experiences having certain nonrepresentational properties. It seems to me that Harman is committed to its being the case that (6) is true if spectrum inversion is possible and the story about George and Mary is coherent. But (6) is my view. And properties that can play the role that (6) assigns to the nonrepresentational properties are what I mean by qualia. So I think that Harman is committed to allowing that if spectrum inversion is possible, there are qualia as I understand them. Let me guard against a misunderstanding. (6) of course will not do as a definition of the terms "Q" and "T". For it does not say what the nonrepresentational properties in question are; it just says that there are such and (by implication) that the one associated with Q is different from the one associated with T. What (6) says, generalized,

is that for each different phenomenal property we can perceive an external thing to have, there is a different nonrepresentational property of perceptual experiences, such that having that phenomenal property consists in standing in the appropriate relation to an experience having the corresponding nonrepresentational property. It is a further part of my account that when an experience has a nonrepresentational property of this kind, it represents the experienced object as having the corresponding phenomenal property, and does so in virtue of having that nonrepresentational property. Supposing 'Q' to name such a phenomenal property, there is no suggestion here that the appropriate way to "define" 'Q', in the sense of explaining its meaning, is by specifying the corresponding nonrepresentational property. I might define it, in the sense of fixing its reference, by saying that it is the phenomenal property my experience of red things represent them as having. And while there may conceivably come a time when we can specify the nonrepresentational properties in neurophysiological terms, the only way of picking them out that we are likely to have in the foreseeable future is by means of such descriptions as "the nonrepresentational property that underlies my experience of red things". Although Harman says that he is uncertain whether the idea of spectrum inversion is coherent, he takes the alleged circularity of my account as raising problems for it. And he mentions other reasons for rejecting qualia. Since the account he finds circular is not mine, and since the notion of qualia he attacks (the mental paint view) is not mine either, I find nothing in his paper that threatens either the claim that spectrum inversion is possible or my account of it in terms of phenomenal properties and qualia. But what is there to be said in favor of my account? As is implied by what I have said already, any case for the possibility of spectrum inversion will at the same time be a case for my account. At any rate, Harman should think so, for given assumptions I share with him, my account gives the only way of accommodating spectrum inversion. I have believed for some time that reflection on the possibility of intrasubjective changes in color experiences provides strong support for the claim that the notion of intersubjective inversion is at least ~ o h e r e n t .The ~ conceivability of a partial intrasubjective inversion -where the appearance of things of most colors remains unchanged, but two narrow bands of colors change place in 3See my "The Inverted Spectrum", Journal of Philosophy, 79, 7, 1982, 357-81. Also my L'Lovelyand Suspect Ideas", Philosophy and Phenomenological Research, LIII, 4 , 1993, 905-910.

4. COLORS, SUBJECTIVE REACTIONS, A N D QUALIA

61

the subject's color quality space- should be uncontroversial. Such a change would involve changes in the subject's discriminatory and recognitional capacities that could be detected without relying on the subject's testimony. But it would seem that a series of such partial inversions could add up to a total inversion. And I think that if one allows the coherence of the supposition that there could be a total intrasubjective inversion, one will be hard pressed to find plausible reasons for denying the coherence of the supposition that there could be total intersubjective inversion. But rather than pursue this line of thought further, I want to argue more directly for the view that a satisfactory account of perceptual experience requires qualia. While, as I have argued, the existence of qualia is implied by the view that spectrum inversion is possible, the converse is not true. I myself am prepared to allow that it may be that the structure of our color quality space is asymmetrical in ways that make it impossible that there could be behaviorally undetectable spectrum inversion for creatures having such a quality space. The asymmetries I have in mind include the fact, or alleged fact, that if we consider the four "unique hues", pure red, pure yellow, pure green, and pure blue, there are fewer discriminable shades between the last two than between the first two. This has the consequence that there is no mapping that maps each unique hue onto a different unique hue, maps any two discriminable shades onto two other discriminable shades, and preserves the similarity ordering of the hues. That our color quality space is asymmetrical in this way -if it is- does not rule out the possibility of there being creatures, otherwise very much like ourselves, whose color quality space is not asymmetrical in such ways and for whom undetectable spectrum inversion is possible. But I want to focus on what it is for there to be, or not to be, such asymmetries. The existence or nonexistence of such asymmetries does not have to do with the structure of the system of colors, thought of as mind independent properties. Nothing in the physical basis of colors in objects or in the light answers to the distinction between unique and nonunique hues, and there are no relations between the physical bases of colors in objects or in the light that fix the similarity relations between colors as we perceive them. We are talking here about something "subjective" -the color "quality space", to use Quine's term, of creatures having a certain sort of perceptual system. A creature's quality space can be thought of as (among other things) imposing a similarity ordering on possible stimuli; the position of stimuli in that ordering will have to do with such things as whether they are discriminable by that creature, and, if they are, how easy or

difficult such discrimination is -also, as Quine points out, with what sorts of inductions the creature is apt to engage in. We know that in the case of color there are stimuli that are physically very different -involving combinations of wavelengths that are metamers- that occupy the same place in the similarity ordering of creatures like us. What I want to suggest is that the notion of a quale is implicit in the very notion of a quality space. Harman has always been willing to admit that our perceptual experiences have, in addition to the representational properties we are introspectively aware of, nonrepresentational properties that in some sense underlie these. These, he says, are physical (physiological) properties to which we have no introspective access. But he would surely allow that there is no a priori reason to think that whenever two visual experiences are phenomenally exactly the same, their physiological properties are the same; as a good functionalist, he presumably allows that being an experience of a certain phenomenal type is "multiply realizable" at the physical level. And assuming physicalism, having a quality space that imposes a similarity ordering on external stimuli must surely go with having a quality space that imposes a similarity ordering on the physical states of the creature that realize the perceptual experiences produced by the stimuli in the first quality space. This will group these states into equivalence classes, the members of each of which all realize experiences of a single phenomenal character, and it will order these equivalence classes in a way that corresponds to the phenomenal similarity ordering of the experiences realized by their members. Now there is a very strong sense in which we lack introspective access to individual physical states of these kinds. While there is a sense in which our perceptual system is sensitive to the presence of such a state, it is not sensitive in the same way to its absence, since it will be in the same conscious perceptual state whether it is in that state or in some other state belonging to the same equivalence class. But for each of these equivalence classes there is a higher-order state a creature is in just in case it is in some state or other belonging to that class. In the case of these higher-order states, there is sensitivity both to their presence and to their absence. Insofar as one is aware of whether two experiences are phenomenally the same or different, one is sensitive to whether they are realized by members of the same equivalence class or by members of different ones. As a first approximation, one could say that qualia are these higher-order states. Equivalently, they are functional states corresponding to these equivalence classes -functional states of which the members of the corresponding equivalence class are realizations.

4. COLORS, SUBJECTIVE REACTIONS, AND QUALIA

63

The reason this is just a first approximation is that experiences are phenomenally similar and different along a number of different dimensions. Experiences that are phenomenally different may be phenomenally exactly alike on one dimension while differing on another. So we need a more fine grained account. Not only must there be a similarity ordering on the physical states that realize experiences having complex contents; there must also be a similarity ordering on the physical properties of these states that contribute to their similarity or difference along particular dimensions. It will be equivalence classes of these, as grouped by such an ordering, that correspond to individual qualia.4 What all of this requires is something I have long insisted on that there must be functionally characterizable relations of phenomenal similarity and difference amongst experiences. Roughly, these are the functional relations between experiences that bestow the behavioral capacities, the recognitional and discriminative capacities, and the dispositions to belief, that go with having a certain quality space in Quine's sense. These similarity and difference relations are closely related to, but must be distinguished from, what might be called relations of "intentional" similarity and difference between experiences -i.e., relations that hold between experiences in virtue of similarities and differences in their representational content. The phenomenal similarities and differences significantly constrain what relations of intentional similarity and difference there can be, but they do not by themselves determine them. For the relations of intentional similarity and difference are partly determined by relations to the environment. Supposing that the neural states of a brain in a vat could duplicate mine over an interval, it would have states that stand to one another in the same relations of phenomenal similarity and difference as mine do, but it would not necessarily have states that stand in the same relations of intentional similarity and difference as mine do -if it were a brain that had always been in a vat, its experiences would not be of color in the sense in which mine are, and so could not be similar or different with respect to what colors they represent. However, given a background of externalist constraints on content, the qualitative similarities and differences can be said to determine the intentional similarities and differences; and, with the same qualification, the qualitative character of an experience can be said to determine its representational content. 4See my "The Phenomenal Character of Experience", Lect. 111 of "Introspection and 'Inner-Sense"', Philosophy and Phenomenologzcal Research, LIV, 2, 1994, 249-314.

I have said that we are "sensitive to" the presence or absence of qualia in our experience. In saying this I a m not reneging on my acceptance of Harman's phenomenological point and returning to the "mental paint" conception of qualia. In part this sensitivity is a subpersonal affair; it amounts to the fact that (with the qualification just mentioned) the qualia instantiated in our experiences, and the relations of phenomenal similarity and difference between them, determine their representational contents and the relations of intentional similarity and difference between them. What in the first instance we are perceptually aware of are properties of things in our environments, and similarities and differences with respect to these; and what in the first instance we are introspectively aware of is the representational content of our experiences, i.e., what properties, similarities and differences there appear to be in our environments. That is what I am calling Harman's phenomenological point. But on reflection we can come to realize that the intentional similarities and differences we are aware of are grounded in phenomenal similarities and differences, and that these are grounded in the nonrepresentational features of experiences I a m calling qualia. So there is a sense in which we can be aware of the latter; knowing what intentional similarities and differences there are, we can know what phenomenal similarities and differences there must be. But this is a far cry from the quasi-perceptual awareness required by the mental paint conception of qualia. I should also say that in speaking as I have done of qualia, or experiences of phenomenal types, as "multiply realizable" I am not going back on my denial, argued elsewhere, that individual qualia are functionally definable. That denial was premised on the assumption that "qualia inversion" is possible. It is compatible with this that the notion of being a quale is functionally definable, and that is enough to give sense to speaking of qualia as multiply realizable, e.g., by neurophysiological states or properties. What the denial does imply is that we cannot specify an individual quale in purely functional terms. Any such specification must involve an indexical element, or an element of reference-fixing a la Kripke. For example, a particular quale might be specified as the one that characterizes experiences of red things in creatures neurophysiologically like me. What does this view say about the nature of color? Given that I think that experiences of the same color can be qualitatively different e i t h e r in different persons, or in the same person at different t i m e s I obviously cannot accept any view that defines colors in terms of the sorts of experiences they are apt to produce. (It should be noted that one doesn't have to believe in the possibility of be-

4. COLORS, SUBJECTIVE REACTIONS, AND QUALIA

65

haviorally undetectable spectrum inversion in order to believe in this possibility. It is clear, I think, that there can be creatures who perceive at least some of the same colors as we do but have differently structured color quality spaces -i.e., differ from us in the similarity and difference relations they perceive between things of the various colors. Clearly, the experiences of at least some colors would have to be qualitatively different in these creatures than they are in us.) What ought to be uncontroversial is that color similarity and color difference are "subjective" in the sense of being perceptual system relative. And perhaps the property of being a color (or a color "realization") is likewise perceptual system relative -maybe there are reflectance properties of objects that are colors for bees and not for us, and ones that are colors for us and not for creatures of some other species. As for our concepts of particular colors, I think I must assign them a bit more complexity than Harman does. Consider again George, who perceives redness by perceiving the phenomenal property Q. George, if he is an ordinary person, won't have a name for Q -the only name he has available for describing his of-Q experiences is "red". But if he is willing to listen to reason, he can come to appreciate that his concept of Q -looking like that - and his concept being red are different. For those of us who can see, the reference of a particular color word gets fixed by reference to whatever phenomenal property is involved in perceiving things with that color. But we take the property the word refers to be one that objects have independently of whether they are being viewed, and we take ourselves to be entitled to say that other persons, and other animals, are perceiving something to be red when we are not in a position to have an informed opinion as to whether their experiences are phenomenally like the ones we have when we view red things. Harman says that objective color "is plausibly identified with a tendency to produce a certain reaction in normal perceivers, where the relevant reaction is identified in part with reference to the mechanisms of color perception". What if normal perceivers were divided equally into the likes of George, who perceives red by perceiving Q, and the likes of Mary, who perceives red by perceiving T? It seems to me that it would then be true to say that red can be "identified" both with the tendency to produce of-Q perceptions in George-like creatures, and with the tendency to produce of-T perceptions in Mary-like creatures. Maybe "identified" isn't the right word; for how can one thing be identified with two different things? But if whatever has the one tendency will necessarily have the other, perhaps "identified" can be allowed to stand. In any case, neither of

these "identifications" will be privileged. If in fact everyone is like George, or everyone is like Mary, then we can identify red with a tendency to produce a certain reaction in normal perceivers. But even still this identification will not be privileged; for it may be that red can also be identified with a tendency to produce some quite different reaction in bees. Yet another way of saying what red is will involve picking some object or kind of objects -e.g., ripe tomatoesand saying that something is red just in case it is, for creatures with our sort of quality space, indistinguishable from such objects with respect to color, where indistinguishability with respect to color is in turn explained in terms of indistinguishability simpliciter in certain circumstances, namely those in which other properties are masked. None of these ways of picking out red gives a meaning equivalence; all are "reference-fixing~"that depend on synthetic truths about the world. And I think that is all we can expect. Here I don't expect Harman to disagree. As best I can tell, such differences as there are between Harman and me about color, and about the concept of color, go back to the differences about color experience discussed earlier.5

5My thanks to Harold Langsam for comments on an earlier draft of these remarks.


Is Color Psychological or Biological? Or Both? Ernest Sosa

'Objective color' means the color of an object in a broad sense of 'object'. To this definition Harman adds an assumption -the observer relativity of objective color:

ORC [We] ". . . have to explain facts about objective color in terms of facts about perceivers" (2).'

I grant both the definition and the assumption. A familiar way of implementing ORC is then considered, which I label color subjectivism: CS An object is red if and only if perception of it would give normal perceivers red sensations under standard viewing conditions. This proposal requires three explanatory notes, a s follows: El T he sense of 'red' that applies to sensations is different from the sense that applies to objects (3).2 'Parenthetical references are all to Gilbert Harman, "Explaining Objective Color in Terms of Subjective Reactions". 'Harman adopts Christopher Peacocke's device of using the term red for the sense of 'red' that applies to sensations.

E2 For the sake of argument we are assuming that there are objective criteria in terms of which we can understand normalcy and standardness (3-4).

E3 The perception of color does not normally involve sensations in any ordinary sense of the term 'sensation' (8). I also grant these three claims. Given E3, accordingly, when I speak of "color sensations" or of "visual sensations", the word 'sensation' is used in a technical sense in which it amounts to 'impression' or 'experience'. Thus Harman sets up a framework for discussion of objective color, and his discussion is at many points insightful. What follows will be restricted to his main thesis and his main line of argument, and here, fortunately for my role a s commentator, I do have some disagreement, or anyhow some questions to raise. However, there is much in the paper that deserves more discussion than I will be able to give it.

Here is a main thesis of Harman's paper: IT For something to be red is for it to have a tendency to have a certain complex effect on a normal perceiver's sensory apparatus, in ways described by the scientific theory of color (12; cf. 1 8 ) . ~ To defend IT Harman attacks CS, with an argument that runs as follows in its main lines: H1 We consider first the qualitative conception of the redness of sensations. According to this conception, ". . . the notion of a red' sensation is the notion of a sensation "like this", where "this" refers to a sensation that one is actually experiencing or imagining" (5). 3However, in the end Harman's position is not entirely clear to me. His three concluding paragraphs seem compatible with the following view: that particular objective colors are tendencies to produce visual experience presenting or representing the environment as characterized in terms of corresponding simple concepts of (objective) color. But it is not clear to me how that view is related to the earlier IT: whether it is meant to replace I T or to supplement IT. In either case, the view courts vicious circularity. See the distinction between CS2 and CS3 below.

H2 Suppose now that different people have different sorts of sensations when perceiving objects they call 'red'. In that case a ripe tomato may have the power to produce sensations "like this" in me under standard conditions, but not in other normal perceivers in standard conditions. And the same may even hold true of every other perceiver (5-6). H3 On this, qualitative, version of CS, therefore, under supposition H2 we get such absurd results as that no objects have any colors (6)! H4 To avoid such absurdity, we must either (a) give up the qualitative version of CS, or else (b) rule out the possibility presented by supposition H2 (6). H5 Opting for H4a, Harman considers an alternative, "functional" conception of color sensations: ". .. a red' sensation would be a sensation that is typically caused by the perception of red things and that typically leads to the belief that one is perceiving something red" (7). H6 But when CS is combined with this functional conception of color sensations, it ". . . reduces to the claim that red objects are those that produce the sort of sensation that red objects produce" (8). And this is unacceptably circular and empty. H7 Harman next considers a second functional conception of color impressions) but dismisses it for beL L ~ e n ~ a t i (experiences, on~" ing, again, unacceptably circular (10-12). H8 Thus we reach eventually proposal IT (12). However, I T seems inappropriate as a response to the sorts of difficulties raised for CS. Just as it seemed possible that people have qualitatively different sensations when perceiving ripe tomatoes, so it seems also possible that they have different complex effects on their visual sensory mechanisms. In such a possible situation we would again have the absurdity that no objects have any colors! So, to avoid such absurdity, we must either give up the likes of IT, or else rule out the possibility that the effects on our visual mechanisms could differ as just supposed above. Presumably, Harman would now take the latter option. He would perhaps say that we have a right to assume that the biological (sensory mechanism) effects of perceiving ripe tomatoes (in standard conditions by normal perceivers) are relevantly uniform. That is to

say, he would hold that the effect on people's sensory mechanisms determinative of the redness of a ripe tomato is an effect that the ripe tomato would have uniformly on everyone normal when in standard conditions. However, the advocate of CS can now plausibly protest: hasn't he a similar right to assume that the psychological (qualitative) effects of perceiving ripe tomatoes (in standard conditions by normal perceivers) are also relevantly uniform? But if so, then the reasoning above can be blocked at step H5: at that stage one could simply opt for H4b rather than H4a. Harman does sketch an argument against option H4b. That option amounts, again, to ruling out the possibility, presented by H2, that different otherwise normal people have relevantly different sorts of sensations when they perceive ripe tomatoes in standard conditions. "This possibility may seem quite likely", we are told ". . . if the relevant aspect of a sensation is its intrinsic qualitative character, .. . if the intrinsic qualitative character of a sensation depends on the exact nature of the underlying physical events in the brain giving rise to the sensation, and . . . if there are differences in these underlying physical events from one brain to the next" (5). That seems the only argument offered in favor of opting for H4a, rather than H4b, at stage H5. Two questions arise about the argument, one general and one specific to Harman's overall project. First, suppose that indeed there are differences in the brain events that give rise to impressions with qualitative character in different people viewing ripe tomatoes. It does not follow that these differences make the brain events "relevantly different", i.e., different in such a way that the resulting impresssions will themselves have relevantly different qualitative character. Secondly, to the extent that there are such differences in the brain events and to the extent that these differences do seem "relevant", this would militate equally against I T as against CS. After all, I T has to postulate brain effects that are relevantly similar in normal perceivers who see ripe tomatoes in standard conditions.

There is a further charge of unfair treatment also available to the color subjectivist. For the attack on color subjectivism is based on the assumption that CS is a certain sort of analysis or definition of objective color terms. This assumption is crucial, for example, to stages H6 and H7 above. But the proposed I T is no such analysis or definition but only a n "identification" of objective color with a

tendency to produce certain biological effects in our sensory mechanisms. What has not been ruled out, then, is the similar "identification" of objective color with a tendency to produce certain psychological effects on the quality of our sensory experience. The attack on CS depends on viewing it as a certain sort of conceptual or meaning analysis. This may be seen, for example, in the following passages: The suggested analysis of objective color in terms of sensations implies that, if there are relevant differences in people's color sensations, then different people have different concepts of the colors of objects and do not mean the same thing by their color terms (5; my emphases). [Your].. . concept of a red object cannot be analyzed into your concept of a red' sensation, meaning the specific quality that your perceptual experience has in order to represent objective redness, because you have no such concept of a red' sensation. You have no idea what specific quality of your perceptual experience is used to represent objective redness (8). Compare with this last passage the following corresponding objection to IT: Your concept of a red object cannot be analyzed into your concept of a certain sort of complex effect on normal perceivers' sensory mechanisms. You have no idea what such complex effect is caused in normal perceivers by viewing red objects in standard conditions. Here no doubt IT is to be defended by stressing that it is no meaning or conceptual analysis. I t is rather a n account of the constitution of objective color properties such as the redness of a tomato. It tells us what such properties are in reality, and does not try to analyze our concepts of such properties, nor the meanings of our terms for such properties. But if the color subjectivist's CS is in competition with IT, then it should be viewed as a competing identification of such properties. How would CS fare under such an interpretation? First of all we could weaken CS to the following (where 'nsc-perceivers' abbreviates 'normal perceivers under standard conditions'):

CS1 For something t o be red is for it to have a tendency to produce a distinctive sort of qualitatively defined sensation (impression, experience) in nsc-perceivers. Of course this does not specify what the relevant quality is but neither does IT specify the relevant complex effect on one's sensory apparatus. We can however do a little better as follows:

CS2 For an object to be red is for it to give nsc-perceivers the qualitative sort of sensation distinctively produced in nsc-perceivers by red objects. This might be rejected as "unacceptably circular and empty", as at H6 above. However, CS2 has two very different readings: one does invite the objection of empty circularity; but the other seems safe from that objection. On that alternative reading CS2 is better formulated as follows: CS3 Consider the sort of sensation distinctively produced in actual nsc-perceivers by red objects. For an object to be red is for perception of it to give nsc-perceivers a sensation of that sort. With CS3 we first pick out a certain sort of sensation: sensation produced by red objects in nsc-perceivers. And we then go on to identzfy what it is to be objectively red as being a producer of that sort of sensation. Note that CS3 therefore has a certain advantage over IT. For CS3 at least is able to pick out the sort of effect it claims to constitute objective redness, at least in part. I T does not get that far. Of course a defender of I T could try for an improved I T as follows: IT1 Consider the sort of effect distinctively produced by red objects in the sensory apparatus of nsc-perceivers. For an object to be red is for perception of it to give nsc-perceivers an effect of that sort. Presumably Harman would reject CS3, arguing that there is no such thing as the intrinsic qualitative sort of any sensation. The qualities we are "aware" of in our visual experience are said to be qualities that our experience represents and not qualities that our experience exemplifies. But this is of doubtful relevance, since CS3 does not require us to be aware of the relevant qualities of our experience.4 If not thus, however, then how might one argue against special qualities 4Except perhaps (perhaps!) in some innocuous sense according to which simply by virtue of our experience's exemplifying or encompassing a quality, that quality thereby qualifies or constitutes our experiential "awareness", and in a corresponding sense we are "aware" of it. If there is such a sense, then of course we are after all aware of such a quality whenever our experience exemplifies or encompasses it. Someone might also conceivably use 'awareness' in such a way that whenever anyone is the subject of any thinking, desiring, sensing, and so on --of any psychological Xing of any sort- then by just having the thought, desire, sensation, or whatever, he is thereby "aware" of it. This might derive from the thought that just as one jumps jumps and smiles smiles, so one experi-

exemplified by or constitutive of our experience, ones that might be of use in a dispositional analysis of objective colors? It might be argued that the burden rests on the proponent of such qualities. This seems especially reasonable if we grant that we are not necessarily aware of such qualities through introspection. But, actually, what has been made plausible is only that we are not aware of such qualities directly, that we do not introspect them as such. Suppose now that our experience is constituted by some intrinsic ontological constitution -and that there are some distinctive qualities, of whatever sort, that are respectively constitutive of our experiences of various sorts. In that case, it would seem that when we are aware that we are having visual experience of something red (as if we saw a ripe tomato), we are in some sense indirectly aware of a sort of experience we are having: it is, say, experience distinctive of seeing something red. Of course we can sometimes be more sure, introspectively, that we are having such experience than that we are then seeing something objectively red. But we are sure of it under its indirect characterization and perhaps not directly; perhaps in such knowledge we do not characterize the experience as it is intrinsically, in itself. Still, for such distinctive experience to be allowable in a proposed analysis of objective color, it may be enough that we have some reason to reach certain conclusions about it, however we characterize it, whether indirectly or directly.If we may conclude that it does exist, that it comes in various sorts, and that the different sorts correlate with corresponding objective colors, perhaps that will suffice. One argument that there is such experience -whatever may be its intrinsic ontological constitution- is now that we are able to distinguish varieties of it, and that we are able to tell that a certain sort, as opposed to others, is present, when it is present, even without rendering a judgment as to the objective origins of that experience on that occasion. Indeed, such accessibility to introspection might even be made a hallmark of "experience" without requiring that the accessibility of the item in question to such introspection must be in terms of the properties that are ontologically constitutive of it. Under such a conception, a quality might qualify as both experiential ences experiences. And since experiencing may reasonably be considered a form of awareness, it follows that whenever one undergoes an experience, one thereby "experiences" it and is hence "aware" of it. And if the experience is constituted by certain sensory properties or ways of sensing that one exemplifies, then it might be argued that in being aware of the experience one is thereby aware of its constitutive properties. (I am not at all sure that there are actually in use any such senses as those required by the reasoning just displayed, but if there are such senses, then we need to put them aside as just indicated.)

and physical: the intrinsic ontological constitution of a distinctive sort of experience might turn out to be just physical, though also experiential. For it might be experiential just in virtue of being accessible to introspection when present, accessible at least indirectly, as being an experience as if there is something red before one, or the like. Which is more plausible: (a) to suppose that there is an experientially distinctive sort of visual experience caused in nsc-perceivers by the likes of ripe tomatoes, or (b) to suppose that there is a distinctive effect caused in the sensory apparatus of nsc-perceivers by the likes of ripe tomatoes? Color subjectivism in the form of CS3 may well opt for (a), while the identity theory in the form of IT1 may instead choose (b). I end with an irenic conclusion: In finding (a) more plausible one need not make any commitment on the nature or physical basis of color sensations (or impressions or experiences), nor on what constitutes or physically underlies the experiential character of such sensations. So it might even be that CS3 is true because IT1 is true.5

5My comments here are in line with my "Perception and Reality", in Information, Semantics, and Epistemology, edited by E. Villanueva (Oxford: Basil Blackwell, 1990), proceedings of the first SOFIA meeting, held in Tepoztlan, MCxico, in 1988.


Qualia and Color Concepts Gilbert Harman

1 Qualia as Mental Paint Can we become directly or introspectively (as opposed t o inferentially) aware of those aspects of perceptual experience -the mental paint, etc.- that serve to represent what we perceive? I say we cannot. Obviously we can become directly aware of intentional of certain intentional or representational features of our experiences, e.g. that a given experience is a visual experience of a red tomato. Ned Block appears to believe that I deny this. He says, "In Harman's lingo, a representational feature of experience is a feature of what is represented, not what is doing the represent.ingV. So let me insist that by "a representational feature of experience" I mean a feature of experience. My point was that we can be aware of such features of experience without being aware of the "mental paint" by virtue of which the experience represents what it represents. Block says that there is a "prior question", namely, whether there is such a thing as mental paint. I take it he means the question whether perception involves mental representations, the answer distinguishing a certain approach to cognitive science from a certain kind of connectionism, on the one hand, and from behaviorism, on the other hand. Of course, this is a "prior" question only for someone who holds, as Block does, that we can directly introspect men-

tal paint. The view that we cannot become introspectively aware of mental paint is compatible with the claim that there is no such thing as mental paint. Much of Block's comments are concerned to argue against what he calls "representationalism", the view that everything about what it is like to have a given experience is determined by the content of the experience, how it represents the world to be. Block claims that I am committed to representationalism in this sense, but he does not say why, and I do not think I am. I am certainly inclined to think that there can be different attitudes with the same content and that what it is like to be in one of these attitudes need not be the same as what it is like to be in another. For example, what it is like to believe that P is not the same as what it is like to desire that P. Similarly, what it is like to see something as ahead and to the right is not normally the same as what it is like to hear something as ahead and to the right. (There may be exceptions. In dim light, one may have the impression that one sees an obstacle ahead and to the right, whereas in fact one has detected it by echolocation.) To feel a pain in one's foot may be a t least in part to perceive a certain sort of event in one's foot. A feeling of pain may also involve a more or less intense desire to escape the cause of the relevant event. Presumably a desire with that content is different from a belief or other attitude with that content. Similarly for Block's orgasm example. An organism is a bodily event of a certain sort, normally, but not always noticed by the person having the orgasm. The experience of orgasm involves a perception of the relevant bodily events and usually includes other things as well, including intense feelings of desire and enjoyment. In any event, let us return to the issue whether we can become directly or introspectively (as opposed to inferentially) aware of those aspects of perceptual experience -the mental paint, etc.- that serve to represent what we perceive. This issue is distinct from the issue of representationalism. Block says, "When I a m looking a t a tomato and attending t o the color of the tomato, I am not attending to the visual sensation, but I believe that we can attend to the visual sensation just as Bach-yRita's subjects could attend to their tactual sensations". The cases are so obviously different, that I have no idea what Block has in mind here. It is certainly not true that, on reflection, we are aware of sensations in our eyes in the way that Bach-y-Rita's subjects were aware, on reflection, of sensations in the skin on their backs, sensations that serve to represent the external objects perceived.

6 . QUALIA A N D COLOR CONCEPTS

77

Block says that phosphene experiences are a challenge to the idea that we cannot attend to visual paint. I fail to see the challenge. What I experience are bright flashing lights spread out before me at an indeterminate distance, the experience being similar to the very end of a fireworks show and moments a t certain rock concerts. It's certainly true that I have no inclination t o believe that these experiences are veridical, but that is irrelevant, since the objects of visual perception may be merely intentional objects. Block suggests that such experiences could not be veridical. I a m not sure why he says that, nor why he thinks it is relevant. It is well known one's visual system sometimes presents one with "impossible objects", (Gregory, 1987) just as one can believe propositions that cannot possibly be true.

2 Inverted Earth I grant the intuitive force of Block's inverted earth example, just as I grant the intuitive force of other inverted spectrum examples. Such examples suggest that there is a real question whether the way colors look the same to two otherwise normal color perceivers. This leaves the apparently insoluble problem of providing a satisfactory account of the relevant respects of sameness. So, it is possible t o doubt whether there is a real question of the suggested sort. Block claims that the burden of proof rests on someone who denies that this is a real question. Someone else might say that the burden of proof lies on someone who thinks that adequate sense can be made of the question. I say that "the burden of proof" is a legal notion, not a philosophical notion. In any event, let me turn now to Shoemaker and Sosa's commentaries.

3 Qualia as Bases of Experiences of Phenomenal Properties Sydney Shoemaker defines a "phenomenal property" as follows:

(P) a phenomenal property is a property an object has in virtue of actually causing, in a certain sort of way (the way involved in normal cases of vision), an experience having a certain qualitative character, i.e., having a certain quale.

He says that this definition avoids the sort of circularity involved in a definition like the following:

(P*) a phenomenal property is a property an object has in virtue of actually causing, in a certain sort of way (the way involved in normal cases of vision), an experience that represents something as having that phenomenal property. Notice that circularity would not be avoided if having a certain quale were defined as follows: (Q) for an experience to have a certain qualitative character or quale is for the experience to represent something has having the corresponding phenomenal property. But Shoemaker does not offer any such definition as (Q). Later Shoemaker says that "the only way of picking.. . out" qualia in this sense is "by means of such descriptions as 'the nonrepresentational property that underlies my experience of red things"'. I reply that there isn't just one such nonrepresentational property; there are infinitely many! Some of these properties underlie the experiences of all other normal perceivers of red things; others underlie only a given person's perceptions of red things. Shoemaker's "only" method for picking out qualia simply does not pick out anything. (I am about to raise a similar objection to Sosa.) Shoemaker could pick out the qualia if he were to add, for example, that a certain qualitative character not only underlies my experience of red things but also underlies the experience of anyone else who perceives something in the way that I perceive red things, that is, anyone who perceives something as having the phenomenal character that I perceive red things as having. But that amounts to accepting (Q) and makes (P) circular.

4

Explaining Color in Terms of Qualities of Experience

Consider (CS) An object is red' if and only if perception of it would give normal perceivers red sensations under standard viewing conditions. What Ernest Sosa calls the "qualitative" version of (CS) takes the notion of a red' sensation to be the notion of a sensation "like this",

6 . QUALIA A N D COLORCONCEPTS

79

where "this" refers to a sensation that one is actually experiencing or imagining. Sosa says that I reject this qualitative version of (CS) on the grounds (from Block, 1990) that it, together with certain other assumptions, leads to such absurd results as that people mean different things by color words and that no objects have any colors. Sosa replies it would be better to reject one of the other assumptions than to reject the qualitative understanding of (CS). I reply instead that the notion of a sensation "like this" is indeterminate, since there are infinitely many ways in which one sensation can be like another (see the first two paragraphs of section 3 in my original paper). The qualitative version of CS needs to be supplemented with an account of the respects in which sensations have to be like each other. (That is why I go on to discuss functionalism and biological criteria of the relevant sort of sameness: color' sensations are the same if and only if they are functionally the same in certain respects, or if and only if they result from the biological mechanisms that are the same in certain indicated respects.) Sosa ultimately puts forward the following thesis: (CS3) Consider the sort of sensation distinctively produced in actual nsc-perceivers by red objects. For an object to be red is for perception of it to give nsc-perceivers a sensation of that sort (where "nsc perceivers" are "normal perceivers under standard conditions"). As his discussion indicates, the relevant "sort of sensation" is not (on pain of circularity) to be defined as a sensation that is produced in actual nsc-perceivers by red objects. As I understand (CS3), it presupposes a prior sorting of sensations and also presupposes that exactly one of the relevant sorts of sensation is such that all sensations produced in actual nsc-perceivers by red objects are of that sort and no sensations produced in actual nsc-perceivers by objects that are not red are of that sort. I object that Sosa has not indicated what this prior sorting might be. In particular, I do not see how he might give a noncircular account of the relevant interpersonal similarities by virtue of which different people's sensations are of the same relevant sort for the purpose of (CS3). This is the same as my objection to Shoemaker, above.

Gregory, R.L. (1987). Illusions. In Gregory, R.L., ed., The Oxford Companion to the Mind. Oxford: Oxford University Press, 337-43.


Layered Perceptual Represent at ion William G. Lycan

My purpose in this paper is to commit a fallacy. -Or rather, to continue defending a position of mine that has been rather grandly labelled a fallacy, by Ned Block (1990, p. 55), who calls it "the fallacy of intentionalizing qualia". But more on the labelling business very shortly. 1. My present effort is part of a larger project: to secure what I call the Hegemony of Representation. That thesis is a weak version of Brentano's doctrine that the mental and the intentional are one and the same: I hold that the mind has no special properties that are not exhausted by its representational properties (together with its underlying constitutive functional properties). It would follow that once representation is eventually understood, then not only "qualia", but the subjectivity of the mental, "what it's like", consciousness in any of that term's eleven or twelve dzflerent senses, and every other aspect of the mental, will be explicable in terms of representation (and the underlying functionally organized neurophysiology), without the positing of any further ingredient not already well understood from the naturalistic point of view.' 'I began this project with respect t o conscious awareness, subjectivity and qualia respectively in Chs. 6, 7 and 8 of Lycan (1987). Parts of it have also been pursued by Harman (1990), Shoemaker (1994), and Tye (1994).

My particular concern here is with qualia a n d , or but, in a n admirably specific sense of that much-abused term: By a "quale" I shall start by meaning roughly what C.I. Lewis (1929) originally did: A quale is the introspectible monadic qualitative property of a phenomenal individual, such as the color of a Russellian sense-datum. For example, if you are visually healthy and looking a t a ripe strawberry in good light, the strawberry will look red to you, and if you introspect and focus your attention on the corresponding subregion of your visual field, you will see that subregion as an individual red patch having a slightly bulgy heart shape. The redness of the phenomenal patch is the quale of the visual sensation that contains or incorporates it.2 One registers such a quale whenever one perceives a colored object as such. 2. Qualia in this comparatively clear sense pose a nasty problem for materialist theories of mind. There is (one hopes) nothing strawberry-red in one's brain, but as everyone knows, one may experience a strawberry-red quale even when there is nothing physically that color in one's environment either, as when one is hallucinating or just after-imaging. So the redness of the latter quale needs explaining, and philosophers have been hard put to explain it in materialistic terms. N.b., not even the most carefree Cartesian Dualist denies that the unveridical occurrence of redness can be explained psychophysically; the problem is rather to account for the redness as a firstorder property. Of what, exactly, is the redness a property? Not of anything physical in the brain, nor (by hypothesis) of anything physical in the neighborhood. So it seems the red patch is a nonphysical thing. In response to this puzzle or to others like it, many philosophers3 have seen fit to deny that there are qualia of this sort. There is in the situation no individual thing that is red, they say; rather, the afflicted subject senses redly, and that manner of sensing can be explicated without reference to any sort of object so weird as a Russellian sense-datum. But for reasons I have set out elsewhere14 I think this "adverbial" maneuver is doomed to failure; there is no getting around the referential character of talk about apparent phenomenal individuals. If you doubt me p e r h a p s because you are 2 0 n e need not endorse Russellian sense-datum metaphysics or epistemology in order t o use t h e term "quale" in this way; just think of t h e color t h a t suffuses a particular subregion of your visual field a t such-and-such a time. 3Notably Chisholm (1957), Sellars (1968), and Smart (1959). 4 ~ y c a n(1987), pp. 85-88. For somewhat different objections, see Jackson (1977) and Butchvarov (1980).

already a satisfied a d v e r b i a l i s t then (a) just take my word for it, or at least (b) please grant me the assumption for the sake of argument, or, best of all, (c) as soon as we are through here, go read my book in a spirit of uncritical awe.5 But this is worse and worse: I am insisting that there is a red individual, specifically a red patch, present to one's consciousness even when nothing physical inside or outside one's brain is red. It sounds as though we have abandoned materialism already. Happily, there is an escape hatch: In the mediaeval tradition revived thirty years ago by Elizabeth Anscombe ( 1 9 6 5 ) , ~we can treat phenomenal individuals as intentional inexistents. Suppose I am having a strawberry-red after-image. Then there is a red patch in my visual field. Do I conclude that I am confronted by something that is actually red? I do not. It merely seems to me that there is a red individual before me; it is in my mind visually as if a red object were before me. And that impression of mine is, in actual fact, false; I am in the grip of optical illusion, even though I am not deluded, taken in. (The illusion can be explained by psychophysiology, for my visual analyzers are functioning in some of the same ways they do when I really do perceive a red physical object.) The red object is a representatum, and "exists" only as an object of thought.7 So we may extend the original C.I. Lewis sense of "quale" analogically: A quale is, not the introspectible monadic qualitative property of an actual phenomenal individual (there being no such things), but that of what seems to be a phenomenal individual and is referred to, individuated and counted as such. Call this the "loose Lewisian" sense. 5lhave been told by people I deeply respect t h a t t h e book is both more pleasant and more invigorating t o read if one is reading a copy t h a t one has bought and paid for (particularly a hardback copy). 'See also Hintikka (1969), Thomason (1973), Adams (1975), Kraut (1982), and Lewis (1983). 7 ~ my n view, t h e best going semantics for intentional inexistence is a possibleworlds semantics: T h e apparent mental references t o phenomenal individuals are really t o objects existing in or "at" alternative worlds. I maintain t h a t t h e latter otherworldly objects are physical a t their containing worlds, so we need not worry anywhere about nonphysical phenomenal individuals. (Granted, we need t o explain t h e physical natures of their qualitative properties in their own containing worlds. Lycan (1987, pp. 88-93) makes a s t a b a t t h a t . ) B u t it is important t o see t h a t t h e possible-worlds metaphysics is inessential t o my view of phenomenal individuals as intentional inexistents. I happen t o think t h a t intentional inexistents are denizens of other possible worlds, but if you have a different metaphysic of intentional inexistents, apply it and d o not waste paper criticizing mine. (Have a nice day.)

The view that qualia such as phenomenal color are the intentional objects of sensory experiences is a n instance of my general doctrine of the Hegemony of Representation, specifically applied to qualia. It is also precisely Block's "fallacy of intentionalizing qualia", mentioned in my opening paragraph: as he characterizes it, "the supposition that experiential contents that can be expressed in public language such as looking red are qualitative contents" (p. 55). Peacocke (1983, pp. 8-10) calls it the "extreme perceptual theory" of sensory experience. Now, in my view, Block has committed the LLfallacyfallacy", that of imputing fallaciousness to a position with which one disagrees but without doing anything to show that that position rests on any error of reasoning (cf. "the naturalistic fallacy", et al.). 3. The position does face a number of objections. Sydney Shoemaker (1981, 1990), Christopher Peacocke (1983) and Block offer arguments designed precisely to enforce a distinction between an experience's intentional content and its more specifically and distinctively "qualitative" content or aspect. (Peacocke's terms are "representational" and "sensational". Such dichotomies go back a t least to Chisholm's (1957) distinction between "comparative" and "noncomparative" senses of "appears".) T h e idea is that while two people's visual experiences may be alike in intentional content -e.g., the two people are both "sensing redly", perceptually-believing that there is something red before them, disposed to shout "Red!", and even visually appeared to as by a red object- the same two experiences may differ in their intrinsic qualitative or "sensational" contents. Shoemaker and Block concentrate on color, and offer inversion hypotheses: one of the foregoing two subjects may be acquainted with a green quale rather than a red one. Thus visual experiences have distinctive intrinsic color contents that do outrun those experiences' merely intentional color contents. Both Shoemaker and Block speak of those intrinsic contents as "qualia". Thus the already abused term "quale" has grown yet another ambiguity.$ One can believe in qualia, as I do -in introspectible '1ts most noticeable ambiguity up tiil t h a t point had been t h a t between its very strict and commissive Lewisian sense, which I defined in section 1, and a more general and more common use, meaning whatever it is about experience t h a t constitutes phenomenal character in any sense, often run together with diffuse notions of "consciousness", "subjectivity" and t h e like. (Considerable harm has been done in t h e philosophy of mind by the use of terms such as "qualia" or (especially) "consciousness" in t h a t diffuse sense.) In section 2 I introduced a third, but fairly precise sense, the "loose Lewisian", which differs only in dispensing with actual phenonemal individuals. Block's use is a fourth.

first-order color properties given phenomenally in experience- while insisting that qualia are merely intentional aspects of sensation, r e p resented properties, and rejecting the possibility of the relevant inverted spectrum;g or one can insist that there are not merely qualia in the sense of intentional color contents, but also "intrinsic" qualitative contents that outrun the intentional. We might call the latter "Q-ualia" with a capital and boldface "Q". It is very important to keep the distinction in mind, and also not to let Shoemaker and Block kidnap Lewis' term and keep it for themselves. 4. But let us confront Block's particularly ingenious inversion scenario. What is distinctive about it is that what inverts is not its victim's inner experience, but the real physical colors in the victim's environment .lo Here is Block's argument against intentionalizing qualia. He posits an "Inverted Earth", a planet exactly like Earth except that its real physical colors are (somehow) inverted with respect to ours. The Twin Earthlings' speech sounds just like English, of course, but their intentional contents in regard to color are inverted relative to ours: When they say "red", they mean green, if it is green Twin objects that correspond to red Earthly objects under the inversion in question, and green things look green to them even though they call those things "red". (Let us suppose there is no split internal to their own population.)

he qualification "the relevant" is massively important here. There is a double relativity t o t h e notion of "inverted spectrum", and it has been widely missed and/or exploited in the literature. We must always ask, what is inverted with respect to what? T h e second parameter just expressed has been more salient, though not salient enough: Many people grant t h e possibility of spectrum inversion with respect to (just) behavioral dispositions or input-utput relations; less obvious is t h e possibility of inversion relative t o behavioral dispositions and underlying functional profile; still less obvious and granted by hardly anyone is t h e possibility of inversion relative t o one's entire molecular constitution. If we are Functionalists, our concern will be t h e second of those three relativized hypotheses, and if we are Functionalists, we will not grant t h e possibility of inversion with respect t o global functional state. One must be very careful, as many writers are not, t o distinguish spectrum inverted merely with respect t o 1-0 relations from t h e much more contentious and usually question-begging inversion with respect t o global functional state. The first parameter of the double relativity can be important too: W h a t is getting inverted? Notably, Shoemaker and Block each owe us an answer t o that, when they posit inversions with respect t o global functional state plus intentional color content, and neither has (in print) supplied a satisfactory answer. "Block claims t o avoid the question of inverted spectrum's metaphysical possibility -successfully, since he does not invert anyone's internal spectrum. But that does nothing to show that his own example is metaphysically possible.

Now, a n Earthling victim is chosen by the customary mad scientists, knocked out, fitted with color inverting lenses, transported to Inverted Earth, and repainted to match that planet's human skin and hair coloring. (Block calls the victim "you".) When you wake up, you experience nothing abnormal, because the inverting lenses cancel out the new planet's color inversion. Your language remains as always, so the natives do not notice anything either. But note that you are now misperceiving; you see green objects as red. You speak falsely, and have false intentional contents. Yet, Block argues, this would change over time: At first, when you look at the sky, thinking the thought that you would express as 'It is as blue as ever', you are expressing the same thought that you would have been expressing yesterday at home, only today you are wrong. . . .Nonetheless, according to me, after enough time has passed on Inverted Earth, your embedding in the physical and linguistic environment of Inverted Earth would dominate, and so your intentional contents would shift so as to be the same as those of the natives. . . .If you were kidnapped at age 15, by the time 50 years have passed, you use 'red' to mean green, just as the natives do. Once your intentional contents have inverted, so do your functional states. The state that is now normally caused by blue things is the same state that earlier was normally caused by red things. So once 50 years have passed, you and your earlier stage at home would exemplify what I want, namely a case of functional and intentional inversion together with the same qualitative contents t h e converse of the [usual] inverted spectrum case. This is enough to refute the functionalist theory of qualitative content and at the same time to establish the intentional/qualitative distinction. [p. 641

5. I shall make two replies to all this. (I have a third as well, but it is dialectically thorny and would take too long to expound here." The first reply is only superficial: Block assumes a nai've psychosemantics for intentional states, done in terms of "normal causes" or causal "grounding" or both, where these notions are understood more or less statistically. That is, he assumes that when for a given type of state, a new kind of cause comes to predominate over the state's original normal cause, the state's intentional content actually changes and the state then refers to the new kind of cause. I reject that psychosemantics. Like Ruth Millikan, I think of "normal cause" not statistically but teleologically, and I tend to understand teleology in terms of selection history. (A state may have its " S e e Ch. 6 of Lycan (in press).

teleologically normal cause but rarely -possibly never, if there is sufficient mismatch between the state's evolutionary origin and its subject's present environment.) Thus, for me, the Inverted Earth transportation victim's intentional states would not change their intentional contents even if that person stayed on Inverted Earth for a very long time and the states ceased having their evolutionarily normal causes altogether. But the nai've psychosemantics is probably inessential to Block's strategy. It would not take much work to construct a variant Inverted-Earth case that could be applied mutatis mutandis against my etiological psychosemantics. Here is my reason for saying that: What I think is really at stake here is a general Putnamian "narrow1,-us.-"wide" issue. ("Narrow" properties supervene on the subject's total molecular state, while "wide" properties can vary despite molecule-for-molecule sameness.) And biosemantics makes intentional content as "wide" as does the naive psychosemantics. As I interpret Block's real objection to intentionalizing qualia, it is that intentional contents are wide while qualia are narrow. So let us move right on. 6. Block's Q-ualia are very like the "narrow content" allegedly possessed by propositional attitudes. (He almost draws the analogy himself, on p. 69.) Viz.: they are like the undisputed intentional contents except that they supervene on molecular constitution and so do not shift when the subject's environment is hypothetically or actually rearranged. Generically, my second objection is that any such notion is dubious. "Narrow content" for propositional attitudes is notoriously disputed, often on the grounds that "narrow 'content' isn't content". By analogy, we should suspect any parallel notion of qualitative character, especially when the suspicion is backed by my general argument against Q-ualia (given in the previous section of this paper). But more specifically, Block's argument depends on the assumption, explicitly admitted by him on p. 66, that qualia supervene on molecular constitution. At least, I think this supervenience assumption is doing some heavy work in shoring up the Inverted Earth example; viz., it nails down Block's claim that despite radical variations in the subject's environment or history or whatever, the subject's qualitative experience remains the same so long as there has been no intrinsic physical change in the subject herlhimself. Here the dialectic gets slightly tricky: On my own view, qualia (first-order monadic properties of ostensible phenomenal individuals) do not supervene on molecular constitution because they are intentional contents. Block believes there to be Q-ualia that out-

run intentional contents, and he chooses to add that those Q-ualia supervene on molecular constitution. (Note that the latter claim is additional, even though it figures in motivating the former.) So the question between us is that of whether there are any Q-ualia, not that of whether qualia in my intentional sense supervene on molecular constitution. Block seems to agree that there are nonsuperveliient qualia in the sense of intentional contents, so we may distinguish "intentional qualia" from Q-ualia. And I see no reason to grant that there are Q-ualia at all, much less supervenient Q-ualia. I know of nothing that particularly warrants the supervenience claim, and it begs the question against the view that all qualia are intentional qualia: If one is already assuming that qualia are nonintentional, "intrinsic" features of sensation, then one will find the supervenience claim plausible; not otherwise.12 Nonetheless, n.b., we should grant the plausibility of the supervenience claim for the case of bodily sensations such as pain. For even if such sensations are intentional (as I believe they are (Armstrong (1962, 1968); Pitcher (1970)), their intentional objects are parts of the subject's own body, and so molecular duplicates can be supposed to share them. I suspect that this is a major source of the notion that qualia generally are supervenient, for I conjecture that philosophers extrapolate from the supervenience of bodily-sensory qualia to that of colors and other world-perceptual qualia; it is tempting to assimilate the "feels" of perceptual states to those of sensory states, so far as we consider perceptual states to have feels. But the "feel" terminology suits perceptual states less well than it does bodily-sensory states. Bodily sensations are themselves conceived as feelings, while perceptual states are not; as Harman (1990) observes, we norrnally "see right through" perceptual states to external objects and do not even notice that we are in perceptual states.13 Thus, it is fallacious to infer that since bodily qualia supervene on molecular constitution, perceptual qualia do also. I turn to Peacocke. '2Actually I have a global objection t o Shoemaker's and Block's idea as described here, b u t it is dialectically tedious; see Ch. 6 of Lycan (in press). 13 "When Eloise sees a tree before her, t h e colors she experiences are all experienced as features of t h e tree and its surroundings. None of them are [sic] experienced as intrinsic features of her experience. And t h a t is t r u e of you t o o . . . . Look a t a tree and t r y t o turn your attention t o intrinsic features of your visual experience. I predict you will find t h a t t h e only features there t o turn your attention t o will b e features of the presented tree, including relational features of the tree 'fro111 here"'. [p. 391

7. Peacocke offers three arguments in favor of Q-ualia, or as he calls them, nonrepresentational sensational properties. As he says, the Hegemony of Representation or "extreme perceptual theory" implies the "Adequacy Thesis (AT)": "that a complete intrinsic characterization of a n experience can be given by ernbedding within an operator like 'it visually appears to the subject t h a t . . . ' some complex condition concerning physical objects", or alternatively, that **thereare [no] intrinsic features of visual experience which are not captured by representational content". His arguments are by counterexample to the AT. Actually I shall address only the first and the third, for the second, though offered as an elaboration of the first, admits of a special and unrelated reply (see fn 22 below). Here is the first alleged counterexample. (1) Suppose you are standing on a road which stretches from you to the horizon. There are two trees at the roadside, one a hundred yards from you, the other two hundred. Your experience represents these objects as being of the same physical height and other dimensions; that is, taking your experience at face value you would judge that the trees are roughly the same physical size. . . . Yet there is also some sense in which the nearest tree occupies more of your visual field than the more distant tree. This is as much a feature of your experience itself as is representing the trees as being of the same height. The experience can possess this feature without your having any concept of the feature or of the visual field: you simply enjoy an experience which has the feature. (p. 12) This is a striking phenomenon and the argument is impressive. How to respond?14 I think that (like Block), Peacocke here relies on an assumption that can fruitfully be challenged. He supposes that what a visual experience represents is an array of everyday states of affairs involving ordinary physical objects and their standard properties (the road, the trees, the objective sizes of the trees and their relative positions, etc., etc.). That supposition itself is not what I wish to contest, since of course the visual experience in question does represent those things. (Or I should say, I agree with Peacocke and Block that it does. Russell would not have agreed that the components of a visual experience represent anything, and possibly not even that the experience itself does.) What Peacocke further seems to assume, and what I do dispute, is that those everyday environmental things are all that is represented in vision. I4Michael Tye (1992, p. 173) has made one doughty response, but I do not find it entirely convincing.

More specifically and at the same time more generally, I want to suggest that a perceptual experience has more than one layer of intentional objects. Indeed a single apparent color patch in one's visual field, does not represent only one kind of external object, but at least two at the same time. This will take a bit of getting used to. Let me introduce the idea by switching to a different sense modality, that of smell. (In another work (Lycan (1989)) I have argued that the olfactory perspective is extremely useful in correcting misapprehensions that arise from overattention to vision, since vision is a radically atypical and unrepresentative sense modality.) In the same essay I argued -what I grant is not obvious- that olfactory sensations represent. To wit, I claim that a smell actually has semantical properties: reference, a truth- and/or satisfaction-condition. A smell be treated formally, a' la Hintikka, as a function from possible worlds to sets. And it can be given a linguistic-functional "dot" characterization, a' la Sellars. A smell can be incorrect, a mzsrepresentation. If these perhaps surprising things are true, then surely smells are indeed representations. 8. And I believe they are true. It may seem that, phenomenally speaking, a smell is just a modification of our consciousness, a qualitative condition or quale or Q-uale in us, lingering uselessly in the mind without representing anything. Disinclination to think of smells as representations increases when we ask what they might be representations of. One might first think that one smell represents natural gas, while another represents roses. The latter is, after all, referred to in English as "the smell of a rose". But if the rose smell represented roses, then (a) it would be true or satisfied or correctly tokened only in response to roses, false or incorrect otherwise; ( b ) it v~oulddetermine a function that, given a world, spit out exactly the set of roses at that world; and/or, (c) it would be, in Sellarsian terms, a .rose -something that plays an internal functional role analogous to the riorm-governed role played by the word "rose" in the public English language. The rose snlell does none of those things in humans. In particular, there are objects other than roses that set off the rose smell -artificial rose smells can be made of any substance whose rnolecules are shaped similarly to those of roses. The point is not that the nose can be fooled. Au contrazre; it is that in the artificial case, the nose is not fooled, and the rose smell is riot incorrectly tokened. An artificial rose that produces the rose smell is smelled correctly, for it does have that smell even though it is not a rose. KO one would ever suggest that if we discover other things on other planets sniell like roses, the initial categorizatioli "rose odor"

was mistaken. (And conversely, of course, if one experiences the rose smell when presented with an odorless variety of rose, one is not smelling correctly, though it does not follow that one is smelling incorrectly.) Thus smells do not represent the external things by reference to which they are usually classified, even though they are usually rightly taken by subjects to indicate those external things as well. Nor do phenomenal smells represent, as some Gibsonians would have it, broader ecologically significant properties of things, for the same reason and for others.15 Thus it is tempting to conclude that smells merely accompany external objects with a fair degree of type-type reliability, but do not represent them. However, there is another candidate for external representatum. Consider what an odor is, in the public sense of the term. It is a vaporous emanation, a diffusing collection of molecules typically given off from a definite physical source. It is itself a determinate physical thing that makes physical contact with the smell receptors in one's olfactory epithelium and sets them to firing. Moreover there is nothing arcane about this. We are publicly and commonsensically aware of odors; they are public physical entities available for sensing by anyone who happens, fortunately or unfortunately, by. (I am sure each of us can think of our own lovely or loathesome examples.) Now, odor is a candidate for representatum, and the idea of an odor as an intentional object of smell resists the objection I have made to the more colloquial candidates. For things other than roses can give off the odor "of roses", and roses can fail to give off that odor. (Again, I am not talking about olfactory deficiencies or misfires, but about the match in the physical world between types of object and types of odor.) Perhaps, then, smells represent odors. But why think that, even so? Perhaps, to the contrary, all there is to the relation is that smells are highly but imperfectly correlated with odors, and that is not enough to make for a case of representing. Actually I can think of two positive arguments for awarding representational status to smell, now that the main objection has been circumvented, though I shall offer only the first here.16 It is that once smells are correlated with odors rather than with types of object, a kind of incorrectness does manifest itself, hence a correctnessor truth-condition. If I hallucinate a rose smell in the absence of any rose or anything else that is giving off the rose odor, I am misperceiving. The point is not just that my belief that a rose is present 15See Ch. 3 of Perkins (1983).

16The other is lengthy, and is given in Lycan (1989).

is erroneous. I may not even have that belief, knowing full well that my olfactory experience is hallucinatory. Something is perceptually wrong; my olfactory bulb is saying "Rose odor" when there is no rose odor physically present, and that report is a lie. Where there is falsehood there is representation.17 9. One may suppose, then, that phenomenal smells represent odors in the sense I have specified. And so I maintained in Lycan (1989). But (in conversation) Ruth Millikan has contended against me that my arguments in the previous section against the idea that smells represent environmental objects must be flawed. If we are t o agree with anything like her pleasantly Panglossian evolutionary-historical psychosemantics (mentioned in section 5 above), we must suppose that if smells do represent anything, they do after all represent environmental objects of potential adaptive significance. Surely that is what olfaction is for, to signal food, predators, shelter, mates, and other objects of interest ultimately derivative from those; and signalling is at least a crude form of representing. I now think Millikan is right about that. But it does not force me to abandon my claim that smells represent odors. What I suggest is that smells represent adaptively significant environmental entities and they also represent odors. In fact, they represent the environmental entities by representing odors. By smelling a certain familiar odor I also smell -veridically or not- a dog. Let us return briefly to my original argument against the idea that the rose smell represents roses. It was essentially that if I experience the rose smell when the rose odor is present but no actual rose is, I a m smelling correctly, and if I experience the smell when a n odorless rose is present, I am not smelling correctly. But in the face of my layeredintentional-objects view, this argument is too simple. For that view (hereafter just the "Layering thesis") introduces the possibility that a mental representation can have more than one truth value at once. And indeed, I think it is fairly plausible to say that in the first case --that of experiencing the rose smell in the absence of any roseI am representing both correctly and incorrectly, the odor correctly and roses incorrectly. "I a m not entirely sure of t h a t last claim. One can grant t h a t a detector or indicator is registering a false positive without being forced t o admit fullbore representation, if one wishes t o place further conditions on what it takes for something t o be a genuine representation. My claim for this first argument is only t h a t smell has a possibly inexistent intentional object in a t least t h e rudimentary sense t h a t detectors and indicators have intentional objects. (However, I would add t h a t smell's strong and multifarious functional connections t o memory and other cognitive agencies suggest a representational connection as well.)

I do not know how to defend this analysis against the objection, voiced above, that my rose representation is the result of unconscious inference from the olfactory representation rather than being olfactory itself. But that is because philosophy has not resolved the general question of whether the incredibly busy pre-processing that goes on in our perceptual modules should be counted as un- (because pre-) conscious inference. A clue may be taken from Jerry Fodor's (1983) notion of a module's operation's being mandatory and informationally encapsulated. Whether or not such-and-such automated perceptual processing counts as "inference" in some legitimate sense of the word, (a) it is not a n inference performed by the subject in the usual way, but is only something that happens within the subject, (b) it is utterly insensitive to other things the subject knows and can use in inference (such as that the Miiller-Lyer illusion is a n illusion and that the two straight lines involved in it are of just the same length), and most importantly, (c) the output of a perceptual module unequivocally counts as a n object of perception in the ordinary sense of the term; indeed, in listening to human speech we perceive phonemes rather than raw vocal sounds, faces rather than geometrical arrangements of particular features, and so on. Thus, even if my rose representation is the result of unconscious inference in one sense, that does not disprove the claim that it is also itself olfactory. My claim, then, is that a given sensory state typically has, not just a single intentional object, but two or more arranged hierarchically by the "by" relation. A good model here is that of deferred linguistic reference: By referring to a numeral I refer to a number (and perhaps by doing that I refer to a building, and perhaps by doing that I refer to a government official, all in the same linguistic act -I see no a priori limit on this layering of designata). 10. Now, how to apply the Layering thesis to vision? As we saw, Peacocke seems to assume that everyday environmental things are all that is represented in vision; that is why the Q-ualia are left over. But the following is a reasonable thing to say: In vision, I see a n array of colored shapes, and by seeing those I see a room full of furniture, and perhaps by seeing that I see something still more concept-laden. (In each case, "see" is colloquial for "visually represent" .) As in the case of smell, optical illusion phenomena afford a t least a weak argument for the Layering thesis as applied to vision. Consider first the example of the Ames chair demonstrations (Ittelson (1952)). Here is a description of Adelbert Ames' device, from Gombrich (1960). [Because of technical difficulties, my originally supplied photographs are not included here.]

Most of these demonstrations are arranged in the form of peep shows. One of them which can be fairly successfully illustrated.. .makes use of three peepholes through which we can look with one eye at each of three objects displayed in the distance. Each time the object looks like a tubular chair. But when we go round and look at the three objects from another angle, we discover that only one of them is a chair of normal shape. The right-hand one is really a distorted, skewy object which only assumes the appearance of a chair from the one angle at which we first looked at it; the middle one presents an even greater surprise: it is not even one coherent object but a variety of wires extended in front of a backdrop on which is painted what we took to be the seat of the chair. One of the three chairs we saw was real, the other two illusions. So much is easy to infer from the [here omitted] photograph. What is hard to imagine is the tenacity of the illusion, the hold it maintains on us even after we have been undeceived. We return to the three peepholes and, whether we want it or not, the illusion is there. (pp. 248-49) A more vivid but still actual example is that of the peep-box, which I believe was a toy of the Victorian period in England. It was a more elaborate version of the Ames chair display.18 A little cabinet roughly the size of a shoe-box had a peep-hole a t one end, and a light source. When one looked inside through the peep-hole, one saw a miniature furnished drawing-room, in the manner of a doll's house; the furnishings were suitably heavy and ornate. But if one then took off the top of the box and looked directly down on the contents, all one saw was seemingly random little bits of wood and wire and cloth. In fact, those bits had been arranged in precisely just such a way as to present a viewer a t the peep-hole with a perfectly credible but utterly illusory Victorian room. Now, when one innocently views the contents of a peep-box through its peep-hole, does one see veridically? Here again, I think a very proper answer is, "Yes and no". Yes, because one does see shapes and textures that are physically real; there really is a dark red plushy object at two o'clock, for example, even though it is not the sort of object one supposes it is and it may be nearer or farther away from one's eye than one thinks; and one sees real edges, lines and expanses. But also, no, because one sees a miniature Victorian drawing-room that simply does not exist. Here again one might protest that the room is not perceived but only inferred from what is perceived. But my reply would be the 1 8 ~ o ar related example, see Dorothy L. Sayers' "The Haunted Policeman", reprinted in Lord P e t e r (New York, NY: Avon Books, 1972).

same: that even if it is "inferred" in some sense, that does not stop it from being seen, an object of the visual experience. 11. If all this is right, then we can answer Peacocke's first argument by denying that the representational content of our visual experience in his example is exhausted by the everyday environmental array of road, trees and background. We do visually represent the trees, and as being of the same size etc., but we do that by representing colored shapes and relations between them. Some of the shapes -in particular those corresponding to the treesare represented as being larger shapes than others, as occluding others, and so forth. As with all intentional objects, it does not follow that there are any actual things that have such relational properties. And an intentional object is normally "incomplete" in Meinong's way: There are many properties P such that the object is represented neither as having P nor as having P's complement; if the object is nonexistent, it may indeed have neither P nor P's complement. I a m maintaining that the experiential features Peacocke claims to be sensational rather than representational are represented contents after all, though the representata are not physical objects of the everyday sort.lg But if they are not physical objects of the everyday sort, what are they? What is a "shape", or a n "expanse"? The first important point to note is that the shapes etc. are (represented as being) objects external to oneself, not as mere contents of one's consciousness. When one looks into the peep-box, one sees shapes arranged in a certain complex design, but one sees them as residing inside the box, on the other side of the peep-hole, not in one's own head. They are external things.20 The second important point is that not all the shapes etc. are actual. Perhaps some of the edges and lines are actual physical edges and lines seen veridically for what that is worth, but shapes such as ''In more recent work, Peacocke (1992) has put forward some ideas t h a t seem not inhospitable t o the Layering thesis, though they d o not include it. (i) He posits a nonconceptual kind of representational content, embodied in his notion of a "spatial type" or "scenario". (ii) He posits a second and more primitive kind of nonconceptual representational content, in t h e form of "protopropositions" incorporating concepts such as "square", "curved" and "parallel to", though those terms apply t o ostensible everyday objects and their surfaces, rather than t o what I a m calling "shapes". 20Jackson (1977) got this right, even though he believed t h e shapes t o be actual ~ h e n o m e n a ilndividuals. s e n s e d a t a . As has been widely pointed out, vision presents us with very finely-grained content t h a t outruns our own ability t o express in words. I d o not suppose t h a t all t h e shapes and expanses we see fall under our own everyday concepts.

Peacocke's larger and smaller tree shapes do not really exist; there are no two actual tree-shaped objects, one larger than the other, in the environment we as his protagonists are observing. (In this way, they differ from the rose odor in virtue of smelling which we smell the rose itself.) The third important point is that the shapes etc. are physical or at least not immaterial. Here as before, that means, represented as being physical or a t least not immaterial, because they are intentional objects and most of them are intentional inexistents. The shapes one sees in the peep-box are visually taken to be physical things of some sort, even if one suspects one is being toyed with and deliberately withholds one's assent to the drawing-room perception. As represented, they may not be very robust physical things; one may think they are mere faqades, or that they are flimsy and filmy, or one may not know what to think of them. This is a good case of Meinongian incompleteness. T h e intentional inexistents are not robustly physical, but they are not nonphysical or immaterial either.21 Thus, I think shapes etc. are unobjectionably ill-behaved intentional objects, representata. Let us turn, then, to Peacocke's third argument for Q - ~ a l i a . ~ ~ 2 1 ~ is t sometimes objected against t h e intentional-inexistent view of sensed a t a t h a t apparent colors inhere in things t h a t are obviously not physical objects. E.g., we say t h a t the sky is blue, but no one thinks there is even any nonexistent physical thing u p there. B u t I contend there is a nonexistent physical thing u p there: It is a n illusion. Vision represents "the sky" as a n object. Poets write about it. Visually, t h e sky is a canopy, or the vault of heaven. 2 2 ~ isecond s alleged counterexample goes as follows.

(2) . . .Suppose you look a t a n array of pieces of furniture with one eye closed. Some of the pieces of furniture may be represented by your experience as being in front of others. Imagine now t h a t you look a t t h e same scene with both eyes. T h e experience is different. It may be tempting t o try t o express this difference by saying t h a t some chairs now appear t o be in front of others, but this cannot suffice: for t h e monocular experience also represented certain objects as being in front of others. (p. 13) T h e idea here is additional t o t h a t which case (1) was meant t o illustrate: Not only d o t h e intrinsic experiential properties or Q-ualia outrun representational content, but they can vary despite sameness of representational content. T h e binocular experience differs from t h e monocular, but not in representational content, since all t h e frontlback relations represented by the binocular experience were already represented by t h e monocular. But I a m not a t all convinced. For consider t h e very point of what stereoscopic vision does. When one eye is located a t a distance t o t h e left of t h e other, it sees just slightly further leftward around the side of a n object t h a t is in view,

(3) . . . [A] wire franework in the shape of a cube is viewed with one eye and is seen first with one of its faces in front, the face parallel to this face being seen as behind it, and is then suddenly seen, without any change in the cube or alteration of its position, with that former face now behind the other. The successive experiences have different representational contents. . . . Yet there seems to be some additional level of classification at which the successive experiences fall under the same type. (p. 16)

Peacocke infers that there is a nonrepresentational sameness between the two experiences, hence a nonrepresentational or intrinsic feature. I have two replies t o this. First, Peacocke seems to be assuming that in such aspect-seeing examples, informational or representational content tracks with the aspectual conceptualization, and all underlying aspects of the experience in question are nonrepresentational. But that would be far too hasty. Seeing-as is a large subject (Lycan (1971)), and I cannot begin to go into the needed subtleties here, but a key point to grasp is that aspectual conceptualization often outruns features of experience that are themselves uncontroversially representational. Under suitably rigged circumstances, I can see almost anything as almost anything else. If we are discussing military history and using everyday objects to represent important events, I can come to see a meerschaum pipe as the Battle of Jena. Perhaps that last example suggests an irrelevantly frivolous use of "see as", and that there is a more strictly visual use. But consider the locus classicus of seeing-as, Jastrow's duck-rabbit figure (known to most philosophers by way of section 11, xi of Wittgenstein's Philosophical Investigations). When we look a t that figure, we visually while t h e right eye sees slightly further rightward around t h e object's other side. It follows, I think, t h a t t h e binocular experience contains more information than did t h e monocular, for one gets a very slightly fuller view of each object t h a t one looks a t , even though no new frontlback relations are represented. (This observation is hardly new. Hear Joseph Harris' Treatise on Opticks, published in 1775: And by t h e parallax on account of t h e distance betwixt our eyes, we can distinguish besides t h e front, part of the two sides of a near object.. . and this gives a visible relievo t o such objects, which helps greatly t o raise or detach them from t h e plane, on which they Lie: Thus, t h e nose on t h e face, is t h e more remarkably raised by our seeing each side of it a t once! [Quoted in Gregory (1986), p. 1561) For this reason, Peacocke's second argument seems t o me a nonstarter. (1992) makes a similar point.)

(Tye

represent it as such, though it is then optional whether we take it in t u r n t o picture a duck or a rabbit. Even in this core case, seeing-as is posterior t o some visual representation. Peacocke nearly anticipates this response (p. 17), agreeing that something is seen first as a representation of a duck, and then is seen as a representation of a rabbit. But then what is so seen, an arrangement of lines on paper, remains constant in the representational content of the successive experiences. . . . [However, i]n the example of the wire cube, this reply is not available: for after the aspect switch, the wires do not all seem to be in the same relative positions as before. But now my second reply is that t h e two cube experiences share some shapes, edges and lines; and, according t o me, all those items are visually represented. So there is after all a representational sameness underlying the aspectually different seeing-as experiences; the Layering thesis saves the day again, and Peacocke's third argument fails. 13. I a m not under the illusion t h a t the debate over the Hegemony of Representation, Q-ualia, and nonrepsentational sensational features is finished. I suppose there will be more arguments for Q-ualia, that will have t o be answered in their turn. But I have stood a t Armageddon and battled for the Lord, I hope sufficiently for today.

REFERENCES Adams, E.M. (1975), Philosophy and the Modern Mind (Chapel Hill, NC: University of North Carolina Press). Anscombe, G.E.M. (1965), "The Intentionality of Sensation: A Grammatical Feature", in R.J. Butler (ed.), Analytical Philosophy: Second Series (Oxford: Basil Blackwell). Block, N.J. (1990), "Inverted Earth", in Tomberlin (1990): 53-79. Butchvarov, P. (1980), "Adverbial Theories of Consciousness", in P.A. French, T.E. Uehling and H. Wettstein (eds.), Midwest Studies i n Philosophy, V: Studies i n Epistemology (Minneapolis, MN: University of Minnesota Press). Chisholm, R. (1957), Perceiving (Ithaca, NY: Cornell University Press). Crane, T . (ed.) (1992), The Contents of Experience (Cambridge: Cambridge University Press). Fodor, J.A. (1983), The Modularity of Mind (Cambridge, MA: Bradford Books / MIT Press). Gombrich, E.H. (1960), Art and Illusion (Princeton, NJ: Princeton University Press).

Gregory, R.L. (1986), Odd Perceptions (London: Methuen). Harman, G. (1990), "The Intrinsic Quality of Experience", in Tomberlin (1990): 31-52. Hintikka, K.J.J. (1969), "On the Logic of Perception7', in N.S. Care and R.H. Grimm (eds.), Perception and Personal Identity (Cleveland, OH: Case Western Reserve University Press). Ittelson, W.H. (1952), The Ames Demonstrations in Perception (Princeton and London: Hafner Publishing). Jackson, F. (1977), Perception (Cambridge: Cambridge University Press). Kraut, R. (1982), "Sensory States and Sensory Objects", Notis 16: 277-295. Lewis, C.I. (1929), Mind and the World Order (New York, NY: C. Scribner's Sons). Lewis, D. (1983), "Individuation by Acquaintance and by Stipulation", Philosophical Review 92: 3-32. Lycan, W.G. (1971), "Gombrich, Wittgenstein and the Duck-Rabbit", Journal of Aesthetics and Art Criticism 30: 229-237; reprinted in J.V. Canfield (ed.), The Philosophy of Wittgenstein: Aesthetics, Ethics and Religion (New York: Garland Publishing, 1985), pp. -. Lycan, W.G. (1987), Consciousness (Cambridge, MA: Bradford Books/ MIT Press). Lycan, W.G. (1989), "Philosophy and Smell", unpublished monograph, excerpts from which constituted the Presidential Address delivered a t the Fifteenth Annual Meeting of the Society for Philosophy and Psychology, Tucson, Arizona (April, 1989). Lycan, W.G. (1990), "What is the 'Subjectivity' of the Mental?", in Tomberlin (1990): 109-130. Lycan, W.G. (1996, in press), Consciousness and Experience (Cambridge, MA: Bradford Books/MIT Press). Peacocke, C. (1983) Sense and Content (Oxford: Oxford University Press). Peacocke, C. (1992), "Scenarios, Concepts and Perception", in Crane (1992): 105-35. Perkins, M. (1983), Sensing the World (Indianapolis, IN: Hackett Publishing). Sellars, W. (1968) Science and Metaphysics (London: Routledge and Kegan Paul). Shoemaker, S. (1981), "The Inverted Spectrum", Journal of Philosophy 74: 357-81. Shoemaker, S. (1990), "Qualities and Qualia: What's in the Mind?", Philosophy and Phenomenological Research 50 (Supplementary Volume): 109-31. Shoemaker, S. (1994), "Phenomenal Character", Notis 28: 21-38. Smart, J.J.C. (1959), "Sensations and Brain Processes", Philosophical Review 68: 141-156.

Thomason, R. (1973), "Perception and Individuation", in M. Munitz (ed.), Logic and Ontology (New York, NY: New York University Press). Tomberlin, J.E. (ed.), Philosophical Perspectives, 4: Action Theory and

Philosophy of Mind (Atascadero, CA: Ridgeview Publishing, 1990).

Tye, M. (1992), "Visual Qualia and Visual Content", in Crane (1992): 158-

76. Tye, M. (1994), "Qualia, Content, and the Inverted Spectrum", N o h 28: 159-83.


On a Defense of the Hegemony of Represent ation Robert Stalnaker

On the general ideological issue, I am on Bill Lycan's side: I would like to believe in the hegemony of representation. But I don't find it quite so easy to believe this (or even to understand it fully) as he apparently does. When it comes to the details of his defense of the general thesis against some specific objections and counterexamples, I am not very clear about how the argument is supposed to go. I have worries both about the response to Ned Block's inverted earth argument for a distinction between qualitative and intentional content of experience, and about the response to Christopher Peacocke's proposed counterexamples to his version of the hegemony of representation. First on Block's inverted earth: I will review Block's argument, as I understand it, and then consider Lycan's response. Block told his story in two versions. In both of them, inverted earth is a place much like earth, except that the actual physical colors of things are inverted -switched with the complementary colors. Further, the language spoken by inverted earthians is an inverted version of English: they use "blue" to refer to the color yellow, which is the color of the sky. In the intersubjective version of the story, our subject

(you) remains on earth, but has a counterpart on inverted earth who is like you, except that he is fitted with inverting lenses. In this version of the story, inverted earth might, in fact, just be a counterfactual possible world. But Lycan focuses on the intrasubjective version of the story, according to which inverted earth is a part of our universe, and you are transported from here to there (abducted, perhaps, by some of those aliens that Harvard psychoanalysts like to investigate). It is part of this version of the story that the abduction takes place without your noticing it. As Lycan says, "when you wake up [after the abduction], you experience nothing abnormal", since you were, without your knowledge, fitted with inverting lenses that cancel out the effects of the color inversion on the planet. Your language remains the same, and so when you say "the sky is blue", you mean that the sky is blue, unlike your new compatriots, who mean by words that sound exactly the same as yours that the sky is yellow, and who mistakenly interpret you to mean that as well. But, Block suggests, after many years go by (during which you never learn about the switch), it will eventually become true to say that you have become a speaker of inverted earth's variant of English, and so to mean what they mean by "blue", which is what earthly English speakers mean by "yellow". But while the intentional content of your thoughts about colors (and of your visual experiences) may change, the qualitative content of your color experiences remains the same. Therefore, there must be a difference between the intentional content of experience and their qualitative content. Now Lycan is skeptical about the assumptions about semantics that lead Block to hypothesize that there would be, after enough time goes by, a semantic shift, but he does not think this is where the real problem with the argument is, so he grants Block his semantics for the purpose of the argument. The real problem, he suggests, is that Block has no reason to assume that there are any mental properties of your color experience that remain the same when the semantics shifts. Only by assuming that Qualia are narrow properties p r o p e r t i e s that supervene on the internal molecular constitution of the subject- is this conclusion justified. But this assumption, he says, begs the question: the defender of the hegemony of representation rejects the assumption that there is anything about the qualitative character of color experience that is narrow in this sense. I don't find this persuasive since while I agree that Block believes (with good reason) that Qualitative properties supervene on the internal molecular constitution of the person whose experiences have

8. ON A DEFENSE OF THE HEGEMONY OF REPRESENTATION103

the properties, I don't think that he assumes this, or that it plays any role in the argument. Let us stipulate that to be a Qualitative property of an experience, a property must satisfy at least the following necessary condition: if two experiences are introspectively indistinguishable, then either both or neither have property $. (At this point, the difference between the intrasubjective and the intersubjective versions of the story becomes important, since the introspective indistinguishability relation applies only to experiences of a single person.) Might one deny that there are any properties meeting this condition? Perhaps, but only, I think, by claiming that we cannot make sense of the relation of introspective indistinguishability (or by claiming at least that we cannot make sense of this relation outside of very limited contexts). One might accept the intelligibility of introspective indistinguishability, while rejecting all but holistic Qualitative properties, but holistic properties of this kind, I think, will be enough for Block's purposes. Notice that, while any reasonable materialist will believe that all properties meeting this condition will also meet the supervenience condition (and so will be narrow or intrinsic properties), this further conclusion need play no role in the argument. It is the assumption of introspective indistinguishability that is doing the work. And if one grants the intelligibility of this assumption, it seems to me difficult to deny that in the story as Block tells it, your experiences are introspectively indistinguishable, after the switch, from what they would have been if the switch had never been made. If it is not obvious that this follows from the story as already told, just add the following elaboration of it: consider the following possible experiment: the aliens fill you in about the details of what has happened to you, and then take you back to earth, removing your inverting lenses. "Attend carefully to the character of your visual experience", they say. Then you are returned to the inverted earth situation, with the inverting lenses, told that you are there, and again asked to attend carefully to the character of your visual experience. Finally, you are again blindfolded, and one of the two locations is selected at random. This time, you are not told which of the two conditions you are in, but are asked to attend to the way things look, and on the basis of this to judge whether you are on earth or inverted earth. Let us assume that, were this experiment to be performed, you would be unable to make an informed judgment. You would report that you were unable to tell any difference at all between the way things look on earth and the way they look on inverted earth. (Don't assume that this experiment is actually performed; just assume that

the counterfactual, if the experiment were performed, this would be the result, is true according to the story.)' Now I can't see what question Block has begged if he claims to have told a story in which two experiences have the same Qualitative content (where Qualitative content is understood to be a property meeting the indistinguishability condition), without having the same intentional content. If he is a materialist, he will be committed to the thesis that the experiences of molecular duplicates share all Qualitative properties, but this will be a conclusion, not an assumption, and it will be a conclusion that is not necessary to make the point. Lycan's reasons for being skeptical about any such property seem to me, in any case, pretty weak: Qualia are something like narrow content, which is "notoriously disputed. . . . We should suspect any parallel notion of qualitative character". (p. 10) I have no sympathy for any notion of narrow content, but I think one needs a better reason to suspect a notion that might be thought to share some of its features than guilt by association. Now on Christopher Peacocke's examples: I am inclined to agree with Lycan that perceptual experiences have multiple representational contents, and that these contents may be layered in the sense that for different properties P and Q, it may perceptually appears to one that P by its appearing to one that Q. Further, I am inclined to suspect that Lycan may be right that this is relevant to the task of reconciling a representationalist thesis with the phenomena that Peacocke's examples direct our attention to. But I don't have a very clear idea exactly how the argument is supposed to go. I am not entirely clear either about exactly what the problem that Peacocke is raising is, or about how a recognition of the fact that perceptual representations are multiple and layered is supposed to solve it. Let me first review Lycan's point about olfactory representation and then turn back to the visual examples. I am happy enough to grant that smells are representational, and that they can have multiple intentional contents. Specifically, that a smell can be correctly said to represent both the presence of an object of the kind that normally gives off the smell (the rose) and also the presence of an odor of the kind that such objects normally give off (where an odor is something physical, something that occupies space). In such a case, I take it that one and the same experiential h he role of introspective indistinguishability in Block's argument is most explicit in his summary of the argument in his article on qualia in Samuel Guttenplan, A C o m p a n i o n t o the Philosophy of Mind (Basil Blackwell, 1994). See especially the paragraph on pp. 517-518.

condition -my particular olfactory experience- realizes different intentional states -on the one hand, the state of representing that a rose is in the vicinity, and the state of representing that there is a rose-odor in my environment. The subject has two different representational properties -its olfactorily appearing to him that a rose is present in the vicinity, and its olfactorily appearing to him that a rose-odor is present in the surrounding atmosphere- and he has both of those properties in virtue of having one sensory experience. The one state might represent correctly while the other does not. One can, as Lycan suggests, contrast three possible scenarios in which I have the same rose-smelling experience: In the first, a rose is giving off a rose-odor, which is causing my experience; in the second, there is a rose-odor, perhaps produced by a synthetic chemical reaction, but no rose; in the third, I am having an olfactory hallucination. In the first case, both of the representational states realized by the experience are veridical; in the second one is but the other is not, and in the third, neither is. The character of the experience is the same in all three cases, and in all three cases the experience has the same intentional contents. Now all of this seems fine to me, but what I don't see is how to turn this into a response to Peacocke's counterexample. Let me first try to say what I think the problem is that Peacocke's first example is supposed to raise, and then consider how the recognition of multiple and layered representations might help to solve it. The visual experience, in the example, is of a pair of trees along a road. One fact about the character of this experience is that the trees appear to be of about the same size, with one closer and one farther away. But it is also a feature of the experience, Peacocke suggests, that the nearer tree occupies a bigger part of the visual field than the farther tree. That is, we are inclined to say in describing such an experience that something -a certain tree shaped part of the visual field- is bigger than something else -a different tree shaped part of the visual field. Now the representationalist strategy is to paraphrase all descriptions of the character of experience in representational terms, where the properties expressed in characterizing the experience are not ascribed to anything, but are used to identify a representational content -a proposition. According to this strategy, the reason we can use such words as "green" and "square" to describe the character of a visual experience without committing ourselves to the existence of anything relevant that is actually green or square is that those words occur (in a perspicuous paraphrase of the characterization of the experience) only in the scope of a "that" clause which identifies an intentional content used to characterize the experience -as

the experience of visually representing that something before one is green, or square. Peacocke's example, as I understand it, is supposed to present a prima facie problem for this strategy since while we are inclined to describe our experience in terms of something (a certain shape in the visual field) being bigger than something else, this judgment of comparative size seems to be about a feature of the representation, and not about what purports to be represented. To solve the problern, it would seem that one would need to find a paraphrase of the description of the visual field into the form "it visually appears to the subject that . . . " where the "bigger than" occurs in the that-clause: something visually appears to be bigger than something else, but nothing is said actually t o be bigger than anything else. The recognition that one experience can have multiple contents allows for the possibility that one and the same visual experience might represent two tree as being the same size, while at the same time represent one of them as being bigger than the other. But that does not seem to be the right description of the experience, and it is not what Lycan is suggesting. Intuitively, one is inclined to say that the trees themselves are represented as the same size, while it is the images of the tree that are of different sizes. But the images are representations, and not (it seems) things represented. It is not clear how the analogies with odors is supposed to be applied here. Both roses and rose odors are, as Lycan emphasized, things that we may represent to be present in the nonmental world. If the trees are analogous to the rose, what is it that is analogous to the rose odor in the layered account of the visual experience in Peacocke's example? Lycan suggests that it is "colored shapes and relations between them. Some of these shapes -in particular those corresponding to the treesare represented as being larger shapes than others". (p. 24) Now there need not actually be any such shapes: we need not be perceiving veridically. While the observer in Peacocke's example is perceiving the trees veridically, Lycan says that "shapes such as Peacocke's larger and smaller tree shapes do not really exist". At this layer of perceptual representation, things are not as they visually appear to be. But to understand what the proposed content is, one has to understand how the world is being represented to be -what the world would have to be like in order for the representation to be veridical. Is the proposal that the content, at this level, of the visual experience is that there are two tree shaped physical objects --one larger than the other? Is that the way, or a way, that things visually appear to be to the observer? One might be inclined to reply that that is not how things seem to appear to the observer.

8. ON A DEFENSE O F T H E HEGEMONY O F REPRESENTATION107

Here is a visual case that seems to me analogous to the smell example: I see a scene (which might be just like the scene described in Peacocke's example) reflected in a mirror (recognizing, let us say, that that is how I am seeing it). In this case, I really do represent both the scene (in which the trees are the same size) and an image -a mirror image- of the scene, in which one tree-image is bigger than the other. Just as in the olfactory example, where my experience represents a rose b y representing an odor, in this case my experience represents trees of the same size b y representing tree images that are of different sizes. As in the rose case, both representational states may be veridical, or one might be veridical, the other not. Suppose I think I am seeing trees reflected in a mirror, but in fact the image is projected on a screen, and there are no real trees. Or it might be that it is the mirror image that is an illusion: in fact I am seeing the trees through an open window that I take for a mirror. But in Peacocke's example, there is neither a mirror, nor the appearance of one. Can one still say that it visually appears to me that there is before me an image (an intermediate physical representation) which contains two tree images, one bigger than the other? This may not seem very plausible, phenomenologically, but it might seem a little better if one said simply that my visual experience is the way it would be if I were seeing an image (a physical image) that contains two tree images, one bigger than the other. This more explicitly externalist characterization puts the description of one shape being bigger than another into a counterfactual condition, rather than a that-clause. Is that change significant, from the point of view of the hegemony of representation? The answer to this question may depend on exactly how the hegemony thesis is formulated, and what work it is intended to do. But this way of paraphrasing the characterization of the experience has this in common with a more straightforward representationalist paraphrase: one explains how the size comparison can be used to describe the intrinsic character of an experience without hypothesizing any actual entities -components of the experience- that are compared with respect to size. If this is what Lycan has in mind, I think I could be talked into this line of response to Peacocke's example. Peacocke emphasized in describing his example that the viewer need have no concept of the visual field in order for it to be correct to describe the viewer's experience in terms of a visual field and various features of it. We are not supposed to assume that the example is one of a sophisticated observer, reflecting on his visual experience and judging that one object takes up more space than the other. The claim is that even for the unreflective viewer who thinks

only about the scene, who does not notice that the one tree takes up more space in his visual field than the other, or perhaps even have the conceptual resources to notice this, there are still features of the experience which we can use the concept of a visual field to describe. This can be granted, I think, without by itself causing any problem for the representationalist. There is no reason why one should not, in order to identify a content for the purpose of attributing an intentional state with that content to a person, make use of concepts that are unavailable to the person. Such externally identified content is perhaps "nonconceptual" in some sense, but it is none the less intentional or representational for that. (Some people find nonconceptual content puzzling. I have no problem with it -it is the other kind that I can't understand.) I will conclude with a very brief comment about Peacocke's second example: the wire frame shaped like a cube that is seen first with one of its faces in front of another, and then with the two reversed. I confess that in this case it is not clear to me what the problem for the representationalist is supposed to be. One does not need a theory of layered representation such as Lycan proposes in order to recognize that one visual experience may have many representational properties. I may describe different features of the way things visually appear to someone, and I might describe them in more or less detail. (I might say that an object before me seems to be blue, and also that it seems t o be navy blue; I might say in addition, describing the same visual experience, that the object seems to be shaped like a cone.) Two visual experiences might be alike in certain respects, and different in others (I might seem to see an object of the very same color, but this time shaped like a sphere), and a single extended visual experience might change in some respects and remain the same in others. (It still looks blue, but now it is a kind of pale sky blue.) In Peacocke's example, the content of my visual experience -how things visually appear to be- changes in some respects, but remains the same in others. I think there must be more to the problem than this but I confess I have a hard time seeing what it might be.


Perception and Possibilia James E. Tomberlin

Actualism, as I will understand it here, is minimally the view that there are no objects that do not actually exist.' By this view, there are no philosophical problems whose solution calls for or requires an ontological commitment to non-actual objects. To his credit, William G. Lycan has seriously challenged this ontological stance. In particular, as regards perception, he has set out and defended a position that the best account of qualia invokes possibilia: phenom'There are in fact grades of actualism. Alvin Plantinga (1985a), for example, endorses actualism as the view that there neither are nor could have been objects that do not actually exist. But Nathan Salmon (1987) accepts actualism only by affirming the first half of Plantinga's characterization while explicitly rejecting the second (and modal) half. As the reader may easily verify, my discussion here applies t o both versions of actualism. The literature on actualism and related modal matters is now extensive. Besides the works of Plantinga and Salmon just cited, there are excellent (even though jointly conflicting) treatments in Adams (1974, 1981), Bealer (1993), Castaiieda (1974, 1983, 1989), Deutsch (1990, 1994), Fine (1985, 1994), Fitch (1988, 1993), Forbes (1985), Kaplan (1975), Kripke (1980), Lewis (1986), Linksy and Zalta (1994), Lycan and Shapiro (1987), McMichael (1983), Menzel (1990, 1991), Parsons (1980), Plantinga (1974, 1976, 1985a, 198513, 1987), Pollock (1985), Rapaport (1978, 1979), Salmon (1981), Stalnaker (1976), van Inwagen (1987), Yagisawa (1988, 1992), and Zalta (1983, 1988).

enal individuals such as sense-data turn out to be possible but nonactual objects (Lycan 1987a, 1987b, and this conference). Now I share Lycan's anti-actualism concerning perception. And in what follows I seek to extend the argument: setting qualia aside, I shall argue, there are ample reasons to think that possibilia are properly invoked for a correct account of certain perception sentences. Consider this scenario. Jones, as it happens, has taken up nouvelle cuisine with its laudable emphasis on fresh and unusual ingredients. One weekend, in seclusion, he opts to prepare for himself the remarkable ragout of wild mushrooms with veal stock and red wine concocted by Alice Waters for her renowned restaurant, Chez Panisse. For the preparation, Jones decides, why not utilize wild mushrooms he gathered from the nearby woods just yesterday? A splendid dish indeed, he observes upon dining. But alas, some time later, Jones, still home alone and miles from the nearest person, is rendered comatose. Several of the wild mushrooms were highly toxic and Jones, alone and physically incapable of conveying his plight, faces certain death. In this situation, I take it, (1) and (2) are true but (3) is false: (1) For any individual x, if x has normal vision and x comes across Jones, x will see that Jones needs aid. (2) No actual individual comes across Jones. (3) For any individual x, if x has normal vision and x comes across Jones, x will not see that Jones needs aid. If so, however, the actualist cannot read the perception sentences (1) and (3) as universally quantified material conditionals. For suppose otherwise. Since no actual individual satisfies the open sentence

x has normal vision and x comes across Jones, every actual individual satisfies both ( l a ) (x has normal vision and x comes across Jones) that Jones needs aid)

> (x will see

and (3a) (x has normal vision and x comes across Jones) see that Jones needs aid).

> (x will

not

But then (1) and (3) are both true, after all. These considerations lead directly to a specific challenge for actualism:

9. PERCEPTION A N D POSSIBILIA111

Challenge. With objectual quantification,2 provide an interpretation of (1) and (3) satisfying these conditions: (1) and (2) are true, (3) is false, and the quantifiers range over actual individuals only. Denied the interpretation of (1) and (3) as universally quantified material conditionals, how exactly might the actualist meet this challenge? Here I consider each of three alternative proposals in turn. Strict Conditionals. By this alternative, (1) and (3) are to be construed as universally quantified strict conditionals, where a strict conditional O ( A > B ) is true (at a world w) if and only if B is true in every logically and/or metaphysically possible world (relative to w) where A is true. So understood, this proposal plainly fails the above challenge: since it is logically possible that some actual individual has normal vision and comes across Jones without seeing that he needs aid, (1) comes out false under this interpretation. Nomic Conditionals. According to this view, (1) and (3) are to be taken as universally quantified conditionals of nomic necessity, where a conditional of nomic (= physical) necessity H ( A > B ) is true (at a world w) exactly on the condition that B is true in every physically possible world (relative to w) in which A is true. While more modest than the previous interpretation, it should nevertheless be clear that the present alternative fails: insofar as no law of nature or statement of nomic necessity is violated under the assumption that some actual individual has normal vision and comes across Jones without seeing that he needs aid, true (1) won't be true, after all. Counterfactuals. With the current alternative, (1) and (3) become the universally quantified counterfactuals ( l b ) and (3b), respectively: ( l b ) For any individual x, if it were the case that x has normal vision and x comes across Jones, then it would be the case that x sees that Jones needs aid. (3b) For any individual x, if it were the case that x has normal vision and x comes across Jones, then it would be the case that x does not see that Jones needs aid. A tempting view indeed for anyone who demands an account of the truth-conditions for our target sentences while insisting on an ontology devoid of non-actual individuals. But any such theoretical attraction notwithstanding, this counterfactual interpretation is fraught with troubles, including each of the following prominent ones: 'As opposed to substitutional quantification. For criticisms of substitutional quantification, see Tomberlin (1990 and 1993).

First, as we learned from Stalnaker (1968) and Lewis (1973), transitivity and contraposition both fail for counterfactuals. And yet, (I) and (11) seem harmlessly valid. (I) For any individual x, if x has normal vision and x comes across Jones, x will see that Jones needs aid. For any individual x, if x sees that Jones needs aid, then x will attempt to help Jones. Thus, for any individual x, if x has normal vision and x comes across Jones, then x will attempt to help Jones. (11) For any individual x, if x has normal vision and x comes across Jones, then x will see that Jones needs aid. Thus, for any individual x , if x does not see that Jones needs aid, then x does not both have normal vision and come across Jones. Second, while there is room for genuine disagreement over the correct rule of truth for a counterfactual A PB, to facilitate matters I assume the one provided in Lewis (1973): A b B is true (at a world w) if and only if either (i) there are no possible A-worlds (in which case A O-+ B is vacuously true) or (ii) some A-world where B holds is closer (to w) than is any A-world where B does not hold. Next, suppose with Kripke (1980) genetic essentialism -any actual individual necessarily has the origin it in fact has. Return now to Jones and let me expand on his background. Rebounding from a failed and childless marriage, Jones, vowing not to contribute to a world of overpopulation, underwent a successful and irreversible vasectomy three years ago. That is, we have the truth of (4): (4) No actual individual is a biological offspring of Jones. Now surely any reason for treating (1) and (3) as ( l b ) and (3b), respectively, should likewise dictate that (5) and (6) are to be parsed as (7) and (8), in turn: (5) For any individual x, if x has normal vision and x comes across Jones and x is a biological offspring of Jones, then x will see that Jones needs aid. (6) For any individual x, if x has normal vision and x comes across Jones and x is a biological offspring of Jones, then x will not see that Jones needs aid. (7) For any individual x , if it were to be the case that x has normal vision and x comes across Jones and x is a biological offspring

9. PERCEPTION AND POSSIBILIA 113

of Jones, then it would be the case that x sees that Jones needs aid.

(8) For any individual x, if it were to be the case that x has normal vision and x comes across Jones and x is a biological offspring of Jones, then it would be the case that x will not see that Jones needs aid. With all of the above, however, actualism comes to grief: owing to the Lewis rule of truth for counterfactuals, genetic essentialism, and the truth of (4), actualism dictates that (7) and (8) are both true. But in the scenario about Jones it seems clear that whereas (5) is true (6) is false. Conclusion. With this negative verdict of four alternative actualistic interpretations of (1) and (3), I scarcely claim to have exhausted all of the positions in logical space open to the actualist. (But when, after all, has anyone exhausted all of the positions in logical space as regards any serious philosophical matter?) Still, I do think I have addressed the most promising ones. If so, until and unless some other interpretation is offered that suits actualism, it seems entirely appropriate to theorize that possiblia are to be invoked for a correct account of perception sentences like (1) and (3).314

Adams, R.M., 1974, "Theories of Actuality," Nozis 8: 211-231.

Adams, R.M., 1981, "Actualism and Thisness," Synthese 49: 3-41.

Bealer, G., 1993, "A Solution to Frege's Puzzle," Philosophical Perspectives

7: 17-60. 3The present discussion sharpens and extends the scepticism as regards actualism in Tomberlin (1988, 1991, 1993, forthcoming) and Tomberlin and McGuinness (1994). In Lycan (1993), there is a fascinating theoretical proposal for indicative conditionals: with Lewis and Stalnaker, he finds antecedent-strengthening, contrap* sition, and transitivity invalid; but unlike Lewis and Stalnaker, Lycan mounts a powerful case against modus ponens. The resulting view is an original and important contribution t o the vexing topic of conditionals. Still, owing to his (essential) quantification over possible but non-actual objects, Lycan's view provides no aid to actualism concerning items like (1) and (3). 4For rewarding discussion and/or correspondence, I am grateful to Bill Lycan, John Biro, Gilbert Harman, Bernie Linsky, Bob Stalnaker, Ed Zalta and my colleagues Frank McGuinness, Jeff Sicha, and Takashi Yagisawa. Linsky and Zalta (1994) contains an original and important version of actualism, one far too intricate for adequate treatment here. But I do critically examine their fascinating proposal in Tomberlin (forthcoming).

Castaiieda, H.N., 1974, "Thinking and the Structure of the World," Philosophia 4: 4-40. Castafieda, H.N., 1983, "Reply to Alvin Plantinga," in J.E. Tomberlin, ed., Agent, Language, and the Structure of the World, Indianapolis, Hackett. Castaiieda, H.N., 1989, Thinking, Language, and Experience, Minneapolis, University of Minnesota. Deutsch, H., 1990, "Contingency and hlodal Logic," Philosophical Studies 60: 89-102. Deutsch, H., 1994, "Logic for Contingent Beings," Journal of Philosophical Research 19: 273-330. Fine, K., 1985, "Plantinga on the Reduction of Possibilist Discourse," in J.E. Tomberlin and P. van Inwagen, eds., Alvin Plantinga, Profiles, Dordrecht, D. Reidel. Fine, K., 1994, "Essence and Modality," Philosophical Perspectives 8: 1-16. Fitch, G.W., 1988, "The Nature of Singular Propositions," in D. Austin, ed., Philosophical Analysis, Dodrecht, Kluwer. Fitch, G.W., 1993, "Non-Denoting," Philosphical Perspectives 7: 461-486. Forbes, G., 1985, The Metaphysics of Modality, Oxford, Clarendon. Kaplan, D., 1975, "How to Russell a FregeChurch," Journal of Philosophy 72: 716-729. Kripke, S., 1980, Naming and Necessity, Cambridge, Harvard. Lewis, D., 1973, Counterfactuals, Cambridge, Harvard. Lewis, D., 1986, On the Plurality of Worlds, Oxford, Blackwell. Linsky, B. and Zalta, E., 1994, "In Defense of the Simplest Quantified Modal Logic," Philosophical Perspectives 8: 431-458. Lycan, W.G., 1987a, Consciousness, Cambridge, MIT. Lycan, W.G., 1987b, "Phenomenal Objects," Philosophical Perspectives 1: 513-526. Lycan, W.G., 1993, "MPP, RIP," Philosophical Perspectives 7: 411-428. Lycan, W.G., this conference, "Levels of Perceptual Representation". Lycan, W.G. and Shapiro, S., 1987, "Actuality and Essence," Midwest Studies in Philosophy 11: 434-477. McMichael, A., 1983, "A Problem for Actualism About Possible Worlds," Philosophical Review 92: 49-66. Menzel, C., 1990, "Actualism and Possible Worlds," Synthese 85: 355-89. Menzel, C., 1991, "The True Modal Logic," Journal of Philosophical Logic 20: 331-374. Parsons, T., 1980, Nonexistent Objects, New Haven, Yale. Plantinga, A., 1974, The Nature of Necessity, Oxford, Clarendon. Plantinga, A., 1976, "Actualism and Possible Worlds," Theoria 42: 139-60. Plantinga, A., 1985a, "Self-Profile," in J.E. Tomberlin and P. van Inwagen, eds., Alvin Plantinga, Dodrecht, D . Reidel.

9. PERCEPTION AND POSSIBILIA 115 Plantinga, A,, 1985b, "Replies," in J.E. Tomberlin and P. van Inwagen, eds., Alvin Plantinga, Dodrecht, D. Reidel. Plantinga, A., 1987, "Two Conceptions of Modality," Philosophical Perspectives 1: 189-232. Pollock, J.L., 1985, "Plantinga on Possible Worlds," in J.E. Tomberlin and P. van Inwagen, eds., Alvin Plantinga, Dodrecht, D. Reidel. Rapaport, W., 1978, "Meinongian Theories and a Russellian Paradox," NoCs 12: 153-180. Rapaport, W., 1979, "How to Make the World Fit our Language: An Essay in Meinongian Semantics," Grazer Philosophische Studien 14: 1-21. Salmon, N., 1981, Reference and Essence, Princeton, Princeton U.P. Salmon, N., 1987, "Existence," Philosophical Perspectives 1: 49-108. Stalnaker, R., 1968, "A Theory of Conditionals," in N. Rescher, ed., Studies in Logical Theoy, Oxford, Blackwell. Stalnaker, R., 1976, "Possible Worlds," No& 10: 65-75. Tomberlin, J.E., 1988, "Semantics, Psychological Attitudes, and Conceptual Role," Philosophical Studies 53: 205-226. Tomberlin, J.E., 1990, "Belief, Nominalism, and Quantification," Philosophical Perspectives 4: 573-579. Tomberlin, J.E., 1991, "Belief, Self-Ascription, and Ontology," in E. Villanueva, ed., Philosophical Issues 1, Atascadero, Ridgeview. Tomberlin, J.E., 1993, "Singular Terms, Quantification and Ontology," in E. Villanueva, ed., Philosophical Issues, 4, Atascadero, Ridgeview. Tomberlin, J.E., and McGuinness, F., 1994, "Troubles with Actualism," Philosophical Perspectives 8: 459-466. Tomberlin, J.E., forthcoming, "Ontology and Possibilia," Philosophical Studies. Van Inwagen, P., 1987, "Two Concepts of Possible Worlds," Midwest Studies in Philosophy 11: 185-213. Yagisawa, T., 1988, "Beyond Possible Worlds," Philosophical Studies 53: 175-204. Yagisawa, T., 1992, "Possible Worlds as Shifting Domains," Erkenntis 36: 83-101 Zalta, E., 1983, Abstract Objects, Dodrecht, D. Reidel. Zalta, E., 1988, Intensional Logic and the Metaphysics of Intentionality, Cambridge, MIT.


Perceptual Experience Is a Many-Layered Thing Michael Tye

Like Bill Lycan, I am a strong representationalist about perceptual experience. I hold that the the phenomenal or 'what it is like' aspects of perceptual experiences (and indeed all experiences and feelings) can be understood in terms of their representational contents. Like Lycan, I believe that perceptual experience is layered. Like Lycan, I am not moved by any of the putative counter-examples adduced by Chris Peacocke (in his 1983). And like Lycan, I have a book coming out shortly with MIT Press, defending the representational approach to qualia.l (It too reads better when bought.) So, Lycan's heart, in my view, is unquestionably in the right place. I am less confident about the location of his other organs and limbs. Lycan's version of representationalism seems to me to place him under considerable strain. What he needs is a good massage (or a few days on the beach in Cancun) -something that will smooth out some of the tensions in his position. My primary aim in this reply is not to play the role of masseur but to explain why Lycan requires the services of one. ' S e e Tye 1995

1 The Layering Thesis According to the sense-datum theory, we do not really directly see ordinary, everyday physical objects at all. Instead, we only directly see the appearances these objects present to us, the colored expanses or sense-data that are supposedly present in our visual fields. By seeing sense-data, we thereby indirectly see tomatoes or tables or trees. So, sense-datum theorists are committed to the view that perceptual experience (broadly understood) is layered. The basic or foundational layer consists in seeing or sensing colored expanses, the upper layer consists in the conceptual representation of commonsense physical objects and their standard properties and relations. Seeing that the tomato is red consists in sensing a red, round, bulgy sense-datum and, on that basis, judging that there is a red tomato present. Lycan, if I understand him correctly, accepts this general, layered conception of perceptual experience with one major twist: there really are no sensedata. In Lycan's view, perceptual experience is representational all the way down. The bottom level consists in the sensory representation of colored shapes. This no more demands that, for each visual experience, there really be such a shape than does my hoping for eternal life demand that there be eternal life. So, sense-data, conceived of as real existents, are not required. Suppose, for example, it appears to me that there is a tomato before me. Where the sense-datum theorist would claim that I sense or directly see a red, round, bulgy sensedatum -a special sensory object that really exists whether or not the experience is veridicalLycan holds that it merely appears to me that there is a red, round bulgy shape before me. This is as much a case of representation as its appearing to me that a tomato is present. So, it too can go wrong. If, for example, I am hallucinating and nothing at all is present before my eyes, then, there really is no red, round shape out there any more than there really is a red, round tomato. Misrepresentation, then, is occurring at both levels. My perceptual experience acquires its distinctive phenomenal character not from the sensory qualities of any sensum I am apprehending, but from the fact that I am undergoing a sensory representation of a certain colored shape, whether or not the representation is accurate and the shape really exists. That, then, is the basic picture Lycan offers us. In my view, it is both too simple and too complex. It is too simple because it ignores some further layers of perceptual experience. It is too

complex because the basic layer, as Lycan conceives it, is not really different from the upper one. Let me begin with the latter charge. If perceptual experience does indeed include a layer of representation of colored shapes, then there should be possible cases of illusion at that level of representation. But what exactly makes for a case of illusion here? What exactly are the truth-conditions or correctness conditions at this level? Consider again the case of its appearing to me that there is a red, round bulgy shape present. This is correct if, and only if, there is a real, red, round bulgy shape. But what makes a shape real? And when it is real, what is the relevant shape a shape of? In the course of discussing the example of the strawberry-red after-image, Lycan says that the imager is in the grip of an illusion (p. 83). It merely seems to the imager that there is a red individual or object present. That impression, however, is false. Later, when Lycan addresses the example of the shapes seen in the peep box and Peacocke's example of the two trees, he comments

. . .not all the shapes etc. are actual. Perhaps some of the edges and lines are actual physical edges and lines. . . , but shapes such as Peacocke's larger and smaller tree shapes do not really exist; there are no two actual tree-shaped objects, one larger than the other, in the environment we as his protagonists are observing. (p. 95-6, my italics) For Lycan, then, a shape is real or actual if, and only if, some physical thing or object has that shape. So, the truth-onditions at the base level of perceptual representation advert to the shapes of physical objects or individuals (or their facing surfaces). In the case that it appears to me that there is red, round, bulgy shape ahead, my experience is accurate -things are as they appear- if, and only if, there is before me a physical object with a red, round, bulgy shape. Not only is this the account to which Lycan is committed by his various comments on shape but it also seems to me that there is no plausible alternative available to him. For example, it will not do to say that, in the above case, my experience is accurate just in case there is before me a red, round, bulgy region of space. There is no such region of space, even if I am viewing a ripe tomato and my experience is veridical. What is red, round, and bulgy is the tomato (or its facing surface). Unfortunately, Lycan is now in trouble. For one thing, perceptual representations no longer come in two wholly distinct layers, one concerned with ordinary, physical objects and their standard properties and the other concerned with colored shapes. Since the latter layer really represents the colors and shapes of physical objec.ts, it is

contained within the former layer. For another, Lycan's response to Peacocke's case of the two trees is now untenable. Let me explain.

2

The Two Trees

There are two trees beside the road that stretches from me to the horizon. The trees are of the same size, but one of them is twice as far away as the other. My visual experience of these trees is such that it appears to me that they are of the same size. Yet, there is a sense in which the nearer tree occupies more of my visual field than the one that is further away. Lycan proposes to account for these facts without the introduction of non-intentional qualia by recourse to the layering thesis. My experience represents that the one tree is of the same size as the the other tree. This is representation at the level of ordinary physical objects and their standard properties. But my experience also represents that a certain colored tree-shape is larger than another such shape. This is representation at a different level. Lycan comments: We do visually represent the trees, and as being of the same size etc., but we do that by representing colored shapes and relations between them. Some of the shapes -in particular those corresponding to the treesare represented as being larger shapes than others, as occluding others, and so forth. (p. 95) T h e problem with this proposal is as follows. For my experience to represent that the one tree shape is larger than the other, it must represent that there are two objects, both tree-shaped, one larger than the other. This is required by the account of the representation of colored shapes in the last section. But my experience also represents that those two very objects are trees of the same size. So, my experience represents that there are objects of both different and the same sizes. That, however, it surely cannot do. It will not help here to insist that there are two very different levels at work. Even leaving aside the point that the levels are not really distinct at all, how could my experience represent that there are two objects of different sizes, and thereby represent that they are of the same size? Nor does it help to qualify the account by saying that, at the alleged bottom level, the experience represents that there are two objects whose facing surfaces are of different sizes. For the experience also represents that those facing surfaces are of the same size.

There is also the further point that intuitively, the person who looks down the road is not in the grip of any sort of illusion. True, there is a sense in which the nearer tree occupies more of the visual field than the one that is further away, even though they appear to be the same size. But intuitively, no illusion exists here. This is not the case on Lycan's view, however. For his account of the above fact, as I am understanding it, is that the experience represents that there are two objects, one larger than the other; and that is a case of misrepresentation. In reality, the objects are of the same size. So, Lycan's response to Peacocke seems to me unconvincing. I remarked earlier that, in my view, his account of the way in which perceptual experience is layered is both too complex and too simple. I have tried to explain how it is too complex: layers Lycan introduces as distinct are not really distinct a t all (at least in the way he supposes). I want next to explain how his account is too simple. This will also enable me to suggest another response to Peacocke that is much more satisfying.

3 There are More Layers t o Perceptual Experience than are Dreamt of in Lycan's P hilosophy2 Perceptual experience, in my view, is a many-layered thing. This too is the view of most cognitive psychologists. Higher level vision begins, so the story normally goes, with the combining of local elements of the visual field into contours, regions, and/or surfaces. The first thing that can go wrong, then, in higher level vision is this process of grouping. The resulting impairment is known as apperceptive agnosia (in the narrow sense). It is typically brought about by damage to the occipital lobes and surrounding regions (by, e.g., carbon monoxide poisoning). Patients with this impairment often have roughly normal visual fields, so their perception of colour, brightness, and local contour is adequate. But they are strikingly impaired in the ability to recognise, match, or even copy simple shapes as well as more complicated figures.3 In general, they have great difficulty in performing any visual tasks that require combining information across local regions 2 0 r , without t h e poetic license: there are more levels t o perceptual experience t h a n are explicitly distinguished in Lycan's paper. 3 ~ o v e m e n tof shapes sometimes helps these patients t o identify them. See, for example, R. Efron 1968, p. 159.

FIGURE 1.

of the visual field.4 For example, when shown figure 1, one patient consistently read it as 7415.~ Evidently he was unable to see two parts of a line with a small gap as parts of a single line.6 The obvious conclusion is that there is a stage in normal vision in which representations of the local elements of the visual field are combined or grouped into an overall representation of the surfaces visible from the given point of view. This latter representation, which specifies lines, edges, ridges, and other surface features, is a vital foundation for nearly all higher level visual processing. Experiments performed by Stephen Kosslyn (1980) and others suggest that it occurs in a medium that is shared with imagery and that it has the structure of a grouped array like Marr's 2 ; - ~ sketch (Marr 1982), the cells of which are devoted to specific lines of sight relative to the viewer (with different cells devoted to different line^).^ As such, it does contain information about local features of the visual field too, e.g., colour, texture, and orientation. The most fundamental level of representation in visual experience, then, consists in what is represented in the array prior to any grouping. The array itself can be thought of as being like a large transparent matrix which is placed over the visual field.* Each cell in the 4They do better a t identifying real objects (e.g., toothbrushes, safety pins), than simple shapes. However, their improved performance here is based on anferences from clues provided by color, texture, etc. 'The patient made this identification by tracing around t h e figure with movements of his hand and relying on local continuity. See here T. Landis e t al. 1982. 6 ~ has t been suggested t h a t t h e visual experiences of these patients are something like those you or I would undergo, were we t o don masks with a large number of pin holes in them. 7See also Tye 1991. 'This is intended as a picture only. T h e array cells need not really be physically contiguous a t all. Instead, Like t h e arrays found inside computers, they can be widely scattered. W h a t is crucially important is t h a t cells representing adjacent regions of t h e visual field be operated upon by routines t h a t treat those cells as if they were adjacent. See here my 1991; also my 1995.

matrix covers a tiny portion of the field and is thereby dedicated to a particular line of sight. Within each cell, we may suppose, there are symbols for various features of any surface at that location in the field (for example, distance away, orientation, determinate color, texture, whether a discontinuity in depth is present there, and so on). So, the bottom level of perceptual representation is one into which a myriad of determinate local surface features enter. The representation of these features, in my view, is what is responsible for local phenomenal differences among perceptual experiences. The next level of perceptual representation is one that brings in the representation of nonlocal elements essential to grouping -the lines, edges, ridges, etc. This is achieved by processes that work over the initial local array and introduce additional representational devices for these nonlocal features. Now the grouped array that is formed here does not itself yet represent the viewpoint-independent shapes of any objects visible to the viewer (e.g., whether they are squares or circles or cubes or spheres). Indeed, at this level, there is no segmentation of the visible scene into distinct objects at all. What are needed in the higher levels of processing, then, are procedures that carve the scene into distinct objects and assign them shapes. Once the scene has been segmented in this manner, there are still further processes that categorize the objects into kinds like tomato, table, and so on. What about the viewpoint-relative representation of shapes (and other such nonlocal properties)? This operates at two different levels, I suggest. First, there is representation at the level of surfaces alone, prior to any segmentation into distinct objects. One surface may be implicitly represented as smaller than other relative to the given viewing position via the information contained in the grouped array. Secondly, there is representation of viewpoint-dependent shape and size properties of given objects. This takes place only after segmentation of the scene has occurred. My suggestion, then, is that there is a whole hierarchy of levels of perceptual representation. Agnosias of one sort or another are associated with each of these levels. The sketch I have presented above is very crude and sketchy, of course, but it will do for my present purposes.

4

The Two Trees Revisited

How does this help us with Peacocke's tree example? What account can the representationalist give of the sense in which the nearer tree

FIGURE 2.

occupies more of the visual field? The answer, I propose, is that the experience represents the nearer tree as having a facing surface that differs in its viewpoint-relative size from the facing surface of the further tree, even though it also represents the two trees as having the same viewpoint-independent size. The nearer tree (or its facing surface) is represented as being larger from here, while also being represented as being the same objective size as the further tree. There really are two different sorts of feature being represented, then, although they both are concerned with physical objects (or surfaces). Moreover, there is an associated difference in levels, a t least insofar as the representation of viewpoint-relative features of surfaces is clearly more basic than the representation of viewpoint^-independent features of objects like trees. The obvious virtue of this approach, apart from its consonance with what cognitive psychologists say about perception, is that it affords the representationalist a reply to Peacocke that neither makes the content of the perceptual experience internally inconsistent nor convicts the perceiver of any sort of error. No illusion is present. The experience is veridical on all levels: the one facing surface really

is larger from here than the other; the one tree really is the same size as the other tree. But what exactly is involved in one of two items being larger from here? Well, the one items subtends a larger visual angle relative to the eyes of the viewer. In the above case, this is encoded in the visual representation that the viewer undergoes via the greater number of active cells devoted to regions of the facing surface of the nearer tree (see figure 2). This is (a longer version of) the reply I myself made to Peacocke a number of years ago.g Lycan says that "he does not find it entirely convincing" (note 14). But he does not say why. For my own part, I see no other way of relieving the pressure which the example places on the position Lycan advocate^.^^^ l 1

DeBellis, M. 1991, "The representational content of musical experience", Philosophy and Phenomenolgical Research 51, 303-324. Efron, R. 1968, "What is perception?", Boston Studies in Philosophy of Science 4, 159. Harman, G. 1990, "The intrinsic quality of experience", in Philosophical Perspectives, 4, J. Tomberlin, ed., Northridge: Ridgeview Publishing Company. Kosslyn, S. 1980, Image and Mind, Cambridge, Mass: Harvard University Press. 'In my 1991, p. 130 (following Harman 1990 and DeBellis 1991). 'O~eacocke,I might add, has recently adduced (in his 1993 review of my 1991) another structurally similar problem case involving a coin presented a t an oblique angle. The coin occupies an elliptical region of the visual field. This is manifest in the experience. But, according to Peacocke, the coin does not look elliptical: the visual experience does not represent it as elliptical. It represents it as circular. The reply to this new case is basically the same. The facing surface of the tilted coin is represented as having boundaries, which would be occluded by an elliptical shape placed in a plane perpendicular to the line of sight of the viewer (or which form the irregular base of a cone, whose apex is the eyes of the viewer and whose regular cross-section is elliptical in shape). In this sense, the coin is represented as being elliptical from here. But it is also simultaneously represented as being a t an angle and as being itself circular. This is why the tilted coin both does, and does not, look like the same coin held perpendicular to the line of sight. "In this reply, I have not addressed the case of Inverted Earth or the question of whether the phenomenal facts supervene upon physical facts internal to the head. I do not agree with Lycan that the supervenience assumption '5s doing some heavy work in shoring up the Inverted Earth example" (p. 87). But I do agree with him that it is metaphysically possible for molecular duplicates to differ phenomenally. For a discussion of these issues, see my 1995, Chapters 5 and 7.

Landis, T., Graves, R., Benson, F., Hebben, N. 1982, "Visual recognition through kinaesthetic mediation", Psychological Medicine 12, 515-531. Marr, D., 1982, Vision, San Francisco: W. H. Freeman. Peacocke, C. 1983, Sense and Content, Oxford: Oxford University Press. Peacocke, C. 1993, "Review of M. Tye, The Imagery Debate1', 60, Philosophy of Science, 675-677. Tye, M. 1991, The Imagery Debate, Cambridge, Mass: The MIT Press, Bradford Books. Tye, M. 1995, Ten Problems of Consciousness, Cambridge, Mass: The MIT Press, Bradford Books.


Replies to Tomberlin, Tye, Stalnaker and Block William G. Lycan

1 To James Tomberlin I dispute little more in Tomberlin's paper than he disputes in mine. But as he notes in his first footnote, "actualism" comes in grades, and his paper brings out an interesting further distinction between three such grades. The strongest grade might be called "Very Strong Actualism", for according to it: There must never be any objectual quantification over nonactual possibilia, i n any sense of "nonactual" at all or o n any pretext, period, neither in our object language nor in our formal metalanguage. That ferocious view is suggested by Tomberlin's opening characterization of "actualism", but it is not the only actualist doctrine he goes on to attack, for the "counterfactual interpretation" as Tomberlin unpacks it is not actualist in the Very Strong sense; in its metalanguage it quantifies over David Lewis' nonactual possible worlds. The next weaker grade of actualism, "Fairly Strong Actualism", allows for objectual quantification over mere possibilia in the metalanguage, so long as there is none i n the object language; that allows for object-language counterfactuals to be interpreted a la Lewis.

Finally, "Weak Actualism" holds that any objectual quantification over nonactual possibilia is all right, so long as it is itself ultimately interpreted i n terms of actual objects. For example, I myself am what Lewis (1986) calls an "Ersatzer", and interpret all quantification over possibilia in terms of sets of actual objects and properties (Lycan (1994)). The distinction betwen Very Strong and Fairly Strong Actualism is not terribly important, since as long as truth-conditions are truthconditions, it is hard to see any rationale for accepting the Fairly Strong view but not also the Very Strong view. However, the distinction between the Strong views and Weak Actualism is important, because Weak Actualism is (a) still significantly and controversially actualist, (b) unscathed by Tomberlin's argument, and (c) held by many modal metaphysicians, including, most importantly, me. I have just one quarrel with Tomberlin's conclusion, or rather an ad hominem for him: If we go straight to possibilia in interpreting ( I ) , allowing (1)'s quantifier to range over all the possibilia there are, then contrary to Tomberlin's own intention, (1) is falsified; for there are many possible if strange persons who come across Jones, but are cognitively defective in such a way that even though they have normal vision, they fail to see that he needs aid.

2

To Michael Tye

Tye understands and expounds my Layering view very well, but I am convinced by neither of his objections. First he tries to convict me of attributing contradictory representations to Peacocke's experience of the two trees. But I am innocent. I did claim that the experience represents two tree shapes, one iarger than the other, and that the "shapes" (as represented) are physical objects having the relevant shapes. But it does not follow, and I did not claim, that "those two very objects are trees.. . " (p. 120), and a fortiori not that they are trees of the same size. As Tye seemed to grasp earlier on, the shapes are not real, even though they are physical, while the represented trees are real. There are no actual tree-shaped physical objects at hand, one larger than the other. Remember that my model for the Layering thesis is that of deferred linguistic reference. To say that I represent a building by representing a number and that I represent the number by representing a numeral is not to say that the numeral = the number = the building. (And if fictionalism holds of numbers, the number in question does not actually exist.)

11. REPLIES T O TOMBERLIN, TYE,STALNAKER AND BLOCK

129

Thus Tye is both right and wrong in saying that the coloredshape layer "is contained within" the ordinary-physical-object layer (p. 120) and that "the levels are not really distinct at all" (p. 121). Right, in that my "shapes" are physical objects in the environment alongside or overlapping the ordinary physical objects there, but wrong in that few of them are identical with any of the ordinary physical objects. Tye's second objection is that "the person who looks down the road is not in the grip of any sort of illusion" (p. 121, italics original), so I am mistaken in positing a kind of systematic visual misrepresentation, viz., nonveridical representation of the differently-sized tree shapes which (unlike the represented trees) do not really exist. I do not find it so obvious that there is no illusion at all. Recall Hintikka's (1969) and Lewis' (1983) notion of a visual alternative, viz., a possible world logically consistent with the content of one's visual state at a time. There is a visual alternative in which there really are two physical shapes of the sort that appear to Peacocke's subject, one larger than the other, and not any actual trees at all. Think of a gigantic peep-box that convincingly presents a whole facing environment to the subject , by containing large cloths and cutouts and facades arranged in just the right ways. The resulting total experience would in one way be a grand illusion, in that it would present a road and same-sized trees, but in another way it would be perfectly veridical: the shapes would all be real shapes, physical objects of the sort that there would appear to the subject to be. Indeed, the subject might know that s/he was looking into a peepbox and not seeing a real road and trees. There would be both an entirely veridical presentation and a forehead-smacking illusion, not only at the same time but in the very same experience. So too, I claim, in Peacocke's example there are a veridical roadand-same-sized trees presentation and an illusory different-sizedtree-shapes presentation made in the same visual experience. The experience has one truth-condition that would be satisfied by the apparant different-sized shapes, and another that is satisfied by the same-sized trees. It happens that the former is actually not satisfied, and in that sense there is an illusion. If Tye thinks there is simply "not.. . any sort of illusion" here, he will have to come up with an asymmetry between Peacocke's case and the gigantic-peepbox case, and motivate the claim that there is no illusion (whatever) in the former though there is layered representation and illusion in the latter. N.b., we can grant that the illusion in the peep-box case is in some sense greater and more surprising and outrageous and authorized by biology and normal visual function; but Tye needs to

show that in Peacocke's case there is no illusion whatever, even a tiny and unsurprising and biologically normal one. To make the same point in a closely related way, think of introducing undergraduates to Descartes' first Meditation. One has to teach them to adopt the first-person perspective, the movie theater model of the mind. That is a major Gestalt shift, but the students can and do learn to perform it, and then they can also come to introspect their visual experience in abstraction from its normal externalworld content. Now, if Peacocke's subject were to snap into Cartesian movie-theater mode, s/he would be obligated by philosophers' trade-union rules to start wondering whether the alleged external world were really there a t all. True, s/he would also have t o wonder about the ostensible physical shapes, but intuitively less so than about the richly three-dimensional world of putative ordinary objects; belief in ordinary objects is more commissive, or commits to more. In Cartesian mode, the ordinary objects are more likely to be illusory than are the mere shapes. But I do not need to make the latter point in any very substantive or precise way, for a much weaker one will do: Every everyday experience admits of Cartesian Gestalt snap, and so in virtually every everyday experience, there is the component of "shape" representation that lets in at least an element of illusion. That said, I can but applaud the rest of Tye's paper and especially his suggestion that visual representation is revealed by cognitive science to have even more layers than I have supposed. I a m not entirely sure that on commonsensical grounds alone I would buy the idea that visual experience represents anything about subtending angles, but cognitive science drives out common sense any day.'

3

To Robert Stalnaker

Stalnaker offers an ingenious reworking of Block's "Inverted Earth" argument that does not rely on the gratuitous assumption that qual' ~ says ~ his e present move "is ( a longer version of) the reply.. . [he himlself made t o Peacocke a number of years ago" (p. 124), citing Tye (1991) which duplicates Tye (1992); he naturally wonders why I did not find t h e previous version entirely convincing. My reasons were not weighty. T h a t version appealed t o representation of a counterfactual: t h a t "if t h e trees were moved into line, t h e nearer one would completely obscure t h e other but not vice versa" (1992, p. 173); and it is far from obvious t h a t visual experiences represent counterfactuals of this sort. B u t it may be t h a t they do.

11. REPLIES TO TOMBERLIN, TYE,S TALNAKER AND BLOCK

131

itative content must be n a r r ~ w It . ~turns on the notion of introspective indistinguishability, which had been invoked but not emphasized by Block himself. It goes roughly as follows. (1) For a certain time interval At following your transportation to Inverted Earth, your color experiences are introspectively indistinguishable (i.i.) from those that you would have continued to have on Earth. [Conceded all around.] Therefore, Let an "IntInd" property be one that satisfies Stalnaker's characterization (p. 103) of his properties a, that must be had or lacked uniformly by i.i. experiences. Then (2) Throughout At, each of your Inverted-Earth color experiences e has at least one IntInd property in common with its corresponding Earthly experience. [Seems a safe inference from 1.1 But (3) During At, any given color experience e has changed all its (relevant) representational properties. [According to Block.] Therefore, (4) Any such e has an IntInd property P that is distinct from any of e's representational properties. [2,3.] So, for all WGL has shown,

(5) P is a Blockian Q-uale and so there exist such. [Whyever not?] This argument makes no appeal to narrowness per se, even though presumably any IntInd property is narrow. So I must address it on its own terms. I would make three replies, of which only the third is direct and particularly important. First, an exactly parallel argument can be applied to propositional-attitude contents, showing that your belief that water is H20 and your Twin's i.i. though false belief that Twin Water is H20 must share either a Blockian Q-uale or a "narrow content" or some further IntInd property. That is an inappropriately short road to a controversial and hotly disputed view, unless the 2 ~ my o knowledge, that assumption had never before been dragged out into the light and examined; but in the course of discussion a t SOFIA'S eighth conference, no fewer than six arguments were produced in its favor (most prominently the one that flows from Stalnaker's new support of Block). That was real philosophical progress, of a sort almost never achieved a t a single conference. Viva SOFIA! --except that I must now eventually confront and vanquish all the remaining arguments.

further IntInd property is just a narrow-functional property of the attitude in question (such as what Stich (1991) calls a "fat-syntactic" property). Which brings me directly to my second reply to Stalnaker's argument: Since narrow-functional properties can be IntInd properties, (5) hardly follows from (4), because (4) leaves it open that the IntInd property P is just some narrow-functional property of the experience e. And Block has argued independently that his Q-ualia are not narrow-functional properties because they are not functional; so a new case must be made for the suggestion that (4)'s IntInd property is a Q-uale rather than a narrow-functional property. My third and more important reply is that we have been given no reason to accept premises (1) and (3) together. Since (1) must be conceded for at least some time interval however small, then my claim is at least that we have no reason to grant (3) during that conceded interval. And that minimal claim is fairly obvious. Upon transportation to Inverted Earth, you notice no color difference introspectively or otherwise, but of course there is also no representational difference at first. The question is, whether as Block contends, a representational difference appears later, before an introspective color difference does. That is, is the interval At referred to in Stalnaker's argument large enough to keep (1) and (2) true while falsifying (3), by featuring representational decay prior to introspective color change? On the Millikanian psychosemantics I favor, representational change would take ages if it were ever to occur at all; as I said in my original paper, I do not buy Block's idea that representational change tracks easily with one's decision to speak as the locals do, or with anything else that is tied to public natural languages. If I am right, that is a reason to deny that there is a At large enough to keep (1) and (2) true while falsifying (3). But I have volunteered not to make an issue of psychosemantics. And even if Millikan and I are wrong, a further move is available. Block needs a longish interval during which all the relevant representational contents change but color sensations remain i.i. Perhaps the representational contents do gradually change, but what shows that the color sensations' Q-ualia (n.b. again, not the colors they involve) do not change with the representational contents? I suspect a transitivity assumption is at work here: If el is i.i. from e2 and e2 is i.i. from e3, then el is i.i. from e3 and introspective indistinguishability is maintained over a long interval by way of very short experience-pair intervals. It will surprise no one to learn that I do not grant any such transitivity principle. First, notice a scope ambiguity in the notion of

11. REPLIES T O TOMBERLIN, TYE,S TALNAKER AND BLOCK

133

"introspective indistinguishability": introspecting-not-different vs. merely not-introspecting-different. The latter notion is clearly not transitive; I could fail to introspect a difference between A and B and fail to introspect one between B and C even though I could never have failed to introspect one between A and C. And I doubt that even the stronger "i.i." notion is transitive either. The psychologists' concept of a just-noticeable-difference seems to apply to introspection as firmly as it does to any other form of feature detection, and, notoriously, there can be a j.n.d. between A and C even though there is no j.n.d. between A and B or between B and C. So we have not yet heard a convincing argument for the claim that the color sensations' Q-ualia do not change with the representational contents. One further argument might be that there is no plausible model for such a change. E.g., suppose that the Q-ualia are supposed to shift from "the way blue looks to S at t" to "the way yellow looks to S at t" (remember as always that Q-ualia are not, and do not correspond to, colors). A shift from the way blue looks to the way yellow looks might reasonably be supposed to be a smooth and gradual shift along the spectrum that passes through green. But it is hardly plausible that you (Block's subject) experience such a shift, or a period of unmistakeable greenness in particular. I reply that we really have no plausible model for the alleged representational shift either. How does something smoothly go from meaning blue to meaning yellow? Presumably not by meaning green in b e t ~ e e n . ~ I pause to address two further issues raised by Stalnaker. First, guilt by association (p. 104): He is quite right to insist that the kinship between Block's Q-ualia and narrow content is not, by itself, an argument against them. My real objection to them requires some Block exegesis and is given in Ch. 6 of Lycan (1996, in press). Very briefly, it is that Q-ualia as Block conceives them are weird. We are told,4 not just that what it is like for me to see a red object might differ from what it is like for you to see one, but that what it is like for me to experience subjective, phenomenal red (such as 3 0 n e suggestion would be t o understand t h e semantic shift in terms of Hartry Field's (1973) notion of "partial reference". I do not find t h a t notion a t all plausible, but there is not space t o digress on t h a t here. 4 ~ h following e characterization of Block's view is based what I consider t h e most natural reading of Block (1990). T h e doctrine he defends in "Mental Paint a n d Mental Latex" (this volume) differs somewhat from t h a t view, especially in lacking what I take t o be t h e view's weirdest feature, though I find t h e more recent doctrine objectionable as well. I shall say a bit more about this in t h e next section below.

the color of an after-image) might differ from what it is like for you to experience it. Once the latter modifiers have been italicized, the suggestion becomes hard to interpret, because (I believe) for most philosophical speakers, "subjective/phenomenal red" already names a particular way a sector of one's visual field looks, or what it is like to have that sector look that way. At best, the alleged distinction is elusive. Q-ualia now do not correspond in any obvious way to colors, for as Berkeley said, the only thing that resembles a color is a color; so what are they? When people think of qualia generally, the color of a visual sense-datum is a paradigm, and when we imagine having our spectrum inverted with respect to whatever, what we are imagining is the inverting of a spectrum, i.e., of a bunch of colors. If it is not that, we do not yet know what we are supposed to imagine, under the heading of "inverted Q-ualia" with respect to the same subjective spectrum. Second, Stalnaker goes on (p. 106) to ask what the world would have to be like in order for the "shape" representations I have ascribed to Peacocke's subject to be veridical, and he questions whether my differently-sized tree shapes do visually appear to the observer: ". .. that is not how things seem to appear to the observer". I agree that without philosophical training, the observer would not report such an appearance. But I think it remains true that the observer's experience visually represents the differently-sized shapes. Recall the giant peep-box I hypothesized in response to Michael Tye. That box is one way the world could be in order for my posited "shape" representations to be veridical, and so that way is one of the observer's visual alternatives, and so it is a member of the set of worlds that is the experience's propositional content. Stalnaker anticipates this point when (p. 107) he offers me the wording, "[Mly visual experience is the way it would be if I were seeing an image (a physical image) that contains two tree images, one bigger than the other"; just so.

4

To Ned Block's "Mental Paint and Mental Latex"

On the whole, Block characterizes my view correctly. In particular, I am what he calls a "quasi-representationist". I am a (teleological Homuncular) Functionalist about all mental features save (a) intentional properties such as attitude contents and representata, and his volume pp. 19-49

11. REPLIES TO TOMBERLIN,TYE,STALNAKER AND BLOCK

135

(b) qualia in my loose-Lewisian sense, since I know of no adequate Functionalist account of qualia in that sense. Wishing to minimize mystery, I assimilate (b) to (a) by hypothesizing that qualia are just representata (externalist of course). So, contra Block but as he seems to acknowledge in his fn 2, I do not "plug a gap that.. . [I] see.. . in representationism with [F]unctionalism"; rather the other way around, and I see nothing "contrived" about that. (Nor, should Block ever convince me that there are nonintentional Q-ualia in his sense, would there be anything contrived or ad hoc about trying to give a Functionalist account of them, for I am already a Functionalist of nearly twenty-five years' standing.) Block does get me wrong on one important point.6 I emphatically do not hold that the representational content of a red after-imageexperience, or of any other experience, is that there is a phenomenal individual of any sort. (Checking back, I see that the passage he cites does misleadingly encourage that interpretation.) My position is that what might have been thought to be a phenomenal individual is actually a physical, if inexistent, thing; the red after-imageexperience falsely presents a filmy, floating, evanescent red physical thing as located out in physical space, and the pain experience does portray disturbance or disorder in the physical leg. Thus, the real "sense-datum fallacy" was to mistake a (sometimes) nonactual physical object for an actual nonphysical one. Now, what arguments does Block offer against the representationist view? Despite being only a quasi-representationist myself, I shall defend pure representationism against him, partly out of amor belli and partly out of the conviction that representationism is much more plausible than it appears at first. (i) Orgasm (pp. 33-4; Block clearly means male orgasm). I grant that representationism is by no means obvious, as is illustrated by the cases of orgasm, smells before one has thought about them, and perhaps even pains (though I think pain is a poor example for Block's purpose, since pains not only represent damage or disorder in a body part, but they do so as one of their main functions7). Also, within limits, I accept responsibility for saying what the representational 61t is worth mentioning that over the years, Block and I have experienced severe mutual difficulty in understanding each other's main claims about the ontological structure of experience. Much of that difficulty has been terminological, but it is also clear that we look at the topic in different ways, surprisingly different given the similarity of our interests, backgrounds and philosophical upbringing. 7Block (1995, p. 281) offers a quick further antirepresentationist argument about pains: He says there are pains that differ phenomenally but "not in any way describable in words". I am not so pessimistic. Some people are very good

contents really are. My books being family publications, I have not investigated orgasm explicitly, but plausible efforts have been made by Tye (1995, 1996) and by Daniel Gilman (1996, forthcoming). Certainly the represented properties are ascribed to a region of one's own body (for that reason I have trouble taking seriously Block's suggestion that one might mistakenly attribute one's own orgasm to one's partner). They include at least warmth, squeezing, throbbing, pumping, and voiding. (On some psychosemantics, I suppose, impregnating is represented as well.) In accepting the burden of articulating representational content, I added the qualification "within limits" because, as Block anticipates, I think it is obviously impracticable to try to capture detailed perceptual and other sensory content in ordinary English words, at least in anything like real time. So the responsibility I accept is just that of giving some plausible general idea of what sorts of things are represented by sensations of this type or that, general but I hope specific enough to remove reasonable suspicion that there are nonintentional qualitative features left over in addition to the functional properties that are already considered characteristic of the sensation. (ii) Phosphenes (pp. 34-5). Phospheneeexperiences present (usually) colored points of light and larger shapes, having some spatial orientation in the subject's visual field. Do they do more than (re)present such things? Not that I can see, though I agree with Block that the example settles nothing on its own. I would add (though I know Block will resist this) that the points and shapes are represented as being in external, though dark, physical space; they are not just in the middle of my mind. Contra me, Block urges that "there is no guarantee that phosphene experiences produced by pressure or by electromagnetic stimulation could be produced by light" (p. 35). Perhaps there is no such guarantee, but why does Block doubt that the hypothetically veridical phosphene experiences could be produced by light, e.g., by a psychedelic movie as I suggested, or by real little lights in an otherwise very dark room?8 a t describing their pains, paradigmatically in representational terms like "throbbing", "burning", "sharp", "dull", "stinging" and the like. It would be even easier to describe the difference between two pains, since a comparison is afforded. No doubt our descriptive powers would run out before long, but the two pains would still feel as though different unfortunate things were happening in the respective parts of the body. ' ~ e r n a r d Kobes (1991) has made a related objection that has considerable merit, but it is science-fictional and succeeds only by being so.

11. REPLIES T O TOMBERLIN, TYE,STALNAKER A N D BLOCK

137

(iii) Bach-y-Rita's subjects' tactual sensations (pp. 35-6). I take Block's word that the subjects can become aware of the specific sensations in their backs (as such) that are doing the quasi-visual representing of external objects, and I agree that if so, the sensations have some mental properties other than their quasi-visual-representational properties. But as a representationist regarding bodily sensations, I hold that what the sensations are additionally doing is just representing the relevant portions of the subjects' backs as tingling, mildly perturbed or whatever. (Remember my Layering thesis. The sensations represent properties of external objects by representing conditions of one's back.) (iv) Marvin (pp. 39-40). Block asks "what, according to the representationist, is the difference between Marvin's concept of red and Marvin's concept of blue?" Answer: That the former represents the (or a) physical property of objective redness while the latter r e p resents objective blueness. How those concepts do that, exactly, remains to be settled by a credible psychosemantics, and as I have admitted, I have none of my own. But Block's only objection to the foregoing answer seems to be that Marvin does not know which colors are called, in English, by the words "red" and "blue". I have already rejected the idea that the use of particular English words has much to do with the psychosemantics of perceptual representations. (v) The memory argument in defense of the Inverted Earth example (pp. 43-5). In my reply to Bob Stalnaker, I suggested that introspective indistinguishability is intransitive and gradually melts away along with (because it is) the representational shift. Block counters that by appealing to memory: [Ylou remember the color of the sky on your birthday last year, the year before that, ten years before that, and so on, and your long-term memory gives you good reason to think that the phenomenal character of the experience has not changed.. .. Of course, memory can go wrong, but why should we suppose that it must go wrong here? (p. 44) As Block anticipates, I rejoin by noting the wideness of memory contents, memories being representations. On Block's own view, memory contents will undergo the representational shift. So when you say or think to yourself, "Yes, the sky is as blue as it was thirty years ago", you are not expressing the same memory content as you would have when you had just arrived on Inverted Earth. You are now remembering or "remembering" that the sky looked yellow, since for you "blue" now means yellow. And that memory is false, since on the long-ago occasion the sky looked blue to you, not yellow. That is my main nonrhetorical answer to Block's rhetorical question

of why we should suppose that memory must go wrong here. Not because of anything special about the relation between memory and phenomenal character, but because of Block's own hypothesis of representational shift: the shift allows i.i. memories to go from true to false.g Though he has anticipated this point, Block does not respond directly. Rather, he adds on my behalf that "[slince memory is powerless to reveal this shift [in phenomenal character], memory is by its nature defective" (p. 44). I agree with this as stated, so long as "defective" is understood externalistically and not as implying broken hardware. But then (here the dialectic gets complicated) Block seems to take the added remark I have just endorsed as a n argument for my rejoinder to his Memory objection to my reply to Stalnaker's Indistinguishability argument. For he says, But this argument is question-begging. The Inverted Earth argument purports to refute externalist representationism, so trotting in externalist representationism about memory to defend it is futile.. . . The defender of Lycan's view that memory is defective must blunt or evade the intuitive appeal of the first person point of view to be successful. It is no good to simply invoke the doctrine that experience is entirely representational. But the reply to the Inverted Earth argument as I presented it above does just that. It says that the memories of the representational contents are wrong, so the memories of the phenomenal contents are wrong too. [p. 44, boldface original] No, it does not do or say any such thing. But before we get to that, notice carefully that the added-and-endorsed remark Block calls question-begging was never an argument of mine against his position or, more to the point, against his Memory defense. My argument invoked only the (generally acknowledged) wideness of memory contents and Block's own stipulations about representational shift, not in any way the doctrine that experience is entirely representational or any other claim about experience. So the remarks of Block's just quoted fail to bear on anything I have said so far. In his next paragraph (p. 45)' Block says something that does address my rejoinder (even though he inaccurately bills it as something

lock adds, "So if my story is right, Lycan (if he is t o be an externalist about memory) should say that the subject's color experience has shifted gradually without the subject's knowing it" (p. 44). That is an odd remark, since I already had said t h a t the subject's color experience has shifted; that was my original response t o the Inverted Earth example. Externalism about memory has come u p only since, in light of Block's present criticism of my position.

11. REPLIES TO TOMBERLIN, TYE, STALNAKER AND BLOCK

139

by which "[wle could dramatize what is question-begging about the argument [viz., about the added-and-endorsed remark that was never a n argument of mine].)" He supposes that the protagonist of the Inverted Earth story understands both that his representational contents shift and, philosophically, why they have done so. But that gives him no reason to back down from his insistence that there is no difference in the way the sky looked to him (in one sense of that phrase), that if he could have both experiences side by side he would not be able to discern a difference. Plainly, he is justified in saying that there is no difference in something, something we could call the phenomenal character of the experience of looking at the sky. (p. 45) T h e argument seems to be that the subject is justified in claiming introspective indistinguishability across memories (and hence, presumably, indistinguishability obtains despite the representational shift). But if the subject does understand both that his representational contents shift and philosophically why they have done so, the subject is not justified in saying that there is no difference in memory content. For he would know that his state(-type) that used to represent blue now represents yellow, hence that his own thought that "the sky is still blue" is really the thought that the sky is now yellow, hence that there is a difference despite the putative introspective indistinguishability of the memories. (Remember, Block's present hypothesis is that the subject is philosophically hip; so the subject would already have predicted representational difference despite introspective sameness.) Now, why is he supposed to be justified in thinking that "if he could have both experiences side by side he would not be able to discern a difference", and that "there is no difference in something, something we could call the phenomenal character of the experience.. . " ? If anything is doing the justifying, it is surely the subject's memory. But on the standard wide construal of memory content, the memory content has changed, and presents a different color, so on that construal the subject's memory would not justify him in thinking that the experiences would be indiscernible or that there is a phenomenal "something" that has not changed. Therefore, either Block's argument presumes some narrow notion of memory content that and that notion accurately tracks a narrow kind of qualitative content, or it simply assumes that the subject has some other kind of access to an unchanging qualitative content over time. But the notion of narrow memory content is controversial all by itself, and that of a narrow and unchanging qualitative content is just the point

at issue. So if I have begged the question, I am not alone; and I do not see which assumption of mine it is that does beg. (Block adds a second argument (pp. 45-6); I parenthesize it because I do not follow it: Suppose he visits "a different language community in a different country, knowing full well that they cut up the colors differently". And suppose he "resolve[s] to reject.. . [his] old language community and be loyal to the new one", and that psychosemantically this shift of practices changes his representational contents as soon as it occurs. T h e change of course includes his memory contents. But there is no time here for 'gradual' but unnoticed change in what it is like to see red. The externalist view of memory would dictate that my memory of what it was like to see ripe tomatoes just a few moments ago before my decision is wrong. Five minutes ago, before my decision to change loyalties, I was looking at a ripe tomato. Now I have made the decision and I am looking at the same ripe tomato. It looks just the same -as far as I can tell. Yet the externalist about memory has to say that I mis-remember. To the obvious objection, that no such superficial linguistic decision could so quickly change the contents of my standard everyday propositional attitudes, Block replies, "My point is that if anything is wrong with the claim about this thought experiment that the representational contents of my experiences changed in a flash while their phenomenal characters stayed the same, it is this theory of color concepts, not taking memory a t face value" (p. 46). Having allied myself in my original paper with Millikanian psychosemantics, I agree that what is most obviously wrong with Block's judgment about this newest example is its guiding theory of coior concepts and representational contents generally (either perceptual contents or cognitive-attitude contents). If I am right, then I do not see the force of the example, since I know of no one who disputes that superficial linguistic decisions and loyalties may change while perceptual and attitude contents remain the same, and I do not see why anyone would doubt that of perceptual contents in particular. What if Millikan and I are, not just wrong, but as radically wrong as Block suggests, and some mental contents can be changed in a flash by a merely verbal decision? Even then, one may take different lines on perceptual contents and cognitive-attitude contents respectively. Perhaps belief and memory contents might shift, while visual contents remain determined by nature and by the structure and teleology of the visual system, not by verbal decision. That would explain Block's intuition (which I share) that the look of the

11. REPLIES TO TOMBERLIN, TYE,STALNAKER A N D BLOCK

141

tomato would remain the same even if belief and memory contents changed; that explanation strikes me as plausible, and it is still representationist and requires no Q-ualia. So if(!) I have understood Block's third argument correctly, it fails.'')

Block, N. (1995). "How Many Concepts of Consciousness?" Behavioral and Brain Sciences 18: 272-84. Field, H. (1973), "Theory Change and the Indeterminacy of Reference". Journal of Philosophy 70: 462-81. Gilman, D. (1996, forthcoming), "Consciousness and Mental R e p resentation", Behavioral and Brain Sciences. Kobes, B. (1991), "Sensory Qualities and 'Homunctionalism': A Review Essay of W.G. Lycan's Consciousness". Philosophical Psychology 4: 147-58. Lycan, W.G. (1994), Modality and Meaning (Dordrecht: Kluwer Academic Publishing). Lycan, W.G. (1996, in press), Consciousness and Experience (Cambridge, MA: Bradford Books / MIT Press). Stich, S.P. (1991) "Narrow Content Meets Fat Syntax", in B. Loewer and G. Rey (eds.), Meaning in Mind: Fodor and his Critics (Oxford: Basil Blackwell). Tye, M. (1991), The Imagery Debate (Cambridge, MA: Bradford Books/MIT Press). 'O~lock adds (p. 46) a parting shot against those whose psychosemantics require that the Inverted Earth subject's representational contents would never switch, no matter how long s/he stayed on Inverted Earth. Though I do not take the latter position myself, the argument is worth mentioning. It is roughly that a representationally unswitched subject would have an inverted spectrum relative to the Inverted Earth aboriginals, even though "you are as functionally similar to them as you like". But if spectrum inversion is achieved so easily, we could accomplish the same without leaving Earth; just raise twins, one with inverting lenses; the twins would "end up functionally and representationally identical, but phenomenally different". Just three quick remarks against that argument: First, remember the double relativity of the notion of an "inverted spectrum". Block's example is so far underdescribed in this regard. Second, whether two twins just one of whom has inverting lenses are "functionally identical" depends on where the lenses are placed, on how they work, and on one's chosen notion of "function" as well. Third, whether the twins are representationally identical depends on details of one's psychosemantics even if that psychosemantics is stipulated to be of the conservative Millikanian sort, though I would not want to rest much weight on that third point.

Tye, h1. (1992), "Visual Qualia and Visual Content", in Crane (1992): 158-76. Tye, M. (1995), "Blindsight, Orgasm and Representational Overlap", Behavioral and Bmin Sciences 18: 268-69.

PHILOSOPIIICAL ISSUES, 7 Perception, 1995

Phenomenal Externalism or If Meanings Ain't in the Head, Where Are Qualia? Fred Dretske

~ a t e r i a l i s t s 'should be willing to tolerate some degree of externalism about the mind. It is hard to see how to avoid it. Beliefs are prominent citizens of the mind, and beliefs are individuated in terms of what they are beliefs about. I know of no plausible psychosemantics, no plausible theory of what makes one thing about another, that isn't externalist in character. It is the relations -causal, informational, historical, or whatever- that, on a given Sunday afternoon, makes something in my brain about football rather than philosophy. Thoughts are in the head, but what makes them the thoughts they are is not there. Though prominent, beliefs (propositional attitudes in general) are not the only inhabitants of the mind. There are also experiences or sensations -the sort of mental state that exhibits a qualitative character. There is a big difference between seeing and hearing, and the difference does not consist in what (if anything) one ends up 'Who are realists (i.e., not eliminativists) about the mind.

believing. Nor does it seem to consist in what the experiences are experiences of. For the experiences associated with seeing Clyde play and hearing him play, though much different, can be experiences of the same thing -Clyde playing the piano. They are, nonetheless, quite different, and the difference seems to lie in the character of the experiences themselves quite apart from what, if anything, they are of or about. So even if externalism about thought is more or less plausible, externalism about experience is not. Even if physically identical heads can harbor different thoughts, they cannot be having different sensations. As I gauge the current situation in the philosophy of mind, this is the accepted wisdom, the orthodox picture (at least among materialists), about perceptual belief and perceptual experience. To put the point in Putnam's familiar form, Fred and Twin Fred may end up with different perceptual beliefs about the puddle they both see Fred that it is water, Twin Fred (miraculously transported to Earth and looking at the same puddle) that it is twater- but they must, if they are physically indistinguishable, be having the same experiences of the puddle. It must look the same to both of them no matter what differences might exist in their resulting beliefs. Perceptual beliefs may not supervene on the biological substrate -this, indeed, is what the Putnam and Burge examples show- but perceptual experience, sensations, qualia, do. If two individuals are physically the same, then, surely, the fact that they grew up in different worlds, had different histories, exist in different linguistic communities, or evolved in different ways, is irrelevant to their phenomenal state. If one has a headache, so does the other. If one is having a visual experience of the kind I have when I see water, so is the other. I want to challenge this orthodox picture of sense experience. I do so, not because I can show it is wrong. I can't. Nor because I regard it as intuitively implausible. I do not. I challenge it, rather, because I have no alternative, no other way of making sense out of another fact about the mind that I find equally (indeed, more) evident -the fact, namely, that the qualities that individuate one experience from another, the qualities that make seeing so much different from hearing, and seeing red so much different from seeing green, are (or need be) nowhere in the person wherein resides the experiences of these qualities. The experiences themselves are in the head (why else would closing one's eyes or stopping one's ears extinguish them?), but nothing in the head (indeed, at the time one is having the experiences, nothing outside the head2) need have the qualities that 'During, say, hallucination.

distinguish these experiences. How is this possible? How is it possible for experiences to be in the head and, yet, for there to be nothing in the head that has the qualities we use to identify and distinguish between them? One possible answer3 is the answer externalism provides: the qualities by means of which we distinguish experiences from one another are relational properties -perhaps (on some accounts of these matters) intentional properties- of the experiences. Just as we distinguish and identify beliefs by what they are beliefs about, and what they are beliefs about in terms of what they stand in the appropriate relations to, so we must distinguish and identify experiences in terms of what they are experiences of. Thus does externalism -and, as far as I can see (if we ignore dualism), only externalism- explain why the properties that individuate experiences (red, green, sour, sweet, hot, cold) are not (or need not be) properties of the experiences. The experiences are in the head, but what makes them the experiences they are -just like what makes beliefs the beliefs they are-- is external. That is why I think the orthodox view- the view that holds that sensations are (whereas beliefs are not) locally supervenient- must be challenged. There seems to be no other plausible way to makes sense of the qualitative character of perceptual experience. But this leaves us with a problem. The problem is to see whether, and if so, how, an externalist account of sensation (what I will call phenomenal externalism) can overcome the powerful intuition that sensations, unlike beliefs, supervene on what is in the head -that physically identical beings must be having the same sensations.

1 Experience as Awareness of Internal Objects Some philosophers may reject phenomenal externalism because of an adherence -if only implicit- to an act-object analysis of experience. According to this familiar (from sense-data accounts of perception) story, an experience of an external object is to be understood as an awareness of an internal image, an image that, in certain respects, and when things are working right, represents or resembles the external object. Seeming to see a red tomato, for example, is 3I ignore dualism. I'm exploring options open t o a materialist. It may b e supposed t h a t short-armed functionalism is another possible answer. I doubt it, but I will not here go into t h e reason (it has t o d o with t h e necessity of going t o long-arm functionalism t o capture t h e relevant distinctions, and long-arm functionalism is a version of externalism).

the mind's awareness of an internal datum that is red, round, and bulgy. For an object to look red is for it to cause a red datum to appear in the theater of the mind. The mind, the only spectator in this theater, becomes directly aware of this datum -its color and shape. One thereby becomes indirectly ("inferentially") aware of the properties of the external object that caused it. According to this way of thinking about experience, an external object's looking red to S is S's awareness of an internal object that is red. If this is what is involved in something looking red, then it is hard to see how anything could look different to two people who were physically the same. For if k (some external object) looks red to Fred, Fred must be directly aware of something inside himself that is red. If it looks green to Twin Fred, then Twin Fred must be aware of something inside him that is green. Since they have differently colored things inside them, the twins cannot -not if materialism is true- be in the same physical state. Physically indistinguishable beings must have the same experiences. This argument doesn't work with thoughts because thoughts are not conceived of as having the properties the objects thought about are thought to have. Thinking that k is red is not itself red nor does it consist of an awareness of something internal that is red. It is, in fact, not quite clear how thoughts must differ in their intrinsic properties to be different thoughts. Maybe thoughts are like utterances; two utterances (in different languages, say) could be the same (consist of exactly the same sounds) and, nonetheless, express quite different propositions, make quite different assertions. Maybe thoughts are like that: two tokens of the same physical type, because they occur in "different languages" (in different heads) manage to be different thoughts. One is the thought that k is red, the other the thought that k is green. Experiences of red and experiences of green, though, have to be physically different. Why? Because an experience of red and an experience of green, unlike a thought about red and green, are constituted by an awareness of an internal image or datum that is the color the experience is an experience of. Since the image or datum we experience when we experience different colors is subjective and, therefore, internal- different color experiences cannot occur in physically indistinguishable beings. If one took phenomenal experience to be an awareness of internal objects that had the properties external things appeared to have, it would be understandable why one would regard an externalist theory of experience as false. But, as we all know, one person's modus ponens is another's modus tollens. This argument is as larne a n d that is very lame, i n d e e d as is a sense-datum theory of perception.

Nobody -at least none of the philosophers I mean to be arguing with- is going to object to phenomenal externalism because (or simply because) it conflicts with a sense-data theory of perception. If there is something wrong with externalism about experience, it is something else. This, though, is only a hunch, a speculation about possible motivations. I suspect that some philosophers (though they would be loath to admit it) resist phenomenal externalism (without resisting conceptual externalism) because they subscribe, if only half-consciously, to an act-object account of experience. If that is, indeed, a motivation for the resistance, my hope is that exposure -making it explicit- will suffice to eliminate it. If different sense experiences are not experiences of different internal objects, why suppose that Fred and Twin Fred could not be having different experiences? They can have different parents, wives, and friends -not to mention (if conceptual externalism is true) different thoughts and memories. Why not different experiences?

2

Phenomenal and Doxastic Appearances

In the case of perceptual experience, qualia, the properties by means of which experiences are identified and distinguished from one another, are the way things s e e m or appear in the sense modality in question. If k looks red to me and green to you, then our experiences differ. We have different qualia. So if yonder puddle (I will call it k) doesn't look the same to Fred as it does to Twin Fred, then their experiences of the puddle are different. They have different visual qualia. If this is what it means to have different experiences, then the question we are asking about Fred and Twin Fred is: despite being in exactly the same physical state, could the puddle they see4 (from the same angle, distance, etc.) look different to them? 4 ~ order n t o keep the focus clear, I will continue to suppose that Twin Fred, after growing u p in a n alien environment, has been miraculously transported t o Fred's habitat, and is looking a t the same stuff: k . Differences in what they believe about what they see will thus be determined not by a difference in the object they see (what their beliefs are beliefs about), but by what they believe about it -the conceptual content of their perceptual beliefs. For the sake of exposition, I will also use the property of being water. I do not, however, think it necessary to run t h e example with natural kinds (if water is, indeed, a natural kind). W h a t is important t o the argument is that what Fred sees k t o be (call it F ) is external in the sense that there are circumstances in which non-Fs can be made t o affect Fred in exactly the way Fs do. If this condition is met, then it will be possible t o construct Twin-Earth type situations

There is, of course, a sense in which k (the puddle of water) doesn't look the same to Fred and Twin Fred. It looks like water to Fred, like twater to Twin Fred. This is how they will describe the appearance of the puddle.5 I will call this sense of appear-verbs the doxastic sense. It signifies a sense of "look" ("appear", "seem", etc.) which implies something about the beliefs (or belief dispositions) of the person to whom something looks F. To say that k looks F to S in this doxastic sense is to say that S believes, or tends to believe, or would believe if he didn't have countervailing evidence, on the basis of his perception of k, that k is F. Since Fred and Twin Fred are caused to believe different things about k on the basis of their perception of it, k looks different to them. If we distinguished experiences on the basis of differences in the way objects (like puddles) looked in this doxastic sense, then Fred and Twin Fred would be having different experiences of the puddle. This, though certainly true, merely shows that experiences should not be classified in terms of how things appear in a doxastic sense. Qualia, the features in terms of which experiences are identified and distinguished from one another, are, instead, to be understood in terms of the way things seem in a phenomenal sense. Whatever else the phenomenal sense of "appear" ("look", "seem", etc.) is supposed to be (it is not always clear) it seems safe to say (at least in this context6) that it is supposed to be independent of the doxastic. It is independent of the doxastic in the sense that k can look, phenomenally, the same to two people without looking the same to them doxastically. If we denote the phenomenal appearances with capitalized words (LOOK, SEEM, APPEAR) and doxastic by words in lower case (look, seem, appear), then things can LOOK the same to S without looking the same (phenomenal sameness does not require doxastic sameness), and they can LOOK different without looking different (phenomenal differences do not require a doxastic difference). For anyone who understands, or thinks she understands, what the phenomenal is supposed to be, the first point is more or less obvious. If I am able to recognize petunias when I see them and a two-year old child cannot, a flower might look like a petunia to me but not to the child. I am led to believe it is a petunia on in which Fred and Twin Fred can be imagined t o be in t h e same physical state while differing in t h e sort of external relations t h a t (according t o conceptual externalism) determine t h e content of their thoughts. 5 0 f course, Twin Fred will utter t h e same words ("It looks Like water"), but t h e word "water" in Twin Fred's mouth means twater, not water. '"Safe t o say" since otherwise there would be no basis for resisting a n externalist theory of experience by those embracing a n externalist theory of belief.

the basis of my perception of it, but the child is not. Yet, though the flower looks different to us, we can easily imagine it LOOKING the same to us: our visual experiences are similar -perhaps, given that we have the same visual acuity and are looking at the flower from the same angle, at the same distance, in the same light, and so on, exactly the same. So phenomenal sameness does not require doxastic sameness. The second point -that phenomenal differences do not require doxastic differences- is an immediate consequence of inverted spectrum possibilities.7 In the phenomenal sense, it is possible for something to LOOK different to us -red to me, green to you- when we both believe (and say) the same thing about it -that it is red.8 So things can LOOK different to us without their looking different to us. Though they LOOK different, we end up believing the same thing. Aside from maintaining this independence of phenomenal from doxastic appearances, I will not attempt to define the phenomenal sense of appearance verbs (I know of no successful effort to do so). It will be enough for my purposes if we operate with an intuitive idea, one that is, in the respects just specified, independent of what we believe, or are disposed to believe, about the things that appear this way. This will be enough, I think, to argue that if you are an externalist about belief -a conceptual externalist- there is no reason not to be an externalist about experience -a phenomenal externalist. If you are an externalist about how things seem- as most materialists are these days -you can also be what few (if any) materialists are- an externalist about how things SEEM.

3 The Accessibility of Phenomenal Appearances Whatever is the case with their phenomenal experience of the puddle (whether it is the same or different), what Fred and Twin Fred think about their phenomenal experience will differ. If we can speak of how 7 0 f course, some people do not believe in the possibility of inverted spectra. But these are precisely the people who are sceptical of phenomenal qualities and should have no trouble accepting an externalist theory of experience if they accept an externalist theory of belief. '1 assume here that what we say (and believe) of the external object is not affected by this phenomenal difference. I assume, that is, that, having grown up in the same linguistic community, etc.), we both mean the same thing by "red". I shall return in a moment t o possible differences in belief about the way things appear in the phenomenal sense (i.e., possible differences in belief about the way they APPEAR).

one's experience of k (as opposed to k itself) seems to one, Fred's experience of the puddle will not seem to him the way Twin Fred's experience of the puddle seems to him. Fred will think c o r r e c t l y as it turns o u t that his experience of the puddle (the way the puddle LOOKS to him) is exactly like his experience of water (is exactly the way water normally LOOKS to him). So Fred will think o n c e again c o r r e c t l y that k LOOKS like water to him. Twin Fred, on the other hand, will think that k LOOKS like twater to him. He will think -correctlythat k LOOKS the way twater always looked to him in normal conditions. The twins' phenomenal experience of the water will prompt in them, not only different beliefs about the water, but different beliefs about their phenomenal experience of it. How their experience seems to them will therefore be different. It may be supposed that this does not follow. Just because water is not twater does not mean that water does not LOOK (phenomenally) like twater. Babs and Betsy are not the same person, but, being twins, they LOOK alike. Although believing something is Babs is not the same as believing it is Betsy, believing k LOOKS like Babs is the same as believing k LOOKS like Betsy since Babs and Betsy LOOK the same. For the same reason, Fred's belief that k LOOKS like water is really no different from Twin Fred's belief that k LOOKS like twater since LOOKING like water is LOOKING like twater. So if Fred thinks that k LOOKS like water and Twin Fred thinks k LOOKS like twater, then what Fred and Twin Fred think about the LOOK of k is the same. This is not so. The belief that k LOOKS like Babs is a different belief from the belief that k LOOKS like my sister even if, in fact, Babs LOOKS like my sister. Even if, in fact, Babs is my sister. And we are now making a claim, not about the LOOK of Babs and the LOOK of my sister (these might be the same), but about the LOOKS-like-Babs-belief and the LOOKS-like-my-sister-belief. These are not the same. I can have the one without having the other. So the twins' beliefs about their experiences of k, their belief about the way k LOOKS to them and, thus, the way their phenomenal experience seems to them, must be different. As a result of his perception of k Fred is prompted to have a LOOKS-like-water belief about k while Twin Fred, seeing the same k, is caused to have a LOOKS-like-twater belief. Hence, their phenomenal experience of k, as defined by the way k LOOKS to them, seems different to them. Their experience of k prompts them to have not only different beliefs about k, but different beliefs about their own phenomenal experience of k. Even if they have the same phenomenal experience of k, it will not seem that way to them.

There is an important philosophical lesson in this result. It is this. The access one has to the quality of one's experience (unlike the access one has to the qualities of the external objects this experience is an experience of) is only through the concepts one has for having thoughts about experience. One does not (I a m assuming) have experiences of one's puddle experiences as one has experiences of puddles. Hence, there is no way one can become aware of the phenomenal qualities of one's experience of the puddle (as one can become aware of the qualities of the puddle itself) except through a belief that it has this quality e x c e p t , that is, via a belief that something (external) LOOKS to have this quality. By experiencing (e.g., seeing) a puddle, S can be made aware of its watery quality (i.e., it can LOOK watery) without S having the concept of water -without the puddle looking like water. One cannot, however, become aware of the watery LOOK of the puddle without such a concept. For in the absence of experience of one's puddle experience, awareness of the watery LOOK of a puddle can only be awareness that the puddle LOOKS watery, that one is having a phenomenal experience of this kind, and this requires the concept WATER. What this means is that If k does not look F to S in the doxastic sense of "look", then even if it LOOKS F to S, S will not (S cannot) be made aware of this fact. If we imagine, furthermore, that the reason k does not look F t o S is that S completely lacks the concept of F so that, until he acquires the concept, nothing can look F to S, then although something can still LOOK F to S, although S can still have F-ish experiences, S cannot be made aware of this fact. Without the concept of F , S is "blind" to the F-ish quality of his phenomenal experience though not, of course, to the F-ish quality of the external objects this experience is an experience of. He cannot be made conscious of it. His experience can have this quality -things can LOOK F to h i m but he cannot be made aware of it. With apologies to Kant, without concepts we are blind to our intuition^.^ Let me illustrate this result with two examples. Consider, first, a case in which k LOOKS F to S and, though S has the concept F (is, therefore, able to believe that k is, or LOOKS, F ) , nothing looks, nothing seems, nothing appears F to S.1° S is in no way caused to

ant said t h a t without concepts, intuitions were blind. If intuitions are experiences, it isn't t h e intuitions t h a t are blind; it is by their means, by t h e fact t h a t they occur in us, t h a t we are made aware of (i.e., see and hear) t h e objects around us. I t is, rather, t h e intuitions t h a t we cannot be made aware of. ''There are a variety of reasons why, though possessing the concept F, things might not look F t o S even though they LOOK F. Inattention is probably t h e

believe that anything is F by k's LOOKING F to him. You hold up seven fingers, and I see all seven. Without enough time to count, but enough time to see all seven, I mistakenly take there to be eight. That is how many I think I see. When asked how many fingers there appear to be, I say eight. On the basis of my perception of the fingers, that is how many I think there are. That, therefore, is how many there appear (doxastically) to me to be. But how many does there APPEAR to me to be? What are the phenomenal facts? Believing that I see eight fingers, I will (if I understand the question) describe the phenomenal appearances by using the number "eight". I have no choice. If I didn't think there APPEARED to be eight fingers, why would I think that I saw -and that, therefore, there w e r e eight fingers? I must think the fingers produced in me an experience of the kind (numberwise) that eight fingers normally produces in me. In this case, though, I am wrong. Given that (by hypothesis) I see only seven, the correct answer to a question about how many fingers there APPEAR to be is seven. At the phenomenal level seeing (exactly) eight fingers is not the same as seeing (exactly) seven fingers. How could it be? There is as much difference between seeing eight fingers and seeing seven fingers as there is between seeing one finger and seeing two fingers. The only difference is that, as the numbers get larger, the difference is harder to appreciate, harder to detect, harder to notice. That is, as the numbers grow larger, things do not as easily seem to be what they SEEM to be. There is, therefore, a fact about my experience t h a t there APPEARS to me to be seven fingersof which I am not conscious. Knowing how to count -having the concept SEVEN- I can, of course, become conscious of (i.e., learn) this fact by counting the fingers.'' I will then become aware that most common reason: I must have seen t h e color of your tie (I looked straight a t it several times), but I was so engrossed in our conversation t h a t I didn't pay any attention. T h a t is t o say, your tie must have LOOKED blue t o me, but, because of inattention, it did not look blue t o me (did not cause me t o believe it was blue). Sensory overload is another: you see more t h a n you can possibly have beliefs about. For this possibility, Sperling's (1960) experiments are particularly suggestive: subjects see a n array of numbers under very brief exposure conditions. They can identify, a t most, three or four items, but t h e fact t h a t they can identify a n y three or four (which ones they identify depends on later cueing) suggests t h a t , a t t h e phenomenal level, there is information about all the letters in t h e subjects' experience of them. T h a t is, a n unidentified "5" must LOOK like a "5" even though, because of such brief exposure, it does not look like a "5". For more exotic instances of t h e disassociation of phenomenal and doxastic consciousness, there is also sleep walking, hypnosis, various types of associative agnosia, commissurotomy, and unilateral visual neglect. "Assuming t h a t one (originally) saw all the fingers one (thereafter) counts.

my experience of the fingers (before I counted) was not what I took it to be: it was a seven-finger experience, not (as it seemed to me a t the time) an eight-finger experience. Hence, when the way things SEEM # the way things seem, the person is necessarily unaware of how things SEEM. If things SEEM F to S, S is not aware that this is so.12 Consider, next, a case in which S lacks the relevant concept the concept needed to make k appear to S to have the properties it APPEARS to S to have. We are listening to a recording of Goible's 93rd Symphony. Thinking there has just been a change of key, but uncertain about it, you turn to me and ask whether it didn't sound that way to me. I am musically ignorant. I do not know what a chan e of key is, and I do not know what a change of key sounds like.'' So I answer "No" to your question. It did not sound to me like they changed key. If truth be told, nothing ever sounds like a change of key to me. To suppose that anything sounds like a change of key to me -someone ignorant of what a change of key is and what a change of key sounds like- is like supposing that an attractive young woman look like your sister to me. If I don't know who your sister is or what she looks like, how can anyone look like your sister to me? Clearly, in this exchange about the music, I am describing how things sound to me in what I have called the doxastic sense -what I believe, or would normally believe, about what I hear on the basis of hearing it. But if we take your question to be a question not about how the music sounds to me, but how it SOUNDS to me, a question about my auditory experience, not my perceptually induced beliefs (or tendencies), then the answer is not so clear. Not knowing what T h e possibility exists, of course, t h a t , in counting t h e fingers, one sees (sequentially) more fingers than one (originally) saw (all a t t h e same time). But, often enough, and despite Dan Dennett's (1991) protests t o the contrary, when t h e numbers are small, it is entirely reasonable t o suppose t h a t one saw all of the fingers one counts. One just doesn't (without counting) know how many one sees. As Farah (1990, 18) points out, counting requires seeing more t h a n one object a t a time. I t seems clear t h a t it also requires seeing more t h a n one knows (though not necessarily more t h a n one thinks) one is seeing. Why else would one count? See Perkins (1983) for helpful discussion on this point. 12For more on this, see Dretske (1993). 1 3 ~ v e nif one knows what a change of key is, one may not be able t o hear it. If I know enough musical theory t o know what a change of key is, but lack t h e capacity t o identify changes of key auditorily, then when I hear a change of key (when it SOUNDS t o me like a change of key), I know what t h e property is t h a t I a m hearing but I d o not know and cannot tell (at least not by hearing) whether I a m hearing it.

a change of key SOUNDS like, I a m not an expert on what the music SOUNDS like to me. CHANGE O F KEY is not a concept I have available for describing my auditory experience. Hence, even if the music SOUNDS to me exactly the way changes of key normally SOUND to me,14 I cannot, become aware that this is so. A conceptual deficiency "deafens" me, not to the quality of the sound (I hear -thus I experience -the change of key), but to this quality of my experience. The fact that a person would not say, does not think, perhaps would even deny, that k LOOKS or SOUNDS F does not show that k does not LOOK or SOUND F. All it shows is that the person isn't aware that it LOOKS or SOUNDS this way. How a m I -someone who has never met your sister- supposed to know whether the woman we now see LOOKS like your sister to me, whether she is causing in me an experience which is like the experience your sister causes (or, if I've never seen her, would cause) in me under normal viewing conditions? YOUare probably in a better position to say whether she LOOKS like your sister to m e than I am. And this holds, not only for concepts like SISTER and CHANGE O F KEY, but for all ways of describing the way things LOOK. If I don't know what it is to be red, I cannot be made aware of the fact that anything LOOKS red to me, that my experience has this qualitative character. Without the concepts, things will still LOOK red and green to me. I will still be aware of these colors, but I will be aware of them in the same way I a m aware of seven fingers (when I mistakenly think there are eight) and a change of key (when I don't know what a change of key is). I will have no awareness that I am aware of them. It will not seem to me as though I a m aware of them. This is not simply a matter of not knowing the right labels or words for perceptual differences. It is, rather, a matter of lacking the right discriminatory powers. We are sometimes aware of a dzfference between the way two things APPEAR (k and h LOOK different, and we realize this) without knowing how to describe the APPEARANCE of either k or h. S can taste the difference between Coke and Pepsi, but he doesn't know which is which. He may not have heard of these soft drinks -may not, therefore, know that this is the Pepsi taste and that the Coke taste. He can discriminate the two 14And this is different from what no change key SOUNDS like t o me. It is important t o add this qualification since nothing need SOUND like a change of key t o me just because it SOUNDS the way changes of key normally SOUND t o me. If I am virtually deaf and almost everything SOUNDS the same to me, then changes of key need not be part of my phenomenal experience even if the music causes in me an experience just like the experience a change of key causes in me. I here ignore this complication since it is not relevant t o the point I a m making.

tastes, consistently distinguishing Pepsi from Coke, but not know that this is the Pepsi and that the Coke taste. Nonetheless, if S responds (or can respond; he may not choose to respond) to one taste in a way different from the way he responds to the other, we give S credit for rudimentary knowledge of taste types and, for present purposes, such discriminatory power with respect to taste types is sufficient for crediting S with rudimentary knowledge of what the liquids TASTE like. Coke and Pepsi not only TASTE different to S, they taste different to S -not like Coke and Pepsi, of course, but like this (call it the P-taste) and that (call it a C-taste). Words and language are irrelevant. The key question is whether S groups or classifies liquids according to taste. If so, then whatever quality or qualities it is that leads him to do so are the qualities that Pand C-tasting liquids seem to S to have. If S does not distinguish between Coke and Pepsi, has no tendency to classify or sort liquids in this way, then even if Coke TASTES different from Pepsi to S, even if S experiences this sensory quality, it will not taste that way to him in the doxastic sense. He will have no thought of the form, "Aha, this tastes the same as that, and these are different". This is the situation I imagined myself to be in when listening to the music. Even if I heard the change of key, I did not -not even mentallygroup or classify changes of key together. I do not, as a result of hearing sounds with this auditory quality, think, believe, or judge them to be the same in some respect. That, and not simply because I lacked the words, is what made me "deaf" to the change of key I heard. That is why it (may have) SOUNDED like a change of key to me without it sounding like a change of key to me -without it being able to sound like a change of key to me until I train myself to hear it.15

4

Conclusions

Where does this leave us with respect to Fred and Twin F'red? None of this shows -it was not intended to show- that they have different experiences of the puddle, that the puddle LOOKS different to them. That was not my purpose. The strategy is not to show that experience is externally grounded (I don't know how to show that), but that it could be and that denials that it could be are based on a 15Strictly speaking, of course, one doesn't train oneself to hear it. One is already hearing it. What training provides is the ability to consistently distinguish -to make things that SOUND different sound (doxastically) different.

faulty picture of the relation between thought and experience. I have tried to do this by showing that qualia -experiential qualities- are not, as it were, on display in the shop window of the mind. Awareness of phenomenal properties (that one is experiencing redness, the taste of strawberries, or a change of key) is not achieved by a process of direct inward inspection, a process in which one becomes aware of the qualities of experience in the way one becomes aware of the qualities of the external objects that that experience is an experience of. An awareness of phenomenal properties (that something LOOKS red, that one is, as they say, experiencing redly) is a much more indirect process, a process that requires the possession and use of the concepts needed to think that something is (or LOOKS) red. When illumination is normal and there is a red object in front of you, you are made aware of the color red by merely opening your eyes. You do not need the concept RED to see red, to experience this quality. But you do need this concept to become aware of the quale red, to become aware that you are having an experience of this sort. This being so, qualia (understood as the qualities in terms of which we identify and distinguish experiences) necessarily remain "hidden", inaccessible, until one acquires the conceptual resources for becoming aware of them. You can become aware of the color red without the concept RED (just as -see a b o v e I can be made phenomenally aware of seven fingers without the concept SEVEN), but you cannot be made aware of the quale red without this concept. For to become aware of the quale red is to become aware that one is having an experience of a reddish sort, and this is something one cannot be made aware of without understanding what it means to be red. If (as we are assuming) Fred and Twin Fred have different concepts for sorting and identifying the objects they perceive, then, even if they are having the same phenomenal experience of k , the quality of this experience, what it is that makes it the same experience, will be inaccessible to them. They cannot be made aware of this quality. What Fred's experience seems like to Fred will not be what Twin Fred's experience seems like to Twin Fred. If, despite this difference in the way their experiences seem to them, these experiences are, nonetheless, the same (they are, let us say, both of the Q-ish sort), then the quale Q will be a quality that neither Fred nor Twin Fred can be made aware of. They will be introspectively blind to that aspect of their experience that makes it the same experience. They will be as completely unaware of the phenomenal quality Q as I was to a change of key. From a subjective standpoint, it will be as if their experience (of the puddle) was not Q.

If a conceptual externalist (that is, an externalist about belief) is willing to accept this result, then she has, I think, taken a big step. She no longer has a reason to resist phenomenal externalism (= externalism about experience). For if one accepts the conclusion of this line of reasoning, then the qualities of experience which allegedly must be the same in physically identical persons are qualities that the persons in question may not be able to become aware of. If Fred and Twin Fred are having the same phenomenal experience of the puddle, the respect in which their experience is the same is not a quality of their experience that they can be made aware of. They cannot be made aware of it because of the conceptual differences that exist between them. If they cannot be made aware of Q, though, why suppose that Q is an intrinsic property of their experience? If the only way Fred and Twin Fred can be made aware of Q is via an appropriate change in their external relations (those that ground their concepts), what better explanation is there for why this is so than that Q itself is externally grounded. This, indeed, seems like the most plausible explanation of why Q, a feature of one's experience, remains inaccessible to one without the right concepts. Even if one concedes this point, though, it does not mean phenomenal externalism is true. It only means that, from the point of view of the conceptual externalist, there is no longer any reason to think it is not true. Hence, it might be true. That, though, is what I set out to show. The conclusion can be put this way. Either phenomenal experiences are identified with thought-like entities -such things as potential beliefs (Armstrong 1969), suppressed inclinations to believe (Pitcher 1971) or micro-judgments (Dennett and Kinsbourne, 1992)-in which case the conclusion that sensations have their p r o p erties externally grounded if thoughts do follows trivially; or (as in the present study) phenomenal appearances (= the way things APPEAR) are distinguished from thoughts (from the way they appear). If experiences are distinguished from thoughts, so that k can LOOK F to S without S believing, or being disposed to believe, that anything is F (i.e., without anything looking F to S), then it turns out that qualia can be completely inaccessible (in the way the music's SOUNDING like it was changing key is completely inaccessible to one who doesn't know what a change of key SOUNDS like). If one takes qualitative states as essentially knowable, as many philosophers do, this is not possible. If, on the other hand, one is willing to tolerate unknowable qualia, what reason is there to insist that physically identical beings share the same qualia? Indeed, what better explanation is there for why conceptually different twins (e.g., Fred

a n d Twin Fred) cannot b e made aware of t h e qualia they allegedly share t h a n t h a t they don't really share t h e s a m e qualia? T h e puddle Fred a n d Twin Fred see doesn't LOOK t h e same t o them.

Armstrong, D.M. 1968. A Materialist Theory of the Mind. New York: Humanities Press. Dennett, D.C. 1991. Consciousness Explained. Little Brown. Dennett, D.C. and Kinsbourne, M. 1992. Time and the observer: The where and when of consciousness in the brain. Behavioral and Brain Sciences, Vol 15, No 2, pp. 183-247. Dretske, F. 1993. Conscious experience. Mind, vol. 102.406, pp.263-283. Farah, M. J. 1990. Visual Agnosia. Cambridge, Mass.: MIT Press. Perkins, M. 1983. Sensing the World. Indianapolis, Ind.: Hackett Publishing Company. Pitcher, G. 1971. A Theory of Perception. Princeton, N.J.: Princeton University Press. Sperling, G. 1960. The information available in brief visual presentations. Psychological Monographs, 74, No. 11.


Dretske's Qualia Externalism Jaegwon Kim

Fred Dretske is surely right when he says that most of those who accept content externalism believe, or simply assume, that qualia are internal, or, if you prefer, are "in the head". In fact, externalists are apt to acknowledge that these are the only purely internal residues of mentality. And if you are a physicalist of any stripe, as most of us are, you would likely believe in the local supervenience of qualia -that is, qualia are supervenient on the internal physical/biological states of the subject. In his interesting and thought-provking paper Dretske wants to undermine this widely shared assumption; he wants to promote "phenomenal externalism", as he calls it. This is the claim that, although experiences are "in the head", "what makes them the experiences they are" (p. 144) -that is, the properties or qualities by which we identify and individuate them- aren't in the head. (Dretske also puts this by saying that these individuating properties of experiences are not properties of experiences, but I think this is needlessly misleading, since on his externalism they turn out to be relational properties of experiences.) Dretske doesn't come flat out and say that phenomenal internalism is false and externalism is true, much less that he has shown this to be true. What he says is more cautious -that if you accept content externalism, there is no reason not to accept phenomenal

externalism. This is a doubly hedged claim: first, if you don't accept content externalism, the argument for phenomenal externalism need not apply. Second, even if you are committed to content externalism, that only permits you (perhaps also encourages you), but it does not compel you, to be a phenomenal externalist. At least, that's what he seems to be saying. At any rate, let us begin by focusing on Dretske's argument for his main claim, that is, the claim that "just as we distinguish and identify beliefs by what they are beliefs about, and what they are beliefs about in terms of what they stand in the appropriate relations to, so we must distinguish and identify experiences in terms of what they are experiences of" (p. 145). So what is Dretske's argument? I should, first of all, confess that I am not at all confident that I got all of his argument right, that I have properly understood just how it's supposed to work. But before we begin, let us remind ourselves of Dretske's distinction between doxastic looks (and other sensory appearings) and phenomenal LOOKS (and appearances), and follow him in using the lower-case "look" for the former sense and the capitalized "LOOK" for the second sense. With my strong caveat about the accuracy of my understanding of Dretske's argument firmly in place, here then is how I think the argument goes (or can go). I have no problem with Dretske's initial statement on how the two types of looks are related to possession of concepts: (1) Although something can (doxastically) look F (like F , like an F , etc.) to Fred only if Fred has the concept F (of F , of being F, being an F, etc.), it can (phenomenally) LOOK F to Fred even if Fred is not in possession of the concept F.

Let Fred and Twin Fred be looking at a puddle of water k. Their experiences of k (k's LOOKING a certain way) prompt them to have beliefs about the phenomenal character of their experiences. (2) Fred forms the belief that k LOOKS like water, and Twin Fred forms the belief that k LOOKS like twater. And these beliefs differ in content. I think Dretske is right in claiming that these two beliefs have different contents. One way of seeing this is to notice that if Fred also holds the belief that alcohol LOOKS like water, he could, or would, infer that k LOOKS like alcohol, but he would not make that inference in conjunction with the belief that alcohol LOOKS like twater. And of course conversely with Twin Fred. However, what isn't clear to me is that the beliefs mentioned in (2) must be the only phenomenal beliefs Fred and Twin Fred can have. It

seems to me that Dretske construes the content sentence "k LOOKS like water" relationally, as having the logical form "x LOOKS like y", and I agree that on this reading it plausibly follows that to have a belief with this content, one must have the concept of water, or some causal/epistemic/historical contact with water. But why isn't it also possible for Fred, and Twin Fred, to form beliefs like "k LOOKS shiny" (the morning sun is shining on puddle k), "k LOOKS round", and, indeed, "k LOOKS watery", where "watery" doesn't have the relational sense of "like water" but a nonrelational sense, something like "having wateriness". What is wateriness? Well, we can think of it as a loose, somewhat indeterminate, combination of the observable properties we associate with water, like transparency, the way it flows and reflects light, the tactile feel when it flows over your skin, its taste, and the like. It is an explicitly shared assumption of the countless twin-earth tales that have flooded philosophy that water and twater are observationally indistinguishable, and I take this to entail that wateriness = twateriness. But none of this needs to be thought to impugn the following conclusion Dretske draws from his discussion of beliefs about phenomenal properties of our experiences: (3) "The access one has to the quality of one's experience (unlike the access one has to the qualities of the external objects this experience is an experience of) is only through the concepts one has for having thoughts about experience" (p. 151). For I am willing to go along with the claim that for Fred to have the belief with content "k LOOKS shiny" or "k LOOKS watery", Fred must have some idea of what it is for something to be shiny or watery. If this is what "having a concept" comes to, I have no problem with (3). Note, and remember, though, that if I am right, Fred and Twin Fred can share phenomenal beliefs with the same content, e.g., belief that k LOOKS shiny, that k LOOKS watery (in the nonrelational sense), etc. If this stands, it may considerably restrict the scope of Dretske's phenomenal externalism -it may apply only to some phenomenal beliefs, not to all. I also have some trouble with the term "access" used in (3). For one thing I am not clear just what Dretske intends to mean by "access" in this context, and why and how he moves from having belief about phenomenal states to having access to them. Is having belief that p the same thing as having access to (the information that) p? Or does having belief only imply having access? And access for what purpose? One way in which the idea of access has been used in discussions of consciousness is in the sense of the availability

of information for verbal reports ("reportability") and for guidance of behavior ("executive access", we may call it).' Plausibly, if the content expressed by "k LOOKS shiny" is to be available for verbal report, the subject needs the concept of shininess. However, it is dubious that for Fred's experience (k's LOOKING shiny) to be available to his "executive module" for behavior guidance, he must possess the concept of shininess. In fact, it isn't clear that he needs to have any beliefs about k or k's shininess. It is part of Dretske's argument that experiences as such do not require concepts, that for k to LOOK watery to Fred, Fred need not possess the concept of water. Dretske may well be right in saying that for Fred to have the belief that he is having an experience of this kind, he needs the concept. What isn't clear is why the concept is needed for Fred to have "access" (in particular, executive access) to this information. And it seems to me that executive access captures one important sense, or kind, of consciousness or awareness. At any rate, as far as I can see, the only sense of access in which Dretske has established (3) is access in the sense of the subject's having a belief, a full-blown propositional belief, about the qualitative character of his/her experience. Notice, though, how he apparently continues to slide on, from "having access" to "being aware of" and "awareness", and then to "being conscious". Does being conscious of the watery LOOK of the puddle require the possession of the concept of water? As I take it, Dretske reasons toward an affirmative answer in something like the following way: being conscious of the watery LOOK of the puddle involves having a belief that the puddle LOOKS like water, and this belief requires the concept of wateq2 hence, being conscious of the watery LOOK requires the concept of water. But this seems wrong: if having executive access is a species of being conscious, this form of consciousness could very well bypass belief in any conceptual sense. Fred is conscious of the watery LOOK of the puddle, so he walks around the puddle rather than stepping h his roughly is the sense of "access consciousness" Ned Block develops in his "On a Confusions About a Function of Consciousness", The Behavioral and Brain Sciences 18 (1995): 1-41. ' ~ o t ethat the principle apparently involved here -namely that for S t o have a belief with content p, S must be in possession of all the concepts t h a t p involves- is anti-externalist. As I take it, Burge-style externalists would reject it: in Tyler Burge's well-known arthritis case, the subject believes that he has arthritis, but doesn't have the concept of arthritis. It seems clear that the rejection of this principle is central t o Burge's externalist arguments. See Burge, "Individualism and the Mental", Midwest Studies i n Philosophy 4 (1979): 73-121.

into it. Animals to which we are not at all inclined to attribute linguistic/conceptual capacities will do exactly Fred does. I don't see why Fred needs to have a belief, in its full-fledged conceptual sense, about the watery LOOK of his phenomenal experience. Apart from this issue, I have a few further questions about Dretske's argument. Remember that the conclusion of Dretske's argument is this: (4) The qualities by which our experiences are identified and individuated are not (intrinsic) properties of experiences. That is, experiences are externally individuated. I have already raised a question about the scope of (4), which as stated covers all experiences. But moving on, I am not clear how (4) follows from (3). What (3) says is only that our beliefs or awarenesses about the phenomenal qualities of our experiences require the possession of concepts (externally individuated concepts, presumably; we will return to this shortly), and perhaps (3) suffices to show that such beliefs are externally individuated. But why does that show that our experiences themselves, which Dretske concedes not to require concepts, are also externally individuated? But isn't this what Dretske needs to show to establish the central claim of his paper? Why isn't phenomenal internalism about experiences, not about our beliefs about experiences, internalism enough? Second, I am not quite clear about what Dretske means by "concept", and just why he moves from "belief that p involves, or requires, concept C" to "belief that p is externally individuated". What is the connection between concepts and externalistic individuation of content? Is Dretske assuming that the acquisition of concepts involve causal and other relations to external objects? Does this apply to all concepts or only to some? Dretske's argument seems to assume that this applies to all concepts. I would like to know whether this is what he believes, and if so, why. Third, It isn't clear to me how far Dretske intends to go with his qualia externalism: Are phenomenal experiences to be fully and wholly individuated externally (that is, in terms of their objects), or are there some fully internal phenomenal aspects to them that play a part in their individuation? The first alternative is prima facie too strong, and at least it isn't obvious how on its basis Dretske can make good on his promise to deliver a principled distinction between hearing F'red play the piano and seeing him play the piano. In both cases the object of the experiencing is the same; how then are the two experiences to be type-differentiated on the basis of their objects? The natural assumption is of course that there are internal,

purely phenomenal differences between the visual and auditory experiences, however the differences are to be described or whether they can be described in objective terms at all. But does Dretske's overall approach leave any room for purely internal phenomenal properties that can help individuate? There is one more question that I have on which I hope Dretske will throw some more light: I do not quite understand why he weakens his externalism about qualia by saying that if you accept content externalism, that allows you, but doesn't oblige you, to accept phenomenal externalism. I have my guesses, but I would like Dretske's reasons for this qualification, an explanation of why he doesn't say that content externalism commits one to phenomenal externalism. In speaking about the qualitative features of our sensations we often resort to relational/causal descriptions; e.g., "the smell of gasoline", "the taste of avocado", "the feel of smooth cold marble", and so on. This needn't be taken to mean, though, that the qualities referred to themselves are relational or extrinsic. We merely use relational/extrinsic descriptions to refer to, or pick out, intrinsic qualities. (The entity referred by "the husband of Xanthippe" is not something relational.) Given the subjective character of experience the use of relational descriptions may be the only way to communicate. But the practice itself is widespread and goes beyond talk of sensory qualities. What does "one kilogram" mean? Its meaning arguably is relational/causal (as Ernst Mach argued): it means something like "would balance the International Prototype Kilogram on an equal-arm balance". But the property it refers to would generally be considered an intrinsic property; mass is an intrinsic property if any physical property is. But that is consistent with its representation, and measurement, being relational and extrinsic. Pain is different from itch. To have a belief about this, or to be aware of the difference, do we really need have concepts (in the externalist sense), as Dretske seems to want to say? Or is he saying that for you to attribute this awareness or belief to me, you must express the content of my belief by the use of externally individuated concepts (which I may not need to possess)? If that is all that Dretske is claiming, I don't see that it supports externalism about qualia; it supports only externalism about descrzptzons of qualia. On the other hand, if I need to have these concepts to be aware of the difference between pains and itches, what kinds of concepts do I need? (Aren't dogs and cats aware of this difference? Do they have the required concepts?) And, finally, why must these concepts be externally individuated ones?

In his paper, Dretske has advanced a surprising and provocative thesis, and supports it with some very subtle and intriguing considerations. As my comments make clear I do not believe I have fully grasped or appreciated his arguments and therefore remain skeptical -skeptical rather than opposed. In any event, we should be thankful to him for his many thought-provoking ideas.


Comment on Dretske Paul Horwich

The conclusion of Fred Dretske's paper is strikingly counterintuitive, as he himself is the first to admit. For there is a strong inclination to hold that our sensations and feelings are internal -implying, if we are materialists, that an exact physical duplicate of a given person must be undergoing the same sensations and feelings as that person. But the point of Dretske's paper it to challenge this idea. He argues that there is, on reflection, no reason to agree with it: no reason to think that the way pains feel to S, or the way red things look to S , are facts solely about S ; no reason to think that what it's like to be a molecular duplicate of S must be the same as what it's like to be S; no reason, in other words, to assume that there is such a thing as purely internal experience. The way Dretske arrives at this conclusion is surprising and ingenious. For the usual basis of the inclination to hold that phenomenal states are internal is the special epistemological access each of us has to what we are experiencing: if I have a pain then I know I do, and vice versa; and how could this be so unless the pain is wholly inside me. But Dretske argues, on the contrary, that if phenomenal states really were internal, then this fact about them would militate against first-person awareness. Thus, according to Dretske, the idea that phenomenal states are internal, far from explaining our special

access to them, would actually limit their accessibility. So our reason for assuming that phenomenal states are internal is undermined. But why does he think that internal phenomenal states might not be perfectly accessible? He begins with the assumption that beliefs are external: that what a person believes is not constituted by facts intrinsic to that person, but depend, as Putnam, Burge and Dretske himself have shown, on facts about the person's environment, linguistic community and history. On this basis he reasons that S's belief that something looks red, tastes sweet, etc., depends upon circumstances that do not supervene on S's bodily state. There could be a molecular duplicate of S that is incapable of such beliefs because the circumstances surrounding the duplicate preclude his possession of the concepts needed to have them. Thus, if we suppose that phenomenal states are internal, then, although the duplicate undergoes the same experiences as S, he cannot come to believe that he has them -he cannot become aware of them. In short, given the external nature of belief, then phenomenal internalism precludes any guarantee of access to one's phenomenal states; so the rationale for phenomenal internalism is undercut. It is not easy to criticise a line of thought whose conclusion you agree with; and it's even harder when you find the argument ingenious and valid. Just about the only option left open in such a case is to maintain that the argument begs the question. And I'm afraid I think that it really does. Let me explain. As I said, Dretske is simply assuming an externalist view of belief. His purpose is to draw some conclusions about the nature of experience. However, one might well wonder whether this assumption is as generally correct as it needs to be in order to be able to support Dretske's argument. For it is one thing to concede that some beliefs are external (e.g. those about water) but quite another thing to agree that all beliefs are -including beliefs about one's own feelings and sensations. Although Dretske does not offer a full-blown justification of his general belief-externalism, he does briefly indicate what, for him, is the basis of its plausibility. It is the combination of two principles (a) that beliefs are individuated an terms of what they are about and (b) that aboutness is constituted by some sort of causal relation Now these principles will very plausibly yield externalism with respect to the belief that something is red, the belief that something

looks red, and even the belief that something looks to me the way that red things normally look (which is roughly what it is for something to 'LOOK red to me' in Dretske's non-doxastic, phenomenal sense). For the concept red is needed to have any of these beliefs and the concept red (let us grant) is not solely in the head. But Dretske's premises leave it open that a belief might be causally related (in the 'right' way) to some purely internal property. And in that case, even if we grant (a) and (b), it remains perfectly possible that there be internal beliefs. But what kind of purely internal property might such beliefs be about? Well, I admit that I can't provide any uncontroversial examples. However, the proponent of internal phenomenal states is likely to maintain that properties of just this sort are what we are aware of when we focus on the qualitative characters of our own experiences. According to this point of view, what it's like for me to look at something red is determined by a property of this kind. Another such property determines what it is like for me to have a toothache. These properties are to be distinguished from properties such as 'being the sort of experience typically produced by red things' or 'being the sort of experience I have when I have a pain' -which are obviously relational in structure. Rather, the properties I am talking about are the intrinsic qualities of an experience that would explain such relational facts about them. It is because my sensations of red share intrinsic feature K (say) that my current experience is of the sort I typically have when looking at red things:- K is the property that defines the sort. Moreover, it is some such intrinsic property that one has in mind when broaching the inverted spectrum question: whether what it's like for you to look at the things we both call "red" is the same as what it's like for me. I am not wondering whether the same things look red to both of us. I am asking whether the intrinsic character, K, of my own experience in the face of these objects, is also exemplified by the experiences that you have in those circumstances. Well, if we were really aware of such properties, then the predicates that stand for them would have meanings that are internal; and so my belief that I am having an experience with some such quality would also be intrinsic to me. Thus my twin would not only have experiences of exactly the same kind that I have, but, no matter how weird his environment, there could be nothing to prevent him from being aware of these experiences. This, it seems to me, is how Dretske's argument would be challenged by a proponent of phenomenal internalism. Now one can well imagine how Dretske would respond to this objection. He would deny that we can be introspectively aware of such

private qualities. He would maintain that to believe that we could be is to be a victim of confusion. But a response of this kind would leave him open to a further charge -that of begging the question. For either he does, or he does not, have up his sleve some way of relieving the intuitive pressure to recognize these private, introspectible phenomenal qualities. If he does, then the argument of his paper is superfluous, since they are the only experiential properties that are prima facie internal. But if he does not, then the argument is unpersuasive. For anyone who disagrees with the conclusion would deny its externalist assumption regarding the awareness of phenomenal states. To repeat my main critical point, it is that there is, arguably, a kind of internal, introspectible, phenomenal property that Dretske ignores and against which his argument doesn't work. If something LOOKS red to me (in the phenomenal sense) then, as he points out, I cannot be aware of this state of affairs unless I have the concept red. But, for something to LOOK red to me is for it to produce in me an experience of the same kind that is typically produced in me by red things. Therefore there is a quality that these experiences have in common. The inverted spectrum problem presupposes that I a m directly aware of this property and consists in the question of whether other people are aware of the same one when they look a t red things. Moreover, this is a simple property which I don't articulate by means of the term "red". Therefore we have been given no argument that my awareness of this property requires having the concept red, and hence no argument that such awareness would be constrained by external factors. Of course, the existence of private phenomenal qualities is extremely controversial. I can sympathise with what I suppose to be Dretske's sceptical attitude toward them. But I don't believe it's possible to remove the lure of phenomenal internalism except by rooting out the inclination to countenance these qualities; and I don't think Dretske has done that here.


Dretske on Phenomenal Externalism John Biro ". . .the king is but a man, as I am; the violet smells to him, as it does to me" (Henry V, IV. ii)

1. Just as there is more than one kind of externalism about thought, there can be more than one kind of externalism about experience. I favor a kind somewhat different from the one advocated by Dretske, and I shall later try briefly to say why. But first I want to explain why Dretske's arguments would not persuade me of the truth of phenomenal externalism if I did not already believe a version of it to be true. Then I shall briefly indicate why I believe that there is a version that is true, though with a twist that may not be agreeable to Dretske or to other champions of externalism. First, a quick word about the status of Dretske's arguments. He indicates in several places that he does not regard himself as having given a proof of phenomenal externalism, only an argument to show that it "might be true" and that "from the point of view of the conceptual externalist, there is no.. . reason to think it is not true". (p. 157) But he also says that there is "no other plausible way to make sense of the qualitative character of perceptual experience". If there really is not, then for a conceptual externalist, such as Dretske, the argument, if sound, must constitute a proof; for such a person, not going externalist about experience is not an option.

But I do not think the argument is sound. After giving a brief summary of what I take it to be, I shall voice a number of misgivings about it. I shall then indicate the direction in which a better argument might be sought, one issuing in the different sort of phenomenal externalism I favor. 2. Dretske7s phenomenal externalism holds that "just as we distinguish and identify beliefs by what they are beliefs about.. . so we must distinguish and identify experiences in terms of what they are experiences 04'. (p. 145) This entails that Twins no more have the same non-conceptual experiences (sensations, qualia) than they have the same thoughts. (For the purposes of the argument, we are assuming the truth of "conceptual externalism", externalism about thought.) While this runs counter to a strong internalist intuition, according to which experience must locally supervene, a second, even stronger intuition dictates that we resist the first. Dretske expresses this second intuition in this way: "The qualities that individuate one experience from another.. . are.. . nowhere in the person wherein resides the experience of these qualities". (p. 144) That this is so is not always as obvious as is the analogous truth about thoughts. The thought of something red is not thought to be red in the way that the experience of red sometimes has been. People even talk of red qualia, though hardly of red thoughts. This leads to the mistake of "[taking] phenomenal experience to be an awareness of internal objects that [have] the properties external things appear to have" (p. 146), a mistake, Dretske suggests, committed by some forms of the representative theory of perception.1 Hence the attraction to many of the hybrid view Dretske labels "the orthodox picture", a combination of conceptual externalism and phenomenal internalism. Here is the skeleton of what I see as Dretske's argument for phenomenal externa~ism:~ ' ~ r e t s k eputs t h e point in terms of a commitment t o an "act/object analysis of experience". B u t naive realists, such as Reid, who famously insisted on a n act/object analysis as a way of countering Berkeleyan idealism, are certainly not committed t o a view such as t h e one Dretske is criticizing. 'Central in this argument is t h e distinction between experience and what Dretske labels 'EXPERIENCE'. T h e latter is non-doxastic experience, experience t h a t does not involve coming t o believe (on t h e basis of one's perceptions or sensations) t h a t something is t h e case. T h e former, in contrast, requires t h e possession of concepts, which are conceded by almost all t o be essentially relational. Having EXPERIENCE requires only having t h e apparatus t o register, as we might say, perceptions in a conceptually neutral, "bracketed", way (deliberately or not, as in the case of painters and pre-conceptual cognizers, respectively).

1. One's "phenomenal experience" (EXPERIENCE) is independent of one's concepts and beliefs.

2. Having EXPERIENCE is not sufficient for knowing what EXPERIENCE one has (i.e., knowing what qualities one's EXPERIENCE has). 3. To know (be aware or be conscious of) what qualities one's EXPERIENCE has, one must have the concepts of the qualities in question.

4. Either one does not know the qualities of one's EXPERIENCE or one's externally grounded concepts are essentially involved. 5. Phenomenal internalism requires local supervenience which means that one's externally grounded concepts cannot be involved. It requires that Twins' EXPERIENCES have the same qualities.

6. But Twins cannot know (be aware or conscious of) the qualities their EXPERIENCES share.

7. If one cannot know an alleged quality of one's EXPERIENCE, there is no reason to think that one's EXPERIENCE has that quality. CONCLUSION:

8. There is no reason to think that the EXPERIENCES of Twins have any locally supervenient qualities. 3. I have not fully articulated the relations among these different theses, and I may well have omitted others that Dretske would regard as essential to his argument. So, I am sure I have not done justice to that argument in this list of what seem to me, nevertheless, essential elements in it, elements that alone provoke enough questions to cast doubt on the force of the argument. Here are some misgivings about the claims just listed: (1) I doubt that phenomenal and doxastic experience -EXPERIENCE and e x p e r i e n c e are as independent as Dretske claims. One I have nothing to complain of concerning this and related distinctions Dretske draws and claims have not been sufficiently attended to. (For more details, see his "Conscious Experience", in Mind 1993). I shall use the terminology of 'experience' and 'EXPERIENCE' and will note here only that one might wonder about how scrupulous in the last part of his paper Dretske himself is in observing his own distinction.

does not have to believe, as some do, that every aspect of experience is thoroughly shaped, indeed, partly constituted, by one's concepts to think that even one's EXPERIENCE, which is surely at least a part of one's experience -what's left when the doxastic dimension is stripped away or "bracketedn- is not wholly unaffected by the concepts one brings to it. Does the petunia really even LOOK the same (or at least very similar) to the gardener and the two-year old (Dretske, p. 148)? But I shall not press this point today. Even if we concede it for the sake of the argument, bigger troubles loom. (2-3) The notion of knowing one's EXPERIENCE, central in Dretske's argument, seems to me an unstable one. At least in the present context, Dretske seems to understand this to come to being aware or being conscious of the quality in question. But how, in turn, are we to understand these latter locutions? Construed as requiring more than mere acquaintance, they immediately take us over to the doxastic side, to experience, rather than EXPERIENCE. And this is, indeed, the work Dretske seems to deploy them to do. But how is this compatible with thinking of EXPERIENCE as concept-free? Experience must be conceded to be relational even by those who insist that EXPERIENCE is not; but if knowing the qualities of one's EXPERIENCE is also some sort of conceptual achievement, there is really no difference between the two. Naively, though, there does seem to be a perfectly good sense of being aware of, or being conscious of, a quality that requires only a relation of acquaintance. In that sense, we think that one surely does know the qualities one is EXPERIENCING. To be conscious of the pain of a tooth-ache is to feel it, not to think about it. To be aware of the redness of the traffic light is to see it. (Not, of course, to see it as red.) This is so even if it is possible to see it and not be aware of it for reasons of inattention and the like. The point is not that perceptual acquaintance is sufficient for awareness, but that when it is not, this is not for want of a doxastic or conceptual element. (4-6) If one reads 'one knows the qualities of one's EXPERIENCE' in the acquaintance way, the dilemma expressed in (5) is clearly a false one. If it is to be read in a way that makes for a real dilemma, we must read (2) and (3) in a non-acquaintance, doxastic, way. But we have just seen that, on the face of it, there is no compelling reason to do that. (7) Again, if 'know' here means what Dretske seems to mean by it, this is far from obvious. In fact, I see no reason for thinking that it is true. Why should what qualities one's EXPERIENCE has

depend on under what concepts one is able to classify those qualities? (And I do not mean here what labels one is able to put on them, but the discriminatory capacities Dretske rightly distinguishes from linguistic competence.) I see no reason to believe that my inability to tell which of two Burgundies has more terroir for want of that concept means that they taste exactly the same to me. The taste (I have) of one may be more earthy (though I will not think of it as such) than the taste (I have) of the other, even if I am unable to index the different tastes to anything that I (and others) normally take to be their externally grounded conceptual correlates. I may even be able to re-identify each on successive samplings as the same without recourse to such conceptual abilities, as I may be able to notice of several pieces of music that they have the same feature, one that is, even if I do not know it to be, a change of key. My lacking the latter concept does not "deafen" me to the quality of my experience, it only prevents me from classifying it as the quality it in fact is. (8) So, where does this leave us with respect to Dretske's conclusion that there is no reason to think that the EXPERIENCES of Twins have any locally supervenient qualities? He says that he finds the intuition on which the claim that there must be is based almost as powerful as the one that leads him to externalism. I, for one, am not as impressed by that internalist intuition. Be that as it may, if I am right in the misgivings I have voiced about the earlier steps in Dretske's arguments, then he has not given any reason to think that the Twins' EXPERIENCES do not have such qualities. And if they do, being objects of Twin EXPERIENCES, these qualities will be (qualitatively, as we may say) the same. As noted earlier, Dretske himself says that the petunia can LOOK the same to the child and to the gardener. Why, then, could it not LOOK the same to Twins? 4. Part of the trouble is that Dretske countenances too few relations one may have to qualities of one's experience. He recognizes one to qualities objectively in what is experienced (that is, to real properties of the objects of experience) but not phenomenologically accessible, and another to a phenomenologically accessible but also fully conceptualized experience. He then insists that when it comes to EXPERIENCE, one can be related to it only in the first way. (By definition, what one would be related to in the second way would not be EXPERIENCE.) On the other hand -and this is the alleged problem- being related to one's EXPERIENCE in the first way is, somehow, epistemically barren. For the relation to be epistemically

rich enough, Dretske seems to require something like having a belief about what one is related to. In the example of SEEING the seven fingers but "taking them to be" eight, not being conscious of what one SEES, must, it seems, come to not having the right belief about what EXPERIENCE one is having. However, to use expressions such as 'is conscious of' and 'is aware of' so as to rule out that one can be conscious of, or aware of, SEEING seven fingers just because one believes that one has seen (and therefore also believes that one has SEEN) eight strikes me as stipulative and unmotivated. Furthermore, there seems to be no reason to think that we are faced with the exhaustive and exclusive dichotomy of relations to experience that Dretske sees here. The facts seem to be quite otherwise: real experience is replete with non-conceptualized and yet phenomenally real and even phenomenologically salient qualities, qualities of which one is perfectly well aware or conscious in the ordinary sense of these terms. In what non-doxastic way does the relation one has to these qualities fall short epistemically? Or is it that nothing non-doxastic counts for Dretske as epistemic? Why should we accept that? It also seems to me that Dretske demands far too much of internal states with respect to epistemic accessibility. He requires that they be a (potential) object of attention, be reportable on the basis of introspection, and be recallable (accurately). As far as I can see, Dretske is right that EXPERIENCE does not meet these requirements. Whether an internalist must require that it do so is, however, far from clear. An important strand in Dretske's argument is his insistence that since being aware of an experience requires its being accessible in the ways just mentioned, it requires thinking that one is having the experience in question. Once we grant him this, we can agree that the thought that one is having EXPERIENCE E, which is distinguished from EXPERIENCE El by involving the property F rather than the property G, requires one to have the concept of F-ness (and, for El, of G-ness). Whatever relational character the concept has then infects E itself. The question is, why should a phenomenal internalist adopt an understanding of locutions like 'knowing E', 'being aware of E', and 'being conscious of E' that requires this kind of conceptual access? Perhaps I can put my complaint about Dretske's argument in a slightly different way. The central claim around which the argument revolves is that only externalism could explain (the fact recorded by) the intuition that whatever is internal in experience does not have the qualities by reference to which experiences are individuated. It

is as much a mistake to think that the EXPERIENCE of red is red as it is to think that the thought of something red is. But, surely, in thinking that in EXPERIENCING something red one is in an internal state that involves acquaintance with the redness of what one is experiencing, one need not be thinking that one is in an internal state that is itself red. Thinking the latter is indeed a mistake, as Dretske reminds us. But the phenomenal internalist need maintain only the former. And being acquainted with a property is no more problematic in the respect relevant here when the acquaintance is perceptual than when it is conceptual. Neither requires that there be something that has the property in the cognizer.

5. Let me finish by suggesting a direction in which to look for a different justification of a kind of externalism about experience. Rather than worrying about TWIN phenomenology, we could ask what I think is a more interesting question: to what extent is externalism inescapable in explaining TWINS' (and, therefore, anyone's) behavior, whatever their unconceptualized phenomenal experience (their EXPERIENCE) may be. And here it seems that the externalist has a different, and more powerful, card to play. In order to give such explanations, we, the explainers, must use concepts and classify our subjects' (putative) EXPERIENCE under these concepts, even if they do not (and perhaps cannot). When we say, under some psychological generalization about the interest or attractiveness of red things to small children, that the child reached for the petunia because it was red, we are relying, in both stating the generalization and in giving our particular causal explanation, on the concept of redness (as well as, of course, on the concept of a petunia). If we did not do this, we would fail to say what about the child's encounter with the object explained his actions and would thus be failing to give an explanation of those actions. This does not mean, as I have argued contra Dretske, that the child does not EXPERIENCE (an instance of) redness. But it does mean that its doing so is in and of itself explanatorily idle. His EXPERIENCE cannot explain his actions -even if it does explain his movements non-intentionally classified. Thus we may have a causal explanation of the movement of his arm towards the object, one for which mention of his EXPERIENCE suffices, without thereby having an explanation of his reaching for the object. And so for all qualities: only when conceptualized can they play a role in explanations of behavior. It is concepts of qualities, not qualities themselves, that figure in our explanations and contribute the external element present in such explanations. Since it is the explainer who uses the concepts, it is he, not the sub-

ject, who brings this external dimension to the explanation. What this means is that externalism is inevitable even in explanations of action that appeal to phenomenal experience (EXPERIENCE). But for this to be so, it is not required that the phenomenal experience itself be somehow sensitive to relational factors in the way that concepts are and in the way Dretske claims it is. So, I suspect that this kind of externalism will not satisfy him.3

3

~ t hya n k s t o Corliss Swain.


Reply to Commentators Fred Dretske

As they will doubtless notice, this is not so much a reply to my four commentators as it is a way of acknowledging their criticisms and indicating the general direction I would take if someone made me compose a more responsible answer. In these remarks I do little more than clarify my view -say where I stand, the position I would argue for, if I had more time. I thank each of my commentators for their perceptive and helpful remarks.

Paul Horwich: Horwich makes several pertinent observations before he accuses me of begging the question. He begins by noting -0rrectlythat I assume an externalist theory of belief. He notes, though, that it is one thing to assume that some beliefs are external (plausible enough), quite another to assume that all are (not so plausible). I agree. For the sake of argument, I was assuming that all beliefs were external. My point was that two individuals share qualia they are aware of (in the only sense it makes sense to speak of "being aware" of qualia) only to the extent that they share concepts for those qualia. Phenomenal Externalism is supported (i.e., arguments against it neutralized) to the degree in which one is a conceptual externalist.

Horwich goes on to say that even if we grant the idea that beliefs are individuated in terms of what they are about, and aboutness is constituted by some sort of causal relation, how can we rule out the possibility that some beliefs are causally related (in the right way) to purely internal properties? Maybe there are intrinsic qualities of experience that we are aware of and which define what it is like to have experiences of that sort. When we have beliefs about what our experiences are like we have beliefs about these intrinsic, these phenomenal, qualities. Everything depends, of course, on what one means by being "aware of" the intrinsic qualities of experience. I argue that we are aware of an apple's intrinsic properties (its color, for instance) merely by seeing (being aware of) the apple under normal conditions. We do not need concepts for the properties of the apple to be aware of these properties. I deny, though, that we are aware of (have experiences of) our experiences in the way we experience external objects. We do not experience experiences the way we experience apples. Hence, we cannot be aware of the intrinsic qualities of our experiences in the way we are of the qualities of the external objects we experience. The only sense in which we can become aware of the qualities of our experience (= the internal states that make us aware of such things as apples) is by becoming aware that our experiences have such-andsuch qualities, and awareness-that is a form of awareness requiring concepts. You can be aware of red without the concept RED, but you cannot be aware that something is red, that something LOOKS red, or that you are having an experience of red, without such a concept. Thus, there is no way you can become aware of qualia -properties of your experience- without concepts for those qualia. I admit, though, that I ran my argument concentrating on the external senses (vision, audition, etc.) and that the argument is easier in these cases. In these cases the properties that serve to distinguish experiences from one another are properties of the (external) things we experience. The argument doesn't work as well (it does not seem to work at all) if we consider, say, pain. John Biro: I, too, doubt that phenomenal and doxastic experiences are independent. Causally speaking, they affect each other. What you know and think (expect, fear, etc.) affects, to some degree at least, what kind of phenomenal experience you have. The opposite is also (obviously) true. I did not intend to deny this. My intention was to represent thought and experience as conceptually independent. I see no reason why a petunia must LOOK different to the experienced

gardener than it does to the two-year old. It looks different, yes, and this may affect the way it LOOKS, but I don't see why it has to. Biro introduces the relation of acquaintance, and says that in that sense (a sense he does not further specify) one knows every quality of one's EXPERIENCE. Putting aside the question of pain (see my comments on Horwich above), Biro seems to think it obvious that there is some sense of awareness (which he calls "acquaintance") in which we are aware of all the qualities of experience. On the contrary, we are aware of (acquainted with) the qualities we experience (the color of the apple, for instance), but this does not mean that we are aware of the qualities of our apple experiences (the qualities that make the experience an experience of the apple's color). The question I was trying to answer is whether one can be aware of the qualities of one's experiences (= qualia) -not the qualities these experiences are experiences of- without the appropriate concepts. Biro says that lack of a concept (e.g., change of key) does not "deafen" one to the quality of one's experience. It only prevents one from classifying it as the quality it in fact is. If I understand him, I agree with this. Not having the concept of CHANGE OF KEY does not mean one is not aware of (acquainted with, to use his word) changes of key. It only means that one is unable to identify (know, be aware that) this quality is the quality one is aware of. One is not aware that one is aware of it. That, though, is precisely what is necessary to be aware of the quality of one's experience (to be carefully distinguished from the quality one experiences). So, without the concept, one is "deaf" not to the quality one experiences (the change of key), but to the quality of the experience (that it sounds like a change of key). Finally, Biro acknowledges that it would be a mistake to think that the EXPERIENCE of red is red, but he suggests that it would also be a mistake to think that such experiences are internal states that involve acquaintance with redness. I confess to not seeing how this could be a mistake. Not unless Biro is using "acquaintance" as a very technical term. Surely, experiences of red are internal states (they go away when we close our eyes) that "involve" acquaintance with (at least an awareness of) the color red. Jaegwon Kim: Kim notes that Fred and Twin-Fred, though they do not share the concept of WATER (Twin-Fred has the concept TWATER), may share other concepts (SHINY, etc.) which gives them some common access to their EXPERIENCE of water. Though there is no aspect

of their experience (qualia) describable with the concepts WATER and TWATER (at least none that they can be aware of) they can both be aware that the puddle they see LOOKS shiny or watery in what Kim describes as a non-relational sense of "watery"). This is an important point (Horwich also made it). My argument is only that if Fred and Twin-Fred have common qualia, their access (awareness of) these qualia are restricted to those for which they share concepts. If they both have the concept of SHINY, then they can both be aware that the puddle LOOKS shiny. Their awareness of the common properties of their experience is restricted t o the properties they share conceptualizations of. In passing, though, Kim makes a related point. He says that it is an explicitly shared assumption of countless twin-earth tales that have flooded philosophy that water and twater are observationally indistinguishable -that, in a word, wateriness = twateriness. As far as I can see, this is (given the way I was expressing myself) the assumption that water and twater LOOK the same to people of similar sensory equipment (i.e., Fred and Twin-Fred). This "shared assumption" is exactly the assumption I meant to challenge. It amounts to the assumption that the LOOK of water = the LOOK of twater, that our experience of, if not our beliefs about, the puddle we both see, are the same. Why should one assume this if one doesn't assume that qualia (the way things LOOK) supervene on the intrinsic character of the experiencer? Kim presses me on what I mean by "access." What I mean by access is epistemological access: knowledge, awareness-that so-andso is the case. To use my example, the person who HEARD a change of key but who lacked concepts for describing this auditory difference, (even as a difference) lacked "access" to that feature of his experience. He was aware of a change of key (he EXPERIENCED the change of key), but he was not aware that it occurred or that he was aware of it. Kim thinks there may be a kind of access -what he calls executive access (which he describes as capable of affecting behavior without necessarily being reportable)- which does not require concepts at all. Maybe we can be aware of experiences in this executive sense without being conscious that we are having them. Kim suggests that Fred may be conscious of the watery LOOK of the puddle (he walks around it rather than stepping into it) without being aware that it LOOKS watery. I do not know why Kim describes this as consciousness of the watery LOOK of the puddle. It sounds to me like an (executive) awareness of the puddle's wateriness. I do not deny that there are features of the world awareness of which can affect our performance but which we have no concept

for. What I deny is that there are features of our experience of the world awareness of which can affect our performance which we have no concepts for. Finally, Kim asks why I have such a weak conclusion -that conceptual externalism allows (but does not compel) one to accept phenomenal externalism. The reason was that I took myself to be removing reasons to be a phenomenal externalist. Even if I removed all such reasons, that would not show that phenomenal externalism was true. It would merely show that there was no longer any reason to think it false. Wilebaldo Lara: [I did not see a written copy of Lara's comments. Nor -thanks to the postal system- did I receive the written comments in time to prepare this reply. So my brief response is from a (probably faulty) memory of the point he was making at the Canclin conference.] Lara suggests that if we take qualia to be externally grounded, then it is hard (impossible?) to see how the qualities of experiences can affect behavior. How could the fact that something LOOKS red to me affect the way I react to it (by, for instance, believing that it is red or sorting it with other red things) if its LOOKING red is constituted by relations to an external world. This is a good question --one I tried to answer (in the case of externally grounded belief) in Explaining Behavior (Dretske 1988). I don't see why the answer I gave there won't work, mutatis mutandis, for experiences.


Is the External World Invisible? Mark Johnston

Why is it such a good thing to see, to hear, to smell, taste and touch? What if anything makes perception intrinsically valuable, as opposed t o just a useful means for getting around our environment? An important part of what seems to make perception intrinsically valuable is that as well as providing us with propositional knowledge about just which properties objects have, perception also acquaints us with the nature of the properties had by those objects. It reveals or purports to reveal what those properties are like. As well as providing a map of our environment which helps us steer our way through the world, perception also seems to do something that no ordinary map could do, acquaint us with or reveal the natures of the perceptible properties of the things mapped. Explaining as I go what I mean by acquaintance with or revelation of the nature of a property, I shall suggest that acquaintance with perceptible properties is something we non-derivatively want from perception, and show that perception seems to provide just such acquaintance. Unfortunately, upon examination, perception's promise to acquaint us with properties of external things seems to be utterly fraudulent.

Just to fix ideas, let us take some examples. I am acquainted with a certain property of digitally reproduced sound (the sound of CDs), I know or a m acquainted with this characteristic property of digitally reproduced sound thanks to having listened to enough of the stuff. Knowing or being acquainted with this property, sometimes called "digital brightness", enables me to recognize digital sound and come out with comparative statements about analog and digital sound, e.g. Don't throw away your LPs, analog is much more pleasing than digital. But what I came to know when I came to know or be acquainted with the brightness of digital sound is not exhausted by my recognitional capacity nor by my capacity to come out with various statements about digital sound. I could have both those capacities without knowing the brightness of digital sound. I could instead be recognizing and comparing on the basis of knowing another characteristic property of digital sound, a certain sucked-out quality in the reproduction of the sound of stringed instruments. By listening, I know what these sound qualities are like. There is a certain way, perhaps impossible to fully characterize linguistically, that hearing represents these sound qualities as being. I not only hear that the Deutsche Grammaphon CD of Herbert Von Karajan conducting the St. Matthew's Passion has these sound qualities. I hear what they are like and so a m acquainted with them. Equivalently, the natures of these two sound qualities are revealed to me. Or so it seems. For another example, take what one knows when one knows what it is like to feel pain. One of the things one knows or is acquainted with is a property or feature of a sensation -what some call the painfulness of pain- the feature that makes pain awful. I claim that perception in general and vision in particular purports to acquaint us with some properties of external objects in the way that sensation acquaints us with features of sensations. But perception cannot carry through on this promise. To locate the notion of acquaintance among more familiar epistemic concerns, consider two familiar philosophical cartoons by which the traditional skeptical problem of the external world is typically presented, the case of the eternal movie buff and the case of the brain in the vat. The eternal movie buff has spent all his life in a dark room watching images on a screen. Never having left his video room he has no idea whether the images correspond to anything outside his room. The brain in the vat is fed a full sensorium by fiendishly clever form of neural stimulation. A computer coordinates the pattern of stimulation so that the brain has a complete and consistent sensory illusion, say as of living an ordinary life in Boise, Idaho.

17. IS

THE

EXTERNAL WORLDINVISIBLE? 187

These bizarre predicaments are typically employed to highlight a skeptical worry about our own predicament. The eternal movie buff cannot be justified in holding any visually generated beliefs about the external world, restricted as he is t o mere images which he cannot check against external reality. He can only check experience against experience. But this, so it is said, is also our predicament. We too can only check our experiences against other experiences. It is no more possible for us to attempt to match our experience against external reality as it is in itself, as it is independently of how it is experienced by us. The case of the brain in the vat is supposed to deflate the natural response that we have an epistemic advantage over the eternal movie buff by having a number of potential windows on the world which we can use to triangulate to an external reality as it is in itself. The triangulations of the envatted brain lead it to false beliefs about a life lived in Boise, Idaho. Now there is no sign in our experience which differentiates our condition from the condition of the brain in the vat. So we possess nothing to rule out the alternative hypothesis that we are brains in a vat. So we are not justified in believing that we are not brains in vats and hence are not justified in believing the things we standardly do believe about the external world. Or so the argument goes. Whatever the force of these cartoons in presenting the traditional problem of the justifiability of our beliefs about the external world, and even if their force is undermined by noting that a spontaneous and utterly natural belief is justified or a t least does not need justification in the absence of a good case against it, the cartoons also serve to illustrate a deeper epistemic anxiety about our own condition, one overlooked by fixating on the problem of justification. This other problem of the external world is the problem of acquaintance, the problem of how, given the nature of information transmission, we could be acquainted with the nature of any of the properties of external things represented by our experience. The nature of any signal received is partly a product of the thing sending the signal and partly a product of the signal receiver. It seems that we cannot separate out the contribution to our experience of our own sensibility from the contribution to our experience of the objects sensed. The case of the brain in the vat shows that our experience does not discriminate between many different kinds of external features so long as their effects on our sensibility are isomorphic in certain ways. Therefore, despite the seductive offer that perception makes, we cannot take our perceptual experiences t o reveal the natures of external things. For no perceptual experience could at the same time reveal two things so intrinsically unalike as

life in Boise and the inner workings of the vat computer. Conclusion: perceptual experience does not reveal the nature of its causes. In other words, it does not acquaint us with the external features causally responsible for our experience but only with their effects in us. We can of course refer to the external features as the features that are standardly causally responsible for our experiences, thereby making them objects of thought and reference -but a t best we know these features by description- we know them as the features that are standardly causally responsible for our experiences, whatever those features might be like in themselves. We could put the conclusion this way: relative to the problem of acquaintance, even if we are not brains in vats, things are as bad as they would be if we were brains in vats.l Both the cartoon of the eternal movie buff and the cartoon of the brain in a vat highlight the problem of acquaintance by inviting us to think of perceptual experience as simply an effect of external causes whose natures are in no way revealed by the experiences they cause. Perceptual experience in no way acquaints the brain or the buff with the nature of the external causes of that experience. In this respect, perceptual experience is unsatisfyingly like morse code reception; both involve interpretable effects at the end of an information-bearing process or signal. But the intrinsic natures of the originators of the signal are not manifest in the signal. This is a depressing comparison. Perception represents itself as (or is at least spontaneously taken by its possessors as) a mode of access to the perceptible natures of things; a mode of acquaintance with their perceptible properties. When I see the sun setting against the magenta expanse of the sky, I seem to have something about the nature of the sky and the sun revealed to me. I seem not to be merely under their causal influence in a way that leaves completely open what their natures might be like. Just as I take myself as knowing the painfulness of pain as it is in itself, and digital brightness as it is in itself, perception encourages the thought that I know sky blue as it is in itself. Russell expresses just this thought in The 'In Reason, Truth and I f i s t o q (Cambridge, 1983), Ch. 2, Hilary Putnarn claims t h a t if we were brains in vats then we couldn't mean t h e standard thing by "WE A R E BRAINS I N VATS" so t h a t we could not formulate t o ourselves t h e traditional problem of t h e external world. Notice t h a t even if this were so, it would not in any way deal with t h e deeper epistemic anxiety associated with t h e problem of acquaintance. Indeed, Putnam's modelktheoretic picture of t h e determination of meaning just capitulates on t h e problem of acquaintance. I see little interest in a proposed solution t o t h e traditional problem if t h e proposal implies t h a t t h e problem of acquaintance cannot be solved.

17. 1.5 T H E EXTERNAL WORLDINVISIBLE? 189

Problems of Philosophy: "the particular shade of colour that I am seeing.. . may have many things to be said about it.. . But such statements, though they make me know truths about the colour, do not make me know the colour itself better than I did before: so far as concerns knowledge of the colour itself, as opposed to knowledge of truths about it, I know the colour perfectly and completely when I see it and no further knowledge of it itself is even theoretically possible" .2 The acquaintance with external features which vision seems to provide is something we have reason to value. I think that it is because vision seems to acquaint us with visible properties of external things and thereby acquaint us with the natures of visible things that we take a certain kind of epistemic pleasure in seeing. My pleasure in seeing color is not simply the pleasure of undergoing certain sensory experiences, it is also the pleasure of having access by sight to the natures of the colors and hence access to part of the nature of colored things. The deeper problem of the external world, as it applies to visibilia, is that this characteristic pleasure of seeing seems inevitably to be a pleasure founded in a false belief, a pleasure which philosophical reflection would have me see through. Once my eyes were covered with bandages for five days. Part of what I longed for in longing to see again was not simply more information by which to negotiate my environment, nor simply more visual sensations. I longed for the cognitive contact with external features which vision seems to provide. It is depressing to conclude that what I longed for -acquaintance with visible properties- can never be had, even with the bandages off. I suspect that such longing for cognitive contact is an important aspect of the longing to know. Consider the metaphors of knowledge: apart from the metaphors of light and sight, they are metaphors of touch and digestion; we grasp things, digest facts, absorb points, assimilate information. (Digestion is of course just touch gone to extremes.) The phenomenology of touch is the phenomenology of getting at the shape and texture of the thing touched and not at all the phenomenology as of having a tactile experience produced at the end of a (very short) causal chain which began with the thing touched. Touch represents itself as providing acquaintance with tactile qualities. For whatever psychological reasons, we place an important value on acquaintance, i.e. not just on knowing that external things have certain properties but also on knowing those properties. Indeed it 'See The Problems of Philosophy (London, 1912), p. 47.

is hard to make sense of having genuine acquaintance with things without being acquainted to some extent with their natures and so with some of their properties. An important part of what seems so bad about the predicament of the brain in the vat, and an important part of what seems eerie about the condition of the inhabitants of IVozick's experience machine, is that there is no acquaintance with sensible properties of external objects. Though the subject's experiences are internally coherent they are not revelations of the nature of the computer processes responsible for those experiences. That is why I believe that if we cannot solve the problem of acquaintance, then the thing to say is that even if we are not brains in vats, we are nonetheless as badly off as brains in vats are. If that is right, knowledge or justified propositional belief about the external world may remain valuable instrumentally, as input for strategies for getting by, but the core of what made knowledge seem valuable in itself has been rotted out. Knowledge of the external world is left schematic and bloodless, if it includes no acquaintance with properties and hence no acquaintance with the natures of things and hence no real acquaintance with things. Well, just why is it that a veridical perception cannot fully reveal the nature of perceptible features of external things? To run through the specific version of the argument in the case of vision, begin with the obvious causal condition on seeing a feature or property of a thing. If a property of a thing is literally seen then the thing's having that property must be part of the causal explanation of the visual experience which counts as seeing that the thing has that property. If I see the greyness of my dog's coat, as opposed to having a (perhaps accidentally veridical) hallucination of the greyness of the coat, then it will be the case that the visual experience which represents my dog's coat as grey is partly caused by my dog's coat being grey. But if my visual experience is to be a revelation of the nature of the greyness of my dog's coat, then that visual experience must correctly represent how that greyness is. Now the way my visual experience represents greyness as being is fixed by the internal states that I a m in at the time of the experience.3 By the causal condition on seeing, those internal states are to be 3 ~ h best e motivation for this local supervenience claim is given by Ned Block's Inverted Earth example. The example simply mobilizes intuitions. The argument for local supervenience of the higher-order "depictions" of the natures of the colors is t h a t the alternative, namely that the content of these depictive representations of the colors of external objects are fixed by the natures of those colors themselves, makes color misperception much harder than it is.

17. IS T H E EXTERNAL WORLDINVISIBLE? 191

causally explained by my dog's being grey. This means that there is a causal process connecting the state of my dog's coat and my internal states. Barring a pre-established harmony no such causal process will preserve and transmit information so as to secure a naturerevealing match between how some feature of the cause, say the greyness of my dog's coat, is and the way I am caused to represent that feature as being. To see involves having the natures of visible properties revealed by a causal process, but this is just what no causal process actually does. The originally unbelievable conclusion now follows: we cannot see color, because our visual experiences as of the colors of things do not reveal to us what the colors of the external causes of our experience are like. But if we do not see color, we do not see color difference, and if we do not see color difference, we see neither edges nor colored areas, and if we see neither edges nor colored areas, we do not see surfaces, and if we do not see surfaces, we do not see anything in the material world.4 Our visual experience is then just a "false imaginary glare", simply an arbitrary medium in which the material world is mapped for the purposes of intentional action. The characteristic pleasure of seeing, the pleasure of having the nature of visible properties and visible things revealed to us, is a false pleasure. The promise of vision now appears totally fraudulent. What is driving the general argument about perception and the specific argument about color seems to be a certain standard of acquaintance with or revelation of the natures of perceptible properties, in effect the standard Russell arrived at by taking visual perception on its own terms. The obvious response is to suggest that the standard of acquaintance naively derived from the case of seeing color, the standard that Russell described, is impossibly high. It is the standard of complete knowledge of a property as it is in itself, a standard not satisfiable even in the case of the colors. Perhaps we can be acquainted with properties by a less demanding but nonetheless significant standard? 41n arguing this I a m not assuming t h a t we see edges or areas or surfaces by seeing color difference. I assume only t h a t certain necessary conditions on seeing these things hold. If we could see color and hence color difference and hence shape, then it might be right t o count as seeing making out shapes with nightvision goggles which misrepresent all t h e colors in t h e scene before t h e eyes. B u t if we never see colors, then this does not count as seeing either. All t h a t is going on is t h e systematic representation in a n arbitrary medium of information in t h e scene before t h e eyes. This is as good as what we call seeing, but neither is seeing in t h e sense of visual acquaintance with t h e scene before t h e eyes.

The notion of acquaintance with a property or better with a family of properties -the shapes, the colors, the textures, etc.- is obscure because acquaintance with properties is not equivalent to knowledge of a statable set of linguistic descriptions of those properties. However, although acquaintance is not exhausted by such linguistic descriptions it does not directly issue in such propositional knowledge. Let us say thak Fl through Fn constitutes a family of properties when there are essential similarities and differences among Fl through F,, similarities and differences that hold in any possible situation. Then if you know or are acquainted with the nature of properties Fl . . . Fn by means of the operation of some faculty K, then (i) you can recognize other instances of Fl . . . Fn just by relying on K, and (ii) you can know a family of similarity and difference relations ("unity principles") holding among Fl through Fn and know these just by relying upon the operation of K . So it seems that I am acquainted with the ordinary Euclidean shape properties on the basis of perception. Having seen pyramids, cubes, spheres, cones, etc. (i) I can recognize other instances of the same shapes, and (ii) I know a family of shape similarities and differences -cubes are not conical- cones, pyramids, and cubes are alike and together unlike spheres in having apexes. Crucial to this operational characterization of acquaintance with properties is that such acquaintance issues in knowledge of comparative similarities among families of properties, e.g. that the property Fl is more similar to the property F2 than either is to the property F3. Again, the similarities knowledge of which is generated by acquaintance are "essential" similarities. They are the similarities which follow from the natures of the properties in the family and so hold among the properties in every possible situation, i.e. no matter how the contingent relations among properties are varied and no matter how the facts about which things have properties are ~ a r i e d . ~ Obviously, on this criterion, acquaintance turns out to be a matter of degree, depending on the range and detail of the knowledge of property similarities it grounds. It may now seem to follow that we are indeed acquainted with the colors, as well as the shapes and the textures. Take the colors. It is often said that just on the basis of visual experience (and maybe a bit of visual memory) we know a host of propositions about the similarity and difference relations among the colors. For example, we know 5 ~ similarities o and differences in how properties look or perceptually seem to a class of perceivers under this or that condition will not count as essential similarities.

17. IS

THE

EXTERNAL WORLDINVISIBLE? 193

that canary yellow is not a shade of blue. And that is to know that the blue shades are more similar (in hue) among themselves than any are to canary yellow. Indeed, for each of the phenomenally evident relations among the shades and the hues which we express by means of the predicate "x is not a shade of y" we know a proposition about comparative essential similarity, a proposition of the form "The y's are more similar in hue among themselves than any is to x". Since the phenomenally evident relations among shades are available in visual perception, visual perception acquaints us with the colors. Or does it? The reason for hesitating is that it is difficult to find a conception of the colors as features of external objects on which the similarities we know to hold just by seeing are indeed similarities among properties of external objects, rather than just similarities in respect of how those properties st,rike us as being. Once we take colors to be categorical properties of external things all we may really be able to know about canary yellow and the shades of blue may be that the blues happen to look more similar among themselves than any of them look to canary yellow. This is not an essential similarity among shades but only a similarity among how the shades happen to appear to us. If that is all we can know by perceiving the colors, then even by more reasonable standards of acquaintance we turn out not to be acquainted with colors. To see how the conception of colors as categorical features is at odds with acquaintance, consider the Primary Quality Account of the colors, the view that the colors are the non4ispositional properties of external objects standardly responsible for the appearances as of the respective colors of external object^.^ So the shade canary yellow is the non-dispositional property which standardly explains the appearances as of canary yellow things. Mutatis mutandis for the various shades of blue. Suppose that color science ends up discovering this: the property which standardly explains the canary yellow appearances and the various properties which standardly explain the various appearances as of the shades of blue are not, when taken together, as similar among themselves as the properties which standardly explain the various appearances of the shades of blue are similar among themselves. (Remember, it is "essential similarity" which is at issue.) On the Primary Quality Account so far adumbrated this would be the discovery that canary yellow is not a shade of blue, i.e., not to be counted a blue. 'This is the view of the colors we find in Saul Kripke's Naming and Necessity (Harvard, 1980), fn. 66.

But is it really a matter of scientific discovery that canary yellow is not a shade of blue? No. Such similarity and difference principles surely have a different status. We take ourselves to know those principles just on the basis of relevant visual experience and ordinary grasp of color language. No one had to wait until the end of the second millennium A.D. to find out whether or not canary yellow is a shade of blue. Any account which implies this along with corresponding things about the other similarity and difference principles concerning the colors thereby implies that perception does not acquaint us with the colors even by our less demanding criterion of acquaintance. That, of course, is just a first move against the Primary Quality Account. The friend of that account should be allowed to answer that indeed it is not a matter of scientific discovery that canary yellow is not a shade of blue. Rather, he might say, such a principle along with other unity principles must be held true as a condition on any family of properties deserving the color names. So the principle that canary yellow is not a shade of blue turns out to be relatively a priori after all. More exactly, what is a priori is a conditional: if any property is to deserve the name of canary yellowness, then it cannot be a property which is as similar to the blues as they are among themselves. On the revised account, some property turns out to count as canary yellow only if a complex condition on that property along with a host of other properties is discovered to hold. The candidate properties to be the blues have to show a natural or genuine similarity among themselves, a similarity which they do not share with the candidate to be canary yellow. Suppose color science discovers this condition holds along with the other unity conditions which the Primary Quality theorist regards as central. Then some (complex, disjunctive) physical properties turn out to be canary yellowness, teal, turquoise, sky blue, and so on. But then what gives one the right to say that there is a property canary yellow and it is not a shade of blue is not simply visual perception but also and crucially recherche facts from color science. That is to say that on the Primary Quality Account perception still does not acquaint one with canary yellow, nor with the other shades, even by our less demanding criterion of acquaintance. Color here is just an example; the same result will follow for any quality conceived of as a primary, i.e. as the microphysical causal basis of a class of perceptual experiences. Physics may tell us that the microphysical causal basis of appearances of Euclidean shape do not themselves have a Euclidean shape and may be features whose adequate description

17. IS THE EXTERNAL WORLDINVISIBLE? 195

involves characterizing things extended in nine or ten dimensions. There is no guarantee derivable from vision that there will be essential similarities and differences among nine dimensional properties which mimic the essential similarities and differences that Euclidean shapes present to vision. Weakening the criterion of acquaintance far enough so that it could turn out that vision acquaints us with the colors as the Primary Quality Account represents them would necessarily involve allowing that one could be acquainted with a property even if the property does not exist, so long as one knows that if the property exists and certain other properties exist, then they are similar and different in certain ways. But to suppose that I am acquainted with a property even when it might turn out that there is no such property breaks the intuitive link between acquaintance and the cognitive contact which makes knowledge non-instrumentally worthwhile. I a m not here following Russell, as he is usually interpreted, in suggesting that there is something about the subjective side of acquaintance which guarantees that acquaintance has an object; I a m only saying that something7 that could occur whether or not there was a property F does not deserve the name of acquaintance with the property F. What then of the account that has it that the colors are dispositions to appear colored, the so-called Secondary Quality Account? Again, colors are here standing in for any secondary quality, so the considerations that follow are in fact perfectly general. The colors conceived of as dispositions are genuine, albeit relational, features of external objects. Their manifestations are the various experiences in various subjects as of the various colors. These sensory manifestations are not simply the effects of the dispositions they manifest. They are or can be manifestations in a more interesting sense -they can be revelations of the nature of the dispositions in question. For about any disposition of objects to produce a given experience, it is plausible to hold that if one has an experience of the kind in question and takes that experience to be a manifestation of the experience i n question, one thereby knows the nature of the disposition in question. Of course one does not thereby know the facts concerning how in general the disposition is specifically secured or realized. But these are 'AS emerges in my response to Alan Gibbard, it would be better t o say "knowledge" rather than "something" here. T h e vaguer term allows a reading t o t h e effect that a n ezpen'ence which could occur whether or not there was a p r o p erty F could not constitute acquaintance with F. Of course, that is the very Russellian claim about acquaintance which I a m concerned to rule out.

facts concerning the disposition's contingent relations to other properties. They do not concern the nature of the dispositional property. So I take myself as having come to understand the complete nature of the property of being nauseating on the high seas one afternoon thirty years ago when I tasted a juicy apricot on a ferry crossing from Melbourne to Hobart. I had the experience of nausea and thought of it as a manifestation of a power of juicy apricots to produce nausea on the high seas. But I did not then know anything about the physical basis of juicy apricots being nauseating. Similarly, if I conceive of the blue of the sky as a disposition to produce a certain visual response in subjects like me, and I now discover myself t o be responding just so, I can be in possession of all there is to know about the essential nature of the dispositional property that is sky blue. I do not thereby know the contingent details of how sky blue might be physically realized here before my eyes or anywhere else. But that is ignorance of the contingent relations between the disposition and the other properties which happen to realize the disposition. I do seem to know everything essential about the disposition that is sky blue. Mutatis mutandis for the other colors. On the basis of visual experience as of canary yellow, teal, turquoise, etc., and conceiving of such experience as the manifestation of the dispositions that are canary yellow, teal, turquoise, etc., I can satisfy the condition for acquaintance with the colors. Just on the basis of visual experience, I can know that the manifestations of the dispositions that are teal, turquoise, sky blue, etc., are more similar than any are to the manifestation of the disposition that is canary yellow. But this means that I can know that the first family of dispositions are more similar in nature, more similar essentially than any are to the disposition canary yellow. Hence I am on the way to satisfying the condition for being visually acquainted with the colors conceived of as dispositions. Suppose for a moment that this works.8 Vision is then a faculty by which we could become acquainted with the properties which are the visual response-dispositions. It is not a faculty by which we could become acquainted with the bases of visual response-dispositions, i.e. it is not a mode of acquaintance with the properties which the Primary Quality Theorist identifies as the colors. Since it is a commonplace in the philosophy of color that we are inevitably in the business of revising our naive color concepts, we should make the revision which allows us to secure a n important cognitive value -the

or

more on this aspect of the dispositional account, see my "How To Speak of The Colors", Philosophical Studies 68, 1992, pp. 221-263.

17. IS T H E EXTERNAL WORLDINVISIBLE? 197

value of acquaintance with those salient, striking and ubiquitous features that are the colors. We should think of colors as dispositions if we are to get what we really want from seeing color. That is the hope: acquaintance with the colors is at last secured by conceiving of the colors as dispositions to produce visual responses. (We can engage in a kind of second-order seeing. We can take our visual representations as manifestations of the dispositions of external things to produce those visual representations.) I believe that the hope is severely threatened by the consequences of the following observation: response-dispositions come in pairs. That is, whenever there is disposition of some external object or objects to produce a sensory, emotional or cognitive response in a class of subjects under certain conditions, there is also the correlative or dual disposition of the subjects in question to issue the response in question upon being presented with the objects under the conditions in question. So as well as the apple's disposition to look red to me, there is my disposition to have a n experience as of the apple looking red when the apple is presented. T h e same perceptual experience is as much a manifestation of my disposition to see the apple as red as it is a manifestation of the apple's disposition to look red to me. So the same response is potentially as much a revelation of a dispositional property of mine as it is a revelation of a dispositional property of the apple. These dispositional properties may not be really distinct but may represent just two ways of characterizing, as it were from different ends, one relational complex comprising features of the apple and features of my sensibility. This odd thing, the relational complex, may be more evenhandedly described as a disposition of a certain subject, me, and of a certain object, the apple, taken together, namely a disposition to interact to produce certain experiences. This, if you like, scattered capacity partly in the object of perception and partly in the perceiving subject, though odd enough not to have any familiar name, is the characteristic kind of property with which vision in particular and perception in general can acquaint us. But perception will only acquaint us with such scattered capacities if we conceive of our perceptual experience as manifestations of such scattered capacities. That requires that we take our perceptual experiences as revealing features which are as m u c h features of ourselves as they are features of external objects. That is none too easy a thing to do. And it is entirely against the phenomenological grain of perception, against what we might call the outerdirectedness of perception as opposed to bodily sensation. To give up on this phenomenological outerdirectedness is t o leave the topic of vision entirely.

So it is still far from clear how acquaintance is possible even with dispositional features of external things and even by reasonable standards of acquaintance. If that remains unclear, the conclusion that threatens is that perceptual knowledge is not what it is cracked up to be. Without being founded in acquaintance with properties, our knowledge of the external world, though vast and probably entirely unthreatened by standard skeptical worries, is nonetheless bloodless and schematic, devoid of the very substance which seemed to make it intrinsically appealing -acquaintance with properties.


Visible Properties of Human Interest Only Allan Gibbard

In this careful and intriguingly argued paper, Mark Johnston makes far-reaching claims.' I don't ever see anything with my eyes, he tells us, or hear anything with my ears, or feel anything with my fingers. What I do with my eyes isn't seeing -and that is depressing. The issue between Johnston and common sense could be linguistic: Whether I count as "seeing" things in front of me must depend on what it means to "see". I'm certainly doing something with my eyes, and if we wanted, we could call whatever that is "seeing".2 But Johnston tells us something that is clearly not linguistic: that things are bad. I and the rest of us have hitherto been deceived: There is a crucial sense of 'see' in which I thought I was seeing, but in which I never have and never will. I've been taken in by an illusion that vision foists on me. And this I should find disappointing and depressing. We're deceived and decepcionados, defrauded and defraudados.

'

~ohnston(this Volume). 2See Johnston (1992) for fuller explanations of the status of the claims he was making.

It's an important philosophical exercise, then, to find in what senses we do see things out there and in what senses we don't. But that leaves other important questions: Have we been deceived? And have we grounds, in this, to be disappointed and depressed? Now there are many things both large and small to be depressed by in the human condition: The law's delays, the proud man's contumely, that the race is not always to the swift. At the large end we have ethnic hatreds and what they lead to; avarice and indifference to the plight of others; the failure of the international community to move toward an order that can foster peace, democracy, and human thriving; the failure of anyone to devise a system that harnesses human propensities to achieve widespread prosperity with reasonable economic security. It's depressing if one's hard and valuable work is not recognized and appreciated by those who matter to one, or if one keeps not dealing with desk clutter, or if the stresses and ego-involvements of work make things that should be fun not so much fun any more. Why, though, do I myself feel no urge to add Johnston's problem to any such list? It's not just that his is an old disappointment that I've accommodated to, like the realization of mortality that recedes into the background for many periods of a good life, and emerges in emotional crises a t times as the fact begins to seem more real. When, though, a t the age of nine or so I began to read about atoms, I found it not depressing but exhilarating -and the same goes for what I've learned since about the science of vision. Johnston not only himself finds our visual plight depressing, I take it; he regards it as a mistake for me not to find it depressing. Or I think this may be what he's suggesting -and I'll certainly be claiming, on the contrary, that it's a mistake on his part to find the main facts of vision depressing. These questions of what to find disappointing and depressing are real questions, I think; they don't just have to be rhetorical flourishes; it's not just "different strokes for different folks" in these matters. As human beings we ponder, separately and together, what norms to accept as governing our feelings towards aspects of life. And Johnston, I claim, when his eyes were bandaged for days, had every reason to welcome the bandages' removal so that he could see again. What is there, then, to be depressed or distressed about in the facts of vision -and of the other sensory modalities, for which the same kind of story obtains? When I see a canary, Johnston tells us, I'm not acquainted with it's specific kind of yellow. I don't see what this property is like; the canary's yellow isn't the way my experience presents it as being; its intrinsic nature isn't manifest in my experience. My experience doesn't "reveal the nature of its

causes". I don't have "access by sight to the natures of the colors and hence access to part of the nature of" the canary. My knowledge of the canary doesn't include the vivid acquaintance it seems to; rather what real knowledge of it I have is schematic and bloodless. Illusions, to be sure, can be well worth having. One of the thrills of the movie Jurassic Park is the illusion of being in the midst of a herd of gracile running dinosaurs. One of the big thrills of the Sistine Chapel is the illusion of seeing God the Father giving life to man. Still, I agree that we'd like not everything to be this kind of make-believe seeing or worse. Is it? Proving Johnston's conclusions might seem a tall order, and we'll have to examine carefully how he means to force us to them, kicking and screaming. I won't scrutinize his initial moves, for with an eye to how we might resist them, Johnston himself devises a response for us t o try: that these moves may depend on an unreasonably high standard for "acquaintance". It is when he responds to this that I want to enter the discussion. Johnston considers two current theories of what colors might be and the sense in which we can be acquainted with them. I'll take up only one of these kinds of theories, the one that I myself find promising.3 The surface color of an object, let's try saying, is a matter of the ways it reflects light at various ~ a v e l e n ~ t hThere's s.~ more to be said t o make this a little precise, but rest with this for now. What, then, is special about color, as opposed to features like shape and mass? It's this: that the interest of these properties is parochial to our species. Reflectance of electromagnetic radiation of various tiny wavelengths is interesting, to be sure, but the particular aspects of this reflectance that constitute specific colors are of no special interest apart from the peculiar construction of the human visual apparatus.5 Moreover, 3 ~ h view e I shall defend here draws much from Kripke, though I won't try to identify where my view matches his and where it might deviate. See Kripke (1980, 140, note 71). Kripke, Johnston, and C.L. Hardin all spoke on color at the University of Michigan Philosophy Department Spring Colloquium, 1989. I arrived on that occasion at some aspects of the view I defend in this commentary, but I now have little track of which aspects of my views were held by speakers there. 4Surface color is to be distinguished from apparent color of luminous objects, and from color reflected from the interior of an object. I thank Johnston for vividly stressing these distinctions at the 1989 Michigan Spring Colloquium. 5~'11speak of "specific" or "exact" colors such as canary yellow, in contrast to such generic colors as red, yellow, green, and blue. A specific color is characterized by three dimensions: brightness, hue, and saturation. English, according to my dictionary, has no term for this, since "shade" indicates degree of lightness or darkness, whereas these have no bearing on "hue". A convenient alternative would be to appropriate the term "shade" t o mean an exact color.

there's a degree of vagueness as to what particular property, say, being yellow is, or being the same exact kind of yellow is. The human visual apparatus is adapted to discriminate objects by means of their reflectances, classifying them in a structured "space" that locates these particular properties. It might have responded to other properties and other kinds of similarity instead -but it doesn't. And it doesn't even make quite the same classifications with all people, or under all normal conditions: Human color agreement and color constancy are remarkable, but far from perfect. This rough color constancy and rough agreement gives a rough veridicality to many of our color judgments. In what sense is this structured array of color properties of parochial interest? Well, suppose there are intelligent beings Centauris, let's call them- on a planet of the star Alpha Centauri. They'll find being spherical important, but being yellow won't be anything they experience. They'll see, perhaps, if we ourselves do, but in seeing they'll discern different aspects of reflectance from the ones that our species peculiarly discerns. As for us, we have a visual concept of a color property -the property of being canary yellow, for instance. Laboriously we might construct a physical concept quite different from this everyday visual concept, and this would be another concept of this same property. Centauris might, if they cared, share our physical concept or something much like it, but not our visual concept. And this physical concept is not one they would bother with -unless they were studying us. It would be one of may arbitrarily constructed physical concepts which they would find pointless. Back, though, to us: Why doesn't my visual response to a canary in plain sight in front of me constitute "acquaintance" with its special color? What could acquaintance be, if not this? I'm having the visual experience I'm adapted to having in response to this exact color. My experience represents these properties, we can say, in the only coherent sense in which an evolved mind can represent its surroundings. Where's the illusion? Issues of philosophical method lie buried in any such retort: My talk of adaptation and representation appeals to recently discovered rnatters of science that no one would claim are manifest in visual experience. I certainly don't want to say that science has no surprises and can never prove us wrong in our everyday beliefs. Any account I offer must allow for what is surprising about the scientific facts of human vision. Given, though, the kind of scientific knowledge that disturbs Johnston, isn't our problem one of interpretation? We now want to know how to place our nai've color concepts in scientific

thought -how, trusting science, to regard them as applying or not to aspects of the world. To assess our nai've concepts, we need to look at them from the standpoint of our scientifically disciplined and informed selves, and ask what properties, if any, those nai've concepts are best interpreted as being concepts of. Now for canary yellow, I would think, the answer would be the property we respond to in conditions we take to let us "see things as they are". Or better yet, we should get to color concepts by their relation to our nai've notion of "seeing things as they are" and the structure of colors that our visual experience presents to us. We should interpret these notions charitably, so long as there is something in nature that roughly fits them. All this of course is sketchy. I'd expect some sort of teleosemantic notion to fill the bill, but I won't try to work it out. We can think of Johnston, the point is, as arguing that charity must be stretched too far, in the case of vision, for any such scheme to be plausible.6 We can't, from a scientific standpoint, self-interpret our longings for acquaintance in any way that offers us reasonable comfort that these longings are often fulfilled. And others agree with Johnston. What, then, stymies all claims that vision acquaints us with aspects of surface reflectance? Johnston, as I say, starts with standards for "acquaintance" that are impossibly demanding, but then he asks whether we can lower these standards. Not enough, he answers: Even the bare minimal standards for any claim of acquaintance with colors can't be met. What stymies meeting them? Johnston's argument before he gets to the kind of view I'm defending is this: Acquaintance, to be acquaintance, must give knowledge of essential similarities: if we're really seeing, we see what's really like what. But with colors, we see not "similarities among properties of external objects", but merely "similarities among sensational features of the visual experiences which we have of properties of things" (p. 193). This kind of similarity is not an essential similarity among surface properties of objects (ibidem). In other words, the similarities we seem to find among, say, specific kinds of blue might mean nothing to a Centauri, even if she had a full understanding of the objective, physical nature of the surfaces in question. And if the similarities among colors that we seem to find aren't real similarities, then what we have isn't acquaintance with properties that are there -and so we aren't really seeing colors things have. ' ~ o h n s t o n (1992) contains sensitive discussions of how t o cope with t h e vagueness of the analytic-synthetic distinction.

Johnston extends this argument, because he wants to cover the possibility that the similarities vision seems to present us do match "essential" similarities of objective properties. But I'll take the simpler argument. After all, it's probably true that phenomenal similarities of colors don't match "essential" similarities. "Essential" similarities, I take it; are ones that our physicists and Centauri physicists can agree upon.7 These might be given in terms of some such mathematical function as the integral of the square of the difference in the reflectance functions of two surfaces: the smaller the integral, the more similar the reflectance functions.' By this standard, I take it, as things actually are, a red and a green surface may be more similar to each other in their reflectance functions than two blue surfaces. I'd guess we're adapted to be highly sensitive to differences between the chlorophyll of leaves and grass and anything tending in its reflectance more toward the longer visible wavelengths. That helps us to discriminate well in verdant surroundings; it's a difference that matters a lot on Earth, and might matter very little for Centauris on their planet. So if Johnston's argument works for an actual mismatch between apparent similarities and "essential" ones, then it probably works for colors and human vision. But does it? I must say first that I don't find myself even mildly annoyed or disappointed, much less depressed, on speculating that the biggest apparent differences in color are a matter of small physical deviations from leaf color. The news is surprising, but I take delight in this; I don't feel that I've been defrauded. I don't feel that I've been walking around in the iron grip of a vast illusion. Should I? I do understand a longing to see how things are: I love seeing what Maya ruins are like, and wish I could see what they were like when they were in full use. Johnston's own longing, on the other hand, strikes me as arcane once its content is cashed out. The content of his avowed longing is this: that the apparent similarities between colors be "essential" similarities. "Essential" similarities, as I read 7"Essential" properties are often defined as properties a thing would have in any possible situation in which it existed. I take it we should not read Johnston as meaning this. Later I discuss a metric of "color similarity" t h a t is "essential" in t h e "possible situations" sense, but not in t h e sense of being of interest apart from t h e peculiarities of human vision. Johnston, as I read him, would not count similarity according t o this metric as "essential" similarity. 'By a "reflectance function" of a surface, I mean t h e function from wavelength t o t h e proportion of light t h e surface reflects a t t h a t wavelength. This may just be called "reflectance" (see Johnston, 1992, 260 note 8), but I mean t o stress t h a t I intend a function of wavelength, and not simply t h e proportion of total light the surface reflects.

him, are ones that are explanatorily important independently of how colors look to us, and so would seem important to Centauris. Now to be sure, I'll happily accept that the real nature of what one longs for might be a recherchC matter, as when a child longs for a piece of ice in his mouth without any understanding of its chemical and lattice structure. We have to ask, though, whether Johnston has identified the real nature of "seeing what something is like" -the real nature it would have to have if there were such a thing. I think that he has not. Different kinds of blue are close to each other in color, whereas any kind of red is quite unlike any kind of green --or so vision presents the case as being. But similarity, we all know, is a matter of respect. Are color similarities presented to us as "essential" similarities? I don't see that: Blues are presented to us as being close to each other in color. I wouldn't have thought there was any way to translate out of this color-similarity language into a language of physics or metaphysics. Color similarities are presented to us as real similarities, let me agree, but as real similarities in color. Now we can construct a distance metric among reflectance functions that pretty well matches differences in color. What would constitute an exact match is a vague matter, since things look to be the same exact color in one normal kind of light look a little different in another kind. But we can get various metrics that more or less match our somewhat varying judgments of identity and similarity of colors. The philosopher's dismissive term "disjunctive" won't illuminate well what is going on with such a metric, since the metric will have a rationale best expressed with integrals and such-like mathematical apparatus, a rationale based on our visual responses and their bases -for instance, the functions that give the sensitivities of the three different kinds of cones in the retina to different wavelengths of light. This metric, I agree, will be arbitrary in terms of any kind of rationale that smart Centauri scientists could appreciate before encountering humans. But the distance metric itself will be a mathematical function defined in a way a Centauri physicist could understand. She wouldn't get the point without hearing of earthlings, but she'd follow the definition. This is a physical metric of no purely physical interest. Similarities in color, I'm saying in short, are physical similarities in one sense -they can be defined in physical terms. This involves some resolution of vagueness, but precise color is a somewhat vague matter anyway. They aren't physical similarities in another sense: they aren't of interest to a physics that excludes all terrestrial " s t a m p collecting".

Note that on such an account, similarity in color is tied not just to the properties that constitute colors. We couldn't just tell Centauris what properties constitute being various specific colors, and then explain that "similarity in color" is similarity in those properties. Given those definitions of properties, they would discern the same topology as we, I presume, but they might well not discern the same metric. Not only are the properties that constitute colors of no interest apart from the specifically human visual system; even given these properties, the similarity relation that constitutes similarity in color is of no interest apart from human vision. Vagueness aside, then: There are physical properties that constitute being various specific versions of blue, yellow, and the like; these are physically uninteresting physical properties. There is a physical metric that constitutes degree of similarity in color between the various physical properties that constitute exact colors. This is a physical metric that is physically uninteresting. This similarity of color, the point is, isn't plain similarity, apart from us, among those properties that constitute exact colors. It is visual similarity. Still, this is a genuine kind of similarity: it is similarity i n a respect. In seeing things we discern this genuine kind of similarity - o r we do when we've got more or less normal light and the like, when we're seeing things, as we'd nai'vely say, in their true colors. I don't find, then, any vindication of the claim that I'm suffering from a massive illusion in my normal visual life. In seeing I discern genuine physical properties, and I discern a genuine vague metric of similarity in color among them as degree in likeness of color. Johnston, though, has another kind of argument. I won't render it exactly, because it comes at a point in the dialectic that I claim we never reach. But an argument he gives suggests the following line of attack: The view I've been giving depends on much recherch6 psycho-physics. Will it not, then, be a difficult empirical discovery that kinds of blue are more similar in color than is any kind of red to any kind of green? Johnston sees roughly the right response to this: that things like this obtain "must be held true as a condition on any family of properties deserving the color names" (p. 194). I've said that we should be charitable to ourselves, in constructing our physical interpretation of color, and this charity is meant to capture this constraint. But then, Johnston complains, what "gives us the right" to the color judgments we make "is not simply visual perception but also and crucially, recherch6 facts from color science" (p. 194). I wouldn't quite say this: We can speak and judge with right before a scientific vindication comes along. But then, I'd agree, we can ask for scientific vindication -and indeed we might not get

it. This, Johnston tells us, defeats all claims that we can see what's in front of us. For it allows "that one could be acquainted with a property even if it [might turn out that] the property does not exist" (p. 195). And "something that could occur whether or not there was a property F does not deserve the name of acquaintance with the property F" . Here I'm baffled. Why not? I'm acquainted with sound of my wife Beth's voice and the look of her smile -so ordinary thought would have it. But suppose she had had a twin Jo with a like voice and smile who substituted herself for Beth just after I first laid eyes on and noticed Beth -in a concert, let's imagine, when Beth was busy playing the bonang and so not smiling or vocalizing. In that case I would have had the same experiences I've actually had, without their being of the sound of Beth's voice or the look of Beth's smile. This possibility is fanciful but not esoteric. Perhaps it raises questions of knowledge: Do I really know that these are the voice and smile of the Beth whom I looked at during that concert? But given that in fact they are, I don't see any problem for my claim to be acquainted with this sound and this look. Perhaps, to be sure, Johnston's dictum applies in a different way to colors. But it's a dictum that in one apparent application is wrong, I've just argued. We shouldn't, then, rely on it without scrutiny: It is not only far less obvious than the fact that I see my hand; it doesn't have any initial credibility until it is argued for. Johnston's charges of fraud, then, fail to hold up, if I am right. His depression is based either on an arcane longing, or more likely, on misidentifying, in a scientific framework, the object of an ordinary longing really to see what things are like. There is nothing schematic or bloodless in my acquaintance with the specific color of a canary; I see and appreciate its vivid yellow. Much is schematic in the scientific and philosophical story I'd tell of what this consists in, but the existence of hidden natures will be of no surprise to ordinary thought. In many realms, human beings the world around assume that hidden natures explain surface appearances. Scott Atran, for instance, finds a universal human propensity to find hidden "species" natures behind the surface appearances of animals -a propensity that has even shaped scientific taxonomy.g Young children show evidence of belief in hidden natures.'' The hidden bases of colors and color vision turn out to be full of delightful surprises -that I won't deny. But for all the arguments we've been considering, the main things ' ~ t r a n(1994).

''See, for instance, Gelman, Coley, and Gottfried (1994)

we seem t o find in seeing the world can, on the right interpretation, be vindicated scientifically. Other arguments perhaps will work where these don't. Certainly I don't know how t o reconcile a n ordinary view of perception with the strangest features of quantum theory. B u t the arguments Johnston gives in this paper, intriguing and marvelously informed as they are, don't themselves succeed, if I a m right, in wresting from us our claim t o see colors -and so t o see a t all.

Atran, Scott (1994). "Core Domains versus Scientific Theories: Evidence from Systematics and Itza-Maya Folkbiology". Hirschfeld and Gelman (1994), 316-340. Gelman, Susan, Coley, J.D., and Gottfried, G.M. (1994). "Essentialist Beliefs in Children: The Acquisition of Concepts and Theories". Hirschfeld and Gelman (1994), 341-365. Johnston, Mark (1992). "How to Speak of the Colors". Philosophical Studies 68, 221-263. Johnston, Mark (this Volume). "Is the External World Invisible". Kripke, Saul (1980). Naming and Necessity (Cambridge, MA: Harvard University Press).


Getting Acquainted with Percept ion David Sosa

In his "Is the External World Invisible", Mark Johnston argues that perception is unable to acquaint us with the perceptible properties of external objects. I believe his argument is interesting and important, and his paper makes its interest and significance explicit. Here I will give some reasons for doubting its success. Johnston is concerned with the problem of acquaintance, "the problem of how, given the nature of information transmission, we could be acquainted with any of the properties of external things represented by our experience" (p. 187). Ultimately, Johnston holds that the problem appears insoluble. "Upon examination", Johnston writes, "perception's promise to acquaint us with properties of external things seems to be utterly fraudulent" (p. 185). What is his argument? Consider two premises: first, the nature of any perceptual experience is partly a product of the thing causing the experience and partly a product of the thing having the experience. And second, we cannot "separate out" the contribution to our perceptual experience of our own sensibility from the contribution to our experience of the objects sensed. Johnston concludes from these premises (p. 187):

"Therefore we cannot take our perceptual experiences to reveal the natures of [properties of] external things". I suspect, however, that Johnston's argument is subject to the following dilemma: either the second premise is false or the conclusion does not follow. Either we can, in the relevant sense, separate out the contribution of our sensibility from the contribution of the object sensed, or we cannot. But the sense in which we cannot is no bar to perception's revealing the nature of properties of external things. What exactly does Johnston have in mind? I think there are two kinds of argument he might be making, one metaphysical, the other epistemic. Let's consider these in turn. Johnston is interested in the fact that any experience could be the result of any of a number of different causes. And these causes might be motley -intrinsically highly dissimilar. Johnston points to the cartoon of the brain in a vat: its experience is as of a life in Boise. The cause of those experiences is not, however, life in Boise but rather the inner workings of the vat computer. Unfortunately, nothing in the experience rules out that it was caused by the computer. For Johnston, the upshot is that experience could not reveal the nature of its causes. But a closer look may reveal a different conclusion. Why does it follow from the fact that something other than life in Boise might have caused a given experience that the experience does not reveal the nature of certain properties of life in Boise? As Johnston correctly notes, "for all that could be revealed in a fully coherent experience either could be the causes of that course of experience" (p. 187, emphasis mine). But from this it does not follow that "perceptual experience does not reveal the nature of its causes" (p. 188, emphasis mine). Of course, I presuppose that a property can be revealed to you in an experience even if that experience might have been caused by some other property. T h e causal process might be nonstandard, and in those cases your experiences in a n important sense do riot reveal the nature of properties of life in Boise. T h a t possibility holds even for cases where the process is in fact standard. But in these cases, the cases in which the causal process is standard, the experience does reveal the nature of the relevant properties. According to me, we are often acquainted with properties of external objects in perception. And this is so even though any perceptual experience could be misleading. The thing is, very often -and often e n o u g h they're not. My response may be understood more clearly in connection with a discussion later in the paper. Consider what Johnston calls a n

19. GETTINGACQUAINTED WITH PERCEPTION211

"obvious causal condition on seeing a feature or property" (p. 190, emphasis mine):

(C) If a property of a thing is literally seen, then the thing's having that property must be part of the causal explanation of the visual experience which counts as seeing that the thing has that property. I highlight the modality in Johnston's condition because the scope of the "must be" in that condition could cause some confusion. What is necessarily the case is that if something's property is perceived, the thing's having that property would be part of the causal explanation of the relevant perceptual experience. The modal operator has wide scope, so to speak. Johnston's condition should not be understood to mean that if a property is perceived, then, necessarily, the thing's having that property would be part of any causal explanation of the relevant perceptual experience. The necessity of the consequence holds; but not the necessity of the consequent, not even given the actuality of the antecedent. So I accept Johnston's condition (C). But I doubt that it is inconsistent with that condition that any given perceptual experience might have been caused by something's having some property very different from the property whose nature the experience (in fact) reveals. One way to understand my response here is as follows: I suggest that Johnston may be putting too strong a metaphysical demand on revelation. It seems to me that according to Johnston, for S's experience e to reveal p (which is a property had by o), necessarily, for any case of e, it must be that 0's having p would be part of the explanation of S's experience e. That Johnston holds this is supported by his rejecting the possibility of perception's providing acquaintance on the basis of the fact that for any experience, it might have any of several intrinsically dissimilar causes. But what cannot fail, for perception to be revelatory, is only the conditional as a whole. Given the necessity of the consequence, however, and given even the actuality of the antecedent, the necessity of the consequent does not follow. It remains possibly true that the experience be caused in a nonstandard way. And this possibility does not undermine perception's capacity to reveal the nature of external properties. Earlier I suggested that there are two interpretations of Johnston's argument. I have suggested a reponse to what might be called the metaphysical interpretation. There is also a kind of epistemological argument to be discerned in the paper. My response to this epistemological interpretation is in some ways similar to my response to

the metaphysial interpretation. Here too it may be that too strong a demand is put on revelation. Compare the following three arguments: Argument 1: (1.a) If perception is to be a mode of revelation of z for S , then there must be a n aspect of S's perceptual experience which matches and is caused by z . (1.b) It is not the case that there is an aspect of S's perceptual experience which matches and is caused by the thing perceived (as opposed to being caused by the viewing conditions or by the particular sensibility of the subject). Therefore,

(J) it is not the case that perception is a mode of revelation of z for S. Argument 2 : (2.a) If perception is to be a mode of revelation of z for S, then there must be a n aspect of S's perceptual experience which S knows to match and be caused by z . (2.b) It is not the case that there is an aspect of S's perceptual experience which S knows to match and be caused by the thing perceived (as opposed to being caused by the viewing conditions or by the particular sensibility of the subject). Therefore,

(J) it is not the case that perception is a mode of revelation of z for S. Argument 3 : (3.a) If perception is to be a mode of revelation of z for S , then there must be an aspect of S's perceptual experience which S can separate out with absolute certainty as matching and being caused by z . (3.b) It is not the case that there is an aspect of S's perceptual experience which S can separate out with absolute certainty as matching and being caused by the thing perceived (as opposed to being caused by the viewing conditions or by the particular sensibility of the subject).

19. GETTING ACQUAINTED WITH PERCEPTION 213

Therefore, ( J ) it is not the case that perception is a mode of revelation of z for S.

I am not sure which of these arguments Johnston means to advance. But while I'm willing to accept (l.a), (2.a) and (3.b)' I am pretty sure that (1.b) and (2.b) are false. (3.a) is not obviously false; but the obvious causal condition on seeing -(C) above- is insufficient for that premise. It may be that Johnston accepts (3.a) as an independent premise. But the explicit defense offerred -(C)is sufficient only for (1.a) or (2.a). As for the independent plausibility of (3.a), I'm inclined to doubt it. Suppose I have a visual experience. And suppose there is an aspect of that visual experience that in fact matches and is caused by a certain feature of an external thing. Finally, suppose I take that aspect to reveal the nature of that feature. Now, Johnston rightly notes that I cannot be sure that the aspect which I take to reveal the nature of the feature does so -it may match or be caused by some feature of the viewing conditions or of my own sensibility. But, for the feature to be revealed in experience, do I need to be sure that I am right about what the aspect reveals or do I need simply to be right about what the aspect reveals, or at most to be right with adequate justification, even if that justification falls short of guaranteeing correctness and warranting absolute certainty? Isn't it enough, for the nature of the feature to be revealed to me, that there be an aspect of the experience that I separate out (with adequate justification) as revealing the nature of the feature and that in fact does match and is caused by that feature? I think so. There is a kind of certainty about the nature of the external world (as such or as cause of our experience) that is just not be had: knowledge without the possibility of error. That is just a chimera and epistemology should be no more embarrassed about our lacking it than metaphysics should be embarrassed about our lack of a round square. Skeptical projects regularly rue our inability to achieve a kind of justification for belief that is incompatible with error. But that inability is no failure. We might pre-analytically have longed for that kind of justification. But we should give up that hope. Returning now to Johnston's position, I do not yet see a good argument for (l.b), (2.b) or (3.a). I'm willing to accept (l.a), (2.a), and (3.b) as they stand. But I don't yet see why we should accept (l.b), (2.b)' or (3.a) as they stand. I know what contribution a given leaf's color makes to my experience. I can, in one sense, separate out the contribution of the

leaf's color from other possible contributions. The phenomenology as of green is provided by the leaf's color. That phenomenology might have been provided by a computer; but it was not. I cannot be sure that what I take to be contributed by the leaf's color is not in fact contributed by something else. So in that sense, I cannot separate out the contribution of the leaf's color from other possible contributions; but nevertheless I know that the phenomenology was contributed by the leaf's color. To highlight my disagreement with Johnston, let me quote from one of the central passages in his paper. If my visual experience is to be a revelation of the nature of the greyness of my dog's coat then that visual experience must correctly represent how that greyness is. Now the way my visual experience represnts greyness as being is fixed by the internal states that I am in at the time of the experience. By the causal condition on seeing [(C) above], those internal states are to be causally explained by my dog's being grey. This means that there is a causal process connecting the state of my dog's coat and my internal states. With all of this, I agree wholeheartedly. But the passage continues in what I consider a puzzling way. Barring a preestablished harmony, no such causal process will preserve and transmit information so as to secure a nature-revealing match between how some feature of the cause, say the greyness of my dog's coat, is and the way I am caused to represent that feature (p. 190-1, emphasis mine). Unless he is demanding too much by "securing" a nature-revealing match, in either of the ways I suggest and respond to above, I do not understand why Johnston is skeptical about causation's providing the requisite information transmission. Causation cannot secure that match in a way that is guaranteed -metaphysically or epistemically necessitated. But it might secure the match for all that. In fact, I think, it does.


Would more acquaintance with the external world relieve epistemic anxiety? E. Villanueva

Professor Mark Johnston faces the epistemic anxiety which arises from the necessity to be acquainted with the nature of the properties of external things represented by our experiences (p. 187) in perception. He reminds us of Russell's ambitious model of acquaintance according to which we aim to be acquainted with the whole nature of external thing's properties and after concluding it to be impossibly high (p. 191) considers a weakened version according to which acquaintance turns to be a matter of degree (p. 192). But what sort of acquaintance are we talking about? Johnston talks of strong acquaintance which has "the standard of complete knowledge of a property as it is in itself.. . " (p. 191) and a weakened acquaintance that "issues in an ability to recognise instances of the properties and in knowledge of comparative [essential] similarities among families of properties. . . " (p. 192) Elsewhere he speaks of "the intuitive link between acquaintance and the cognitive contact which helps to make knowledge non-instrumentally worthwhile" (p. 195)

and of acquaintance giving birth to propositional descriptive linguistic knowledge (p. 192).l But this weakened version of the acquaintance's model, Johnston argues, betrays a tension either between colors as categorical properties of external things (p. 193) or as dispositions of external objects (p. 195) on the one hand, and acquaintance, on the other. Acquaintance cannot penetrate more beyond a minimum of each of the properties of external things; it cannot penetrate and get till the core of these properties or till much more than what experience delivers. What Johnson is aiming at is to get a larger amount of acquaintance of the external property, a larger share than we are currently getting. Johnson is voicing a complaint when he says that we are getting a small amount of acquaintance, far from what we are in need of. Johnson concludes sceptically that perception fails to acquaint us with or reveal the natures of the properties of external objects (p. 185) providing only a schematic and bloodless knowledge (p. 190 & 198) of the external world, a knowledge that is only instrumentally valuable as a sketch of that world in order not to get lost in it; a sketch which allows for carrying on intentional actions (p. 191) but does not provide the real thing, the knowledge by direct contact that could remove the epistemic anxiety pointed above and thus, even if still partial, provide proper knowledge. I want to rise some questions to Johnston concerning both his complaint and scepticism. Why is it that visual perception of shades does not acquaint us with the colors even in this weakened criterion of acquaintance? In the case of categorical primary qualities2 it is because similarities are given at the sensational level and grounding these similarities [i.e. trying to go beyond these similarities to answer why for example a shade of canary yellow is not a shade of blue] relies on facts discovered by science, not on the weakened criterion of acquaintance (p. 194). An affinity or any other strong link cannot be secured among microphysical facts and shapes present to vision. There seems to be an abyss that acquaintance cannot bridge. We seem not to be able of direct contact with the complex condition (ibidem) which explains the essential difference among colors. And this shortcoming is what produces epistemic anxiety, according to Johnston, for even if it is authentic justified knowledge invulnerable to the sceptic threat it is a bloodless and schematic knowledge and nothing short ' ~ h u sJohnston does not claim t h a t acquaintance is knowledge.

'1 shall ignore t h e view of colors as dispositions.

20. WOULDM O R E A C Q U A I N T A N C E . . .

217

of acquaintance with the external property would confer substance to that knowledge (p. 198). Acquaintance on this reading is the only thing that could give birth to or validate a knowledge that could be capable of satisfying the epistemic anxiety pointed above. If we were talking in terms of supervenience the explanation of color differences would lie at the subvenient base of microphysical facts that would not be available to acquaintance. Now what prevents us from being acquainted with those base facts? Is it operating here a metaphysical divide that is brought about by the way the mechanism of consciousness3 or experience works? That is, appearances cannot constitute the base subvenient facts out of metaphysical impossibility; on the contrary, what is aimed at is to make available to acquaintance both the subvenient base facts, the microphysical ones, on the one hand, and on the other, make available the way, the mechanism by means of which, color differences supervene on these microphysical base facts. The picture would be like this: we are acquainted with microphysical facts but the mechanism of consciousness makes appear these microphysical facts as color differences in a partial acquaintance that leaves us unacquainted, for example, with how the microphysical facts determine these differences. Now in that picture it seems reasonably clear where are we pointing at, where the difficulty seems to lie. But do we? Do we have a reasonable clear idea of what would be to be acquainted or have direct cognitive contact with that part of the color property, that is, the complex condition (p. 194) that provides knowledge of (and explains) color differences? It seems as if that complex condition is set at a different level of that of the colors and thus the idea of making direct cognitive contact with it is crucially unclear. It seems as if the metaphysical divide between things as given in experience or acquaintance and things as they are independently of experience is to be overthrown; as if a revisionist trend is on its way and Johnston wants that properties show or exhibit their real natures in experience, only that Johnston demands that most of the real natures -not the whole of them- become apparent.4 It is as if Johnston demands that most of the level of microphysical facts abandon its place and make themselves into appearances delivered in the direct contact of conscious experience or acquaintance. Again, in a less pretentious way, in an epistemic read30ddly, consciousness never appears in Johnston's text. 4 0 n this metaphysical reading of Johnston it could be charged that he aims at repelling the human epistemic condition as far as this condition is rooted in that metaphysical divide.

ing, it is as if Johnston wants most of descriptive knowledge to be derived from acquaintance only on pain of becoming schematic and bloodless. It is as if knowledge of the microphysical level (where essential differences among colors lie) despite being authentic knowledge -not just luck or mere coincidence- for it stands sceptical threats, remains a schematic knowledge only, necessarily incapable of calming down our epistemic anxiety. But why it is not enough to relax the epistemic anxiety pointed above to accept that we get acquainted only with a part of the property perceived and that then that part has to be supplemented with some other inferential means to get knowledge of the rest of the property or at least with a further substantial amount of such knowledge? How could it be that knowledge that does not spring out of acquaintance is declared by Johnston schematic and bloodless, with instrumental value only, even if it can overcome the fierce sceptical doubts? To put it in another way: Why is it that after realizing that we have these two complementing ways of reaching knowledge of perceptual properties, knowledge intended to satisfy any epistemic queries -including those characteristic of the s c e p t i c there is still room for epistemic anxiety? Why is it that only if we get direct contact with most of the property's nature we can relax that epistemic anxiety? In sum, why from all things, it has to be acquaintance? Presumably acquaintance is a necessary grounding of knowledge because only its direct contact provides enough grounding basis for carrying on intentional actions. It is only if a person is acquainted with most of the properties' nature that she can act thoroughly in full intentional confidence and thus can carry on successfully his actions till the very end. But there is something missing here for if humans have a kind of knowledge that is capable of defeating the sticky doubts of the sceptic, surely that sort of knowledge will prove capable of providing enough ground to carry on a full extent their most cherished intentions in complete actions through the world. Surely that kind of knowledge establishes a good enough link with microphysical facts to erase any legitimate epistemic anxiety remaining. Thus Johnston's complaint seems to me strange and excessively demanding and the subsequent scepticism misplaced. But before advancing anymore I would have to know a lot more about the sort of acquaintance involved, about its connection with knowledge, about the nature and legitimacy of that epistemic anxiety and about what appears a distinct fascination with the idea of getting in close touch, of getting in direct cognitive contact, as the only source of reaching a knowledge that is substantive and thus is able to remove that epistemic kind of anxiety out of our epistemic beliefs.


A Mind-Body Problem at the Surface of Objects Mark Johnston "Seems? Nay Madam, is. I know not 'seems'." Hamlet. Act 1, Scene 2.

Along with others at this conference, I take it that the visual representation of color is layered. However the layers or tiers of r e p resentation that I wish to highlight here are not those discussed by Michael Tye and William Lycan. First there are representations to the effect that such and such objects have such and such colors and shapes. Then there are representations of what these colors and shapes are like, what their natures are like. If you wish, this could be understood as the attribution of higher-order properties to the color properties themselves.' Vision purports to tell us not just ' ~ i k eThau, who is writing a dissertation with me on perception and from whom I have learnt much on these issues. believes that I should instead think of perception as attributing to surfaces both colors, which surfaces do have, and a distinct set of colorlike properties which surfaces could not have or be seen to have. Thau believes that externailsm holds for the colors and internalism for the colorlike properties. I find no evidence for this double attribution on the part of perception. In coming to master color terms we never find ourselves confused as between which of the two properties which visual experience represents surfaces as having is supposed to be denoted by the term we are mastering. The best explanation of this is that the two properties are not two but one. Some readers will note that in "How TO Speak Of The Colors" I think that we can still talk truly of the world being colored even if some of the h i g h e r ~ r d e rattributions

that things are canary yellow but also what canary yellow is like. That is why vision seems to give us a particularly intimate contact with the visible aspects of visible things -it seems to lay the natures of these aspects bare.2 Unfortunately, we lack the vocabulary to well articulate the content of these higher-order representations. Some talk of the dissectiveness of the colors, their particular "grain", the way red and yellow come toward you while blue and green discreetly "hang back", the way we can see red and blue in purple, and so forth. In the conference paper I tried to bring out the obscure content of these higher-order representations. My strategy was to focus upon the color similarity and difference claims that we seem to know simply from a course of color experience. The so-called "color solid", which locates all the shades in a three-dimensional array structured by the dimensions of hue, saturation and brightness, is structured out of such similarity and difference claims about each shade. So for example we seem to know that canary yellow is not as similar in hue to teal and turquoise as they are similar in hue t o each other. I suggested that we would have the rich kind of knowledge of canary yellow which would constitute visual acquaintance with its nature if we knew a host of such similarity and difference claims and knew them simply on the basis of a course of experience of the colors of things. If we had such knowledge in that way I would say that vision acquaints us with the nature of canary yellow; that is, shows us how canary yellow is in itself and not merely how it appears to us and to creatures with our type of sensory system. Clearly, if knowledge of such similarity and difference principles is supposed to be the upshot of knowing canary yellow as it is in itself and not merely as it appears -the upshot of being acquainted with canary yellow- then similarity in respect of hue had better not turn out to be merely apparent similarity in respect of hue. Mutatis mutandis for similarity in respect of brightness and saturation. So I a m a little surprised to find in Alan Gibbard's subtle paper this proposal: Vagueriess aside, then: There are physical properties that constitute being various specific versions of blue, yellow and the like; these are physically uninteresting physical properties. There is a physical metric are false of the colors. In t h e paper read a t Canccn I raised one difficulty which the earlier paper does not deal with, namely t h e fact t h a t on t h e dispositional account vision acquaints one as much with oneself as with a surface. his way of putting t h e point is due t o Greg Harding from whose deep work on color perception I have learnt much.

21. A MIND-BODYPROBLEM AT THE SURFACE OF OBJECTS

221

that constitues degree of similarity in color between the various physical properties that constitute exact colors. This is a physical metric that is physically uninteresting. This similarity of color, the point is, isn't plain similarity, apart from us, among those properties that constitute exact colors. It is visual similarity. Still this is a genuine kind of similarity: it is similarity i n a respect. In seeing things we discern this genuine kind of similarity-or we do when we've got more or less normal light and the like, when we are seeing things, as we'd naively say, in their true colors. I don't find, then, any vindication of the claim that I'm suffering from a massive illusion in my normal visual life. In seeing I discern genuine physical properties, and I discern a genuine vague metric of similarity of color among them as degree in likeness in color. (p. 206) "Visual similarity" is Alan's name for visually apparent similarity in respect of hue. It is a perfectly genuine kind of similarity, but it is precisely the kind of genuine similarity that is ill-fitted to explicate acquaintance with the colors. The question was: are we acquainted with the natures of the colors or do we simply know them by description as the properties whatever they might turn out to be which visually appear to us to be similar and different in such and such ways? When Alan offers his proposal about similarity in hue, thereby suggesting that we can do no better, he is answering that question in the same way as I did. Of course he urges that we should still call our relation t o the colors "acquaintance" since all we have discovered is that the colors are idiosyncratic physical properties with natures that vision does not reveal. T h e usage is a little strange if we mean t o be falling in with Russell's use of the term in Problems of Philosophy. What Alan calls knowledge by acquaintance is what Russell calls knowledge by description. T h e Russellian tradition of use is not engaged by Alan's observing that one is in the ordinary sense "acquainted" with the sound of one's wife's voice and the look of her smile. To fully understand Alan's physicalist proposal (which takes its inspiration from Saul Kripke's work) we must understand that among the color properties the proposal aims t o identify with physical properties are the most general color determinables: hue, saturation and brightness. Among the principles that have t o be held true as a condition on any three physical properties deserving the names of hue, saturation and brightness is the principle that any example of one of these properties is also an example of the other two; i.e., there are no saturationless hues or hues with no position on the brightness

scale.3 The physicalist account of color will then take this form: If a good deal of PI . . . PN turn out to be true of some triple of physical properties H, S and B then Hue = H Saturation = S Brightness = B , otherwise these property names do not denote.

I a m struck by the structural similarity between the above account of color and a reference-fixing account of legendary kinds, kinds that for all we know may or may not exist. "Greek Fire" is a legendary kind name. Among the legends concerning Greek Fire we find that it exploded into flame upon hitting the water. We do not know for sure how much if any of the legends surrounding Greek Fire are true. Could the Greeks have discovered a way to refine raw potassium? In any case, if enough of the legend turns out to be true of some stuff then "Greek Fire" denotes that stuff, otherwise "Greek Fire" does not denote anything. The very fact that there is such a structural analogy suggests that there is something missing from the physicalist's account of the kind of contact with the colors that vision provides. If the preconditions on the physicalist account of color aren't well enough satisfied -C.L. Hardin for one thinks that they are not4then there are no colors and so we are not acquainted with the colors. There is simply a physicalist story of how a physical world without color produces certain illusory experiences in us. If the preconditions are well enough satisfied then there are colors. We do not yet know whether the preconditions are well enough satisfied. So according to color physicalism we do not know whether there are colors. (For if the physicalist account is true then it is knowable and if we knew there were colors then we could by inference know now that the preconditions are satisfied.) Now since we do not know whether there are colors, we cannot know on the basis of a course of experience of 3These kinds of relatively a priori interelations among t h e color properties make finding physical indentificands for hue, saturation and brightness a somewhat risky business. T h e highly disjunctive nature of any proposed physical candidates for hue, saturation and brightness makes it more likely t h a t something could satisfy one of the disjuncts for our best candidate for saturation and none of t h e disjuncts for our best candidate for brightness. 4See his Color For Philsophers (Hackett, 1988) pp. 127-34.


223

the colors such things as canary yellow is not similar to teal in hue. All we can know, according to color physicalism is that if the colors exist and if in particular canary yellow and teal exist then canary yellow is not similar in hue to teal. If that is the best that visual experience can do by way of providing us with knowledge of the colors then it evidently does not acquaint us with the colors. For that conditional knowledge can be had even in case there are no colors. As I urge in the conference paper, what sense is there in calling a state of knowledge that can exist whether there are colors or not knowledge by acquaintance with the colors. No sense, at least if we are interested in explicating the metaphor of cognitive contact. Probably because of my careless wording, Alan takes me to task for supposing that some experience could not constitute acquaintance with some thing unless the experience guaranteed that the thing was there to be experienced. (See his example of Beth and her twin Jo. Alan says rightly that he is acquainted with Beth even if his experience could have been produced by Jo.) I do not suppose that the experience guarantees that the object of the experience exists and caused it. In my own defense, I should point out that Alan omits the first half of the sentence that he quotes when setting me up for the Beth and J o argument. In the first half of that very sentence, I wrote "I am not here following Russell, as he is usually interpreted, in suggesting that there is something about the subjective side of acquaintance which guarantees that acquaintance has an object. .. " What I do suppose is that knowledge that could be had whether or not a thing exists is not knowledge by acquaintance with that thing. Such knowledge does not constitute cognitive contact with the thing. I think that a color physicalist might agree with this. The physicalist will simply ask why such acquaintance or cognitive contact matters. Both Alan and Enrique Villanueva rightly urge this question against me. I shall say something about it, but first I must answer David Sosa's questions and criticisms. By way of an elegant provocation to get me to say exactly what I mean, David sets out various unsavory alternatives. Despite David's care I cannot comfortably force my own argument into any one of his strictures. As against David's Argument 2, I do not require for acquaintance with a color, or better, a color shade that one know that one's perceptual representations match and are caused by that shade. Acquaintance with some property does not require that you know that you are acquainted with that property. As against David's Argument 3, I do not require that one must be able with certainty to separate out some aspect of one's experience as matching and being caused

by the shade. Acquaintance with a property does not require any such certainty. Nor do I require, as David's discussion of Principle C suggests, that if one's perceptual experience on a particular occasion is to acquaint one with the nature of a shade had by the object seen on that occasion then necessarily the thing's having that shade is part of the causal explanation of the perceptual experience. If that reading of Principle C is correct -which I doubt- then it is correct only because of something about the necessity of the causal origins of perceivings, something irrelevant to the matter a t hand. I do require that for acquaintance with a shade there be a "nature revealing match" between how the shade is and how the shade is represented by one's perceptual experience of the thing with that shade. And I require that the thing's having that shade be part of the causal explanation of the experience. (So David's Argument 1 is closest to my intent, although the vague reference to aspects and the requirement that only the object of acquaintance cause the perceptual experience are at odds with my strategy.) Features of my presentation may have led David to consider his Argument 2 and Argument 3. I mentioned the brain in the vat scenario as one in which two different but related epistemic anxieties are presented. The anxiety about justified belief or propositional knowledge is the more familiar anxiety. Here some say: since I could be a brain in a vat then I do not know any of the ordinary things I claim to know, e.g. that I have visited Boise, Idaho. Others respond that this argument places unreasonably strong metaphysical or epistemological demands on knowledge. They say: contrary to the brain in the vat argument, my present state of belief that I have visited Boise can constitute knowledge even though I could be in that state while never having visited Boise. Likewise they say: my present state of belief that I have visited Boise can constitute knowledge even though I could be in that state without my being certain that I have not always been a brain in a vat in, say, Tabriz. Propositional knowledge about visiting Boise requires that my belief be non-adventitiously related to the fact of my visiting Boise. It does not require that my present state of belief either necessitate or provide me with certainty that I have visited Boise. There may be something to those arguments. David presses versions of those arguments against my discussion of a different anxiety, the anxiety concerning acquaintance. The anxiety concerning acquaintance is felt when thanks to the brain in the vat scenario we vividly realize the extent to which our style of representing external objects and their properties is dependent on the nature of our own sensibility. For the brain in the vat


225

scenario shows that things with very different natures, i.e. the vat computer and life in Boise -can be represented in the same way. So we, or at least I, come to anxiously feel that our style of representing the properties of external things is too much due to our sensibility to represent those properties as they are in themselves. This argument does not turn on what the brain in the vat scenario shows might be the case or might be the case for all we know. Instead it turns on what the brain in the vat scenario suggests is actually the case about our style of representing the properties of external things, even their observational properties. It is a style of representing shaped by our brain and central nervous system. There is good reason to believe that the way the colors are thereby represented as being matches no property of objects and hence no property which figures in the causal explanation of visual experience. At the end of his paper David asks what sense can be made of this talk of matching on which I place so much weight. Matching is sharing properties. So if we can articulate the content of the higher order representations by which vision depicts the colors -the properties it attributes to the colors- then we have only to determine whether any properties of surfaces which figure in the causal explanation of visual experience have the higher-order properties in question. The higher-order representations are difficult to articulate directly. My suggestion was to articulate them indirectly in terms of the similarity and difference claims which they support. Then we ask whether these claims (and not just their "seeming similarity" counterparts) can be made true by the relevant properties of surfaces. In discussion at the conference I considered another account of how the colors as we see them could figure in the causal explanation of visual experience. They could supervene upon properties which are more directly causally relevant; e.g. reflectance profiles. Just as the table before me gets to play a role in explaining why I see the table even though the molecular and perhaps quantum mechanical structure of its surfaces are what directly determine what I see, so canary yellow gets to play a role in explaining visual experiences even though the reflectance profiles are what directly determine the character of those visual experiences. On this theory, the property that gets to play this role is a property which really deserves the name of "canary yellow". It is a property whose properties match the higher-order properties which visual experience represents canary yellow as having. A number of remarks need to be made about the prospects for this account. First, the analogy with the table may convey the general idea of supervenient causal explanation but it cannot motivate that

idea for the case of the colors. In talking of the table as a supervenient explanatory cause we are implicitly assuming that the sensible qualities -color among them- which figure in our representation of the table can be truly attributed to the table. Secondly, we must not let a lazy reliance on talk of property constitution obscure the richness of the representations of the color properties which have to be constituted by reflectance profiles. It is no easier to make intelligible how the color properties as we represent them could be constituted by reflectance profiles then it is to make intelligible how the property of being in pain could be constituted by the property of having one's C-fibers firing. The supervenience account leaves us with a variant of the mind-body problem a t the surfaces of objects. Thirdly, in order for the supervening colors to be visible there must be a non-accidental regularity to the effect that other things being equal subjects looking a t a surface will visually represent its color as canary yellow when and only when the surface is canary yellow, i.e. only when a property of the surface matches the higher-order representations of canary yellow. This is the "pre-established harmony" mentioned in the conference paper. So far as fitting us or any other creatures for survival, such matching of properties would never do any work over and above the work done by reflectances causing the rich perceptual representations which make up our experience. This is not an argument to the effect that we might be wrong due to the holding a highly adventitious alternative explanation of our experience. It is an argument that what is required for our representations to match surface properties is itself highly adventitious. The fourth remark is prompted by considering that the representation of color in our visual experience is not as of a family of lightand observer-dependent properties, but of a family of properties whose natures are seen when the light is right and the observer is well-placed. (Contrast the highlights; a course of experience as of the highlights represents them as light and observer dependent. As one moves the highlights change, when the lights dim they cease to exist, while the colors get harder to see.) The colors are not visually represented as relations and any account of the colors which takes our visual experience as a definite guide to the natures of the colors should not represent them as relations. Combine that fact with the fact of color variation, the fact that different but perfectly adequate perceivers in different but perfectly good viewing conditions see the same things as having different colors. So from your side of the causeway the sea looks grey and from mine the same body of water looks green. (This is of course a case of volume color but the same could be said for both surface and radiant color.) You have a


227

veridical experience of the sea's greyness. I have a veridical experience of the same body of water being green. So that part of the sea which we each look at is both grey and green. Now the pre-established harmony gets considerably more delicate. The reflectance profiles of the water constitute both greenness and greyness. When I look at the sea it is the greyness and not the greenness of the sea which explains my veridical experience of the sea's being grey. When you look at the same part of the sea it is the greenness and not the greyness of the sea which explains your veridical experience of the sea's being grey. This is, to say the least, an odd kind of explanatory selectivity. Believe in it if you can. The fifth remark is that once we consider the different quality spaces of other animals with color vision then the problem of multiple colors on surfaces playing their selective explanatory roles ramifies. Are we for instance to include among the colors of any surface all the colors which any possible animal with color vision could normally see it as having? Allowing this leads to the sixth remark, namely that the resultant account of color seems to make sense of questions which seem silly. Suppose for example that a surface is purple as well as blue and red. I have an experience of it as purple. It seems to make sense for me to wonder whether this experience of purple is to be causally explained by the surface being purple or jointly by the surface being blue and the surface being red. These remarks can be left as questions for any account, such as Colin McGinn's which tries to respect the richness of our color representations while attributing matching properties to surface^.^ Short of satisfactory answers to these questions I a m still inclined to conclude that we do not see surface properties which match our representations of the colors. Why then do I think that this matters so much? Isn't my interest in what I call 'acquaintance with the colors' an arcane interest, as Alan suggests? Alan represents my interest as the demand that beings with very different sensory systems -his Alpha C e n t u a r a n s would nonetheless find color properties important in their explanations. He then says: they wouldn't and so what. But I never mention such creatures, nor am I particularly concerned with what form their science might take. Rather, I'm concerned with the consequences of the fact that no properties of surfaces match the colors as they are represented as being by visual experience. It follows that either surfaces are not colored or vision does not tell us what surface color is like. Either way vision does not tell us what surfaces are like and if it 'See Colin McGinn "A New Look At The Colors" Ms.

doesn't tell us what surfaces are like it doesn't tell us what bodies are like. Mutatis mutandis for touch and tactile qualities. The upshot is that our senses don't tell us what bodies are like but only how they appear. Why is that bad? In the same vein Enrique writes Johnston's complaint seems to me strange and excessively demanding and the subsequent skepticism misplaced. But before advancing anymore I would have to know [among other things]. . . about what appears to be a distinct fascination with the idea of getting in close touch, of getting in direct cognitive contact.. . . (p. 218) I cannot give a full account of my "distinct fascination" here, but a partial account might begin with a range of emotions t h e aesthetic, sensual and erotic emotions, which are ways of taking pleasure in the revealed sensible natures of things. As such they presuppose that their objects are as they are represented by the senses. The upshot of our indictment of vision and by implication the other senses is that the aesthetic, sensual and erotic emotions should now be replaced by their relational surrogates: instead of taking pleasure in watching my greyhound run I should now take pleasure in being under the causal influence of something whose nature my senses fail to reveal to me but which produces in me certain pleasing appearances as of a greyhound in full stride. Here the pleasure one animal takes in watching another has been replaced by a pleasure taken in certain appearances. A fundamental role of the aesthetic, sensual and erotic emotions t o draw our attention away from ourselves and our inner lives by pleasing us with the presence of others- has been philosophically undermined. What is left are just ways of being pleased with our own reactions to things, whatever those things might be like. The philosophy of perception leads us from what Husserl called the natural attitude towards what we might call the pornographic attitude. To measure the human cost of this try substituting the relational surrogates in your own life. Try explaining to the mother of a newborn that her sensory perception of the child, which drives her feelings for it, leaves the nature of the child completely unrevealed except for the fact that the child is something or other which causes appearances and feelings in her. Alternatively, consider a fast approaching realm in which we are promised an endless variety of ways in which we are turned on by the effects of others upon us. In virtual reality, we are told, the homeliest pair will be able to present to each other as Gregory Peck and Ingrid Bergman. A fun way to spend a half hour, but a hideous

21. A MIND-BODY PROBLEM AT THE SURFACE OF OBJECTS

229

way to spend a life. Why hideous? Because inevitably the emotional focus comes to be on whether and in what way the consistent Peckish and Bergmanesque appearances please. This is a generalization of the pornographic attitude, an attitude that is limited and wrong not because it involves sexual pleasure but because it involves taking sexual pleasure in the sensory appearances rather than in the person appearing. Yet the upshot of our reflections on acquaintance is that we are already inhabitants of a virtual reality. Of course, if I had learnt to care for persons not as intelligent animals whose living bodies manifest that intelligence, but as Kantian somethings I know not what, then this would not matter as much. But I haven't and it does.


Precis of M i n d a n d w o r l d 1 John McDowell

1. The idea of world-directedness -that is, content in one s e n s e is intelligible only in terms of a normative context that is its primary home. We must be able to work with the notion of a posture or stance that is correctly or incorrectly adopted according to whether or not things are thus and so. Only so can we understand the posture or stance as a judgement or belief t o the effect that things are thus and so. (If we can make sense of judgement or belief as directed a t the world in that way, we need have no trouble with other kinds of content-bearing postures or stances.) We might express the point like this: thinking that aims a t judgement, or a t the fixation of belief, is answerable to the world for whether or not it is correctly executed. And now a small step away from that abstract formulation takes us t o a minimal, and one might think undisputable, empiricism: in the sorts of case that must come first for reflection on the very idea of directedness a t the world, the world's verdict, to which thinking must be answerable if it is to be 'Harvard University Press, Cambridge, Mass., 1994. I restrict myself here to those aspects of the book that are relevant to the topic of the SOFIA conference; and, with a view to capturing the gist of the book in a much smaller space, I allow myself to approach its themes in different ways.

thinking at all, is delivered by way of a pronouncement from (in Quine's phrase) "the tribunal of experience" .2 Minimal empiricism promises to cast light on certain sorts of philosophical anxiety. If there is an obstacle in the way of seeing how experience could serve as a tribunal, then by the same token that obstacle will seem to render urgent the question: how is empirical content (as we might put it) so much as possible? And only our small step separates that question from the question how content -world-directednessis possible at all. When one thinks about philosophical anxieties that arise in the neighbourhood of empiricism, that may not be the first question that comes t o mind. Such anxieties are more familiar in the shape of questions like this: how is empirical knowledge possible? That is, in terms of the juridical metaphor: how can experience, sitting in judgement on, say, a belief, return a verdict sufficiently favourable for the belief t o count as a case of knowledge? But suppose we find it puzzling how experience can be such as to return any verdicts on our thinking a t all. Such puzzlement would clearly be prior to any question about how experience can return a verdict that reaches some high level of favourableness. I think it is helpful to see the problems about knowledge in particular that pervade modern philosophy as more or less inept expressions of a deeper anxiety -an inchoately felt threat that a way of thinking we find ourselves falling into leaves minds simply out of touch with the rest of reality, not just questionably capable of getting to know about it. This underlying anxiety is not well captured by questions about knowledge, but it is well captured by asking how content is possible -which is achieved, in the context of minimal empiricism, by asking how empirical content is possible.

2. I &m suggesting that if we can find something that makes it hard to see how experience could serve as a tribunal, then, given the attractiveness of minimal empiricism, we shall be able to see that as the origin of the characteristic anxieties of modern philosophy. The prospect is that we can trace those anxieties to a single source, so that in principle they can be exorcized together. And the source I want to point to can be brought to light by considering Sellars's attack on the Given. 2 " T ~D o ogmas of Empiricism", in W.V. Quine, From a Logical Point of View (Harvard University Press, Cambridge, Mass., 1961; 1st ed. 1953), pp. 20-46, at p. 41.

22. P R ~ C OF I S MIND A N D W O R L D

233

Sellars insists that the concept of knowledge belongs in a normative context, and it is an implication of something I have urged that we should take him to be stressing just one aspect of the normative context that is necessary for the idea of being in touch with the world at all, whether knowledgeably or not. Focusing, as he does, on knowledge in particular, Sellars writes: "In characterizing an episode or a state as that of knowing, we are not giving an empirical description of that episode or state; we are placing it in the logical space of reasons, of justifying and being able to justify what one says."3 One way of putting what Sellars is driving at is to say that epistemology --or, in the more general version of the thought that I have insisted on, reflection about world-directedness as such- is vulnerable to a naturalistic fallacy.4 If we put the thought like that, we are identifying the natural, as indeed Sellars sometimes does, with the subject matter of empirical description as he conceives it in the passage I have quoted, where he contrasts empirical description with placing something in the normative framework constituted by the logical space of reasons. Sellars, then, draws a distinction between, on the one hand, concepts that are intelligible only in terms of how they serve to place things in the logical space of reasons and, on the other hand, concepts that can be employed in empirical description. And we can equate empirical description, as Sellars conceives it, with placing things in the logical space of nature, to coin a phrase that is Sellarsian a t least in spirit. Putting things this way, we can achieve the effect Sellars is after by identifying the logical space of nature with the logical space in which the natural sciences function, as we have been enabled to conceive them by a well-charted, and in itself admirable, development of modern thought. Positively, we can say that to place something in nature, on the relevant conception, is to place it in the realm of law. But what is really important is not such positive characterizations, but the negative point that the relations between elements of nature, on the relevant conception, are not the normative relations that constitute the logical space of reasons. The relations that constitute the logical space of nature on this conception (in the way spatial relations constitute space literally so called) do not include 3"~mpiricismand the Philosophy of Mind", in Herbert Feigl and Michael Scriven, eds., Minnesota Studies i n the Philosophy of Science, vol. 1 (University of Minnesota Press, Minneapolis, 1956), pp. 253-329, at pp. 298-9. 4See p. 257 of "Empiricism and the Philosophy of Mind" for a formulation on these lines.

relations such as one thing's being warranted, or -for the general case- correct, in the light of another. Now, given this framework, which logical space would be the primary home of the concept of experience? Experience, we can plausibly say, is made up of impressions, effects of impacts made by the world on a subject's sensory equipment. Surely talk of impacts of the world on the senses is empirical description; or, to put the point in the variant terms I have introduced, the idea of receiving an impression is the idea of a transaction in nature. On Sellars's principles, then, talk of impressions does not operate in the logical space in which talk of knowledge -or, to keep the general case in view, talk of world-directednessoperates. The logical space in which talk of impressions primarily belongs is not one in which things are connected by relations such as one thing's being warranted or correct in the light of another. So experience, conceived as made up of impressions, cannot serve as a tribunal, something to which empirical thinking is answerable. In fact the idea that it can is exactly what Sellars rejects as the Myth of the Given. I should mention that in Mind and World my main representative of this kind of thinking is Davidson rather than Sellars. Either would have served my purpose. There is a correspondence between Sellars's attack on the Given and Davidson's attack on "the third dogma of empiricism" -the dualism of conceptual scheme and (in a sense other than the one I have used here) empirical content. And Davidson explicitly takes it that the thought dislodges even a minimal empiricism; he claims that the third dogma of empiricism is "perhaps the last, for if we give it up it is not clear that there is anything distinctive left to call empiricism" .5

3. So far I have suggested that we can trace the distinctive anxieties of modern philosophy to a tension between two temptations that our thinking is subject to. One is a minimal empiricism, which links empirical content with the idea that empirical thinking is answerable to the tribunal of experience. The other is a tendency, which I have exploited Sellars in trying to make intelligible, for it to seem impossible that experience could be a tribunal; the idea of experience evidently belongs in a logical space of natural connections, and that can easily be made to seem alien to a logical space in which one thing is warranted or correct in the light of another. 5"On the Very Idea o f a Conceptual Scheme", in Donald Davidson, Inquiries into Truth and Interpretation (Clarendon Press, Oxford, 1984), pp. 183-98, at p. 189.

22. PRBCIS OF M I N D A N D W O R L D

235

The exorcism I have projected would require resolving the tension, and the description I have given leaves various options for doing that. I shall describe three. One course, which Sellars and Davidson follow, is to discard the linkage of empirical content with answerability to impressions; that is, to renounce minimal empiricism, at least with experience construed in terms of impressions. But I do not believe a position on these lines can be genuinely comfortable. It is true that the SellarsDavidson option leaves room for minimal empiricism on a different construal; according to the principles that govern this option, though empirical thinking cannot be answerable to impressions, it can be answerable to appearings. But if there is no such thing as answerability to impressions, I think that ought to make it just as problematic how appearings are possible as how any other mode of possession of empirical content is possible. Sellars and Davidson make it seem that the tension amounts to an incompatibility by insisting that the logical space of reasons is sui generis, as compared with the logical space in which Sellars sees "empirical description" as functioning, which I have identified on Sellars's behalf with the logical space of n a t ~ r e .This ~ points to a second possible way to resolve the tension: namely rejecting that insistence. On this second option, we are to accept that the concept of experience has its primary home in the logical space of nature, conceived in the way that figures in Sellars and Davidson on the other side of a contrast with the logical space of reasons. But on this option, we are to reject the contrast; we are to deny that the logical space of reasons is sui generis in the way Sellars and Davidson claim. This denial is what figures in Mind and World as "bald naturalism". Bald naturalism is a programmatic conviction to this effect: the normative relations that constitute the logical space of reasons can be reconstructed out of conceptual materials whose primary home is the logical space in which empirical description, as Sellars conceives it, functions. (The label "naturalism" is appropriate just to the extent that it is appropriate to identify that logical space as the logical space of nature.) Suppose the bald naturalist programme were executed. That would vindicate moves of the kind that, according to Sellars, commit a naturalistic fallacy in epistemology, and it would vindicate

avidso son's counterpart to what figures in Sellars as the sui genen's character of the logical space of reasons is the sui generis character of what Davidson calls "the constitutive ideal of rationality". See especially "Mental Events", in Donald Davidson, Essays o n Actions a n d Events (Clarendon Press, Oxford, 1980), pp. 207-25; the quoted phrase is from p. 223.

counterparts to those moves in reflection about world-directedness in general. In particular, it would vindicate talk of answerability to experience, even conceived in terms of impressions. Such talk would be naturalistic, by all means, but an execution of the bald naturalist programme would undermine the imputation of a naturalistic fallacy, perpetrated by combining the naturalistic idea of impressions with the normative idea of answerability. I shall say a little more about bald naturalism in a moment, but first I want to outline the different way of resolving the tension that I recommend, the third of the three options I undertook to distinguish. My alternative aims at the same overall effect bald naturalism would achieve: that, without fear of a naturalistic fallacy, we can understand empirical thinking as answerable to experience, even conceived in terms of impressions, impacts from the world on a subject's receptive capacities. But I aim to make room for this without denying, as bald naturalism does, that the structure of the logical space of reasons is sui generis, as compared with the logical space within which the natural sciences confer their distinctive kind of intelligibility on things. The modern scientific revolution made possible a clear conception of that distinctive kind of intelligibility, with the clarity consisting largely, I claim, in something close to Sellars's basic or structural thought: namely, an understanding that natural-scientific intelligibility must be held separate from the kind of intelligibility something acquires when we place it in relation to other things in the logical space of reasons. But we do not need to equate the very idea of nature or the natural with the idea of instantiations of concepts whose primary home is the logical space in which natural-scientific intelligibility emerges. Sellars is right, then, that there is a logical space that is alien to the logical space of reasons; indeed that is a fundamental insight, a prime lesson from the development of modern science. But to equate that logical space with the logical space of nature, as Sellars at least implicitly does, is to forget that nature includes second nature. The natural, in the sense of second nature, embraces concepts that function in the logical space of reasons, sui generis though that logical space is. This makes it unthreatening to acknowledge that the idea of receiving an impression is the idea of a transaction in nature. If we resist the conception of nature that is implicit in Sellars, we undermine the inference to the Sellarsian conclusion, that the idea of receiving an impression is foreign to the logical space in which concepts like that of answerability function. We make room for impressions to be appearings, even as Sellars conceives appearings. Conceptual capacities, talk of which belongs in

22. PRECISOF M I N D A N D W O R L D

237

the sui generis logical space of reasons, can be operative not just in judgements -results of a subject's actively making up her mind about something- but already in the transactions in nature that are constituted by the world's impacts on the receptive propensities of a subject who possesses the relevant concepts. Receiving impressions can be a matter of being open to the way things manifestly are, and that yields a satisfying interpretation for the image of postures that are answerable to the world through being answerable to experience. 4. In Mind and World, I concern myself with bald naturalism only as a potential competitor with the outlook I have just sketched, in the project of exorcizing certain philosophical anxieties. I align myself with bald naturalism in the conviction that the project is a good one. Philosophy is forced into some peculiar and unattractive shapes, if we suppose it needs to answer the questions that express those anxieties -which I have suggested we can collect into the question "How is empirical content possible?". It would be better not to seem obliged to engage in that familiar activity, and that yields a philosophical motivation for bald naturalism that, as far as it goes, I respect.7 But my alternative seems to me a more satisfying response to that motivation than faith in the bald naturalist programme -faith that we can domesticate the logical space of reasons within the logical space in which the natural sciences confer their distinctive kind of intelligibility on things. I have tried to make it plausible that what underlies the anxieties is a sense --often no doubt only inchoatethat the structure of the logical space of reasons is sui generis, as compared with the logical framework in which natural-scientific understanding is achieved. This allows us to see it as non-accidental that the period in which dealing with these supposed difficulties came to seem the dominant obligation of philosophy coincides with the period in which natural-scientific understanding, as we are now equipped to conceive it, was being separated out from a hitherto undifferentiated conception of understanding in general. I claim that the separation was effected largely by way of an increasingly firm grasp of what is in effect Sellars's basic structural insight: that natural-scientific understanding must be held separate from the kind 7 ~ h label e "bald naturalism" is perhaps infelicitous for a position with a sophisticated motivation on these lines; that is what I acknowledge in the footnote on pp. 88-9. I took myself to be stuck with the label even so, since I had given it a thematic prominence in the lectures of which the book is a version. (For the same reason, I felt unable to accede to a plea for less discriminatory terminology made by Nicholas Rescher.)

of understanding that is achieved by placing what is understood in relation to other things in the logical space of reasons. Now according to bald naturalism, that perhaps inchoate sense of a conceptual divide was simply wrong, and the execution of the programme would reveal it as such. My alternative, by contrast, gives modern philosophy more credit, even while, no less than bald naturalism, it enables us to disown those supposed intellectual obligations. (I also give, in one way, more credit to the development of modern science for the conceptual revolution it effected.) According to me, people who think philosophy must centre on problems about how minds can be in touch with the world are not wrong in the thought they take to pose those problems -the thought that is crystallized in Sellars's thesis that the logical space of reasons is sui generis. They are wrong -and the mistake is quite intelligible only in supposing that if we endorse that thought, we are stuck with the intellectual obligations that modern philosophy characteristically, and unsatisfactorily, sets for itself. I do not pretend to have an argument that the bald naturalist ~ point is rather this: the line programme cannot be e ~ e c u t e d .The of thought I have just indicated undercuts the only motivation I consider in my book for supposing the programme must be feasible. As far as that motivation goes, my attitude to the programme can be, not "I know it cannot be carried through" but rather "Given that the motivation is better fulfilled by a different way of thinking, why bother?" Of course that invites alternative motivations. The invitation might initiate a discussion, which would be potentially open-ended; but this further discussion is not my concern in Mznd and World. However, I shall end by allowing myself a brief foray into it. It is not a contribution to the discussion I have in mind to say something like this: "Natural-scientific truth is the only truth there is."9 That is not an argument for bald naturalism, but a mere profession of scientistic faith, the very thing about which the question of motivation arises. Of course we know better than to believe in paramatter as the stuff of minds, and we know better than to suppose ' ~ e r r yFodor insinuates t h a t I aim t o give "the unwary reader" this impression, in his review of Mznd and World: "Encounters with Trees", London Review of Books, 20 April 1995. 'As Fodor in effect does. Outlining t h e position he wants t o defend against me, he writes: "All t h a t ever happens, our being rational included, is t h e conformity of natural things t o natural laws." And later: "[Ilf it's literally true t h a t rationality, intentionality, normativity and t h e like belong t o t h e mind essentially, then they must all be phenomena within t h e natural realm t h a t scientists explore."

22. PRECISOF M I N D A N D WORLD

239

that para-physical forces -forces not needed for explaining "merely material", or perhaps "merely biological", phenomena- are operative at the level of the material constitution of organisms with minds. Something on those lines seems to me to be unquestionable. It sets the agenda for a fine branch of science: to make the material constitution of living things with minds perspicuous, so as to render it intelligible that their lives exemplify mindedness. But it goes no distance towards showing that, if we are to give due honour to the way science has freed us from superstition, we must embrace the conceptual monism -at least about concepts that pull their weight in describing reality- on which bald naturalism insists.


Perception and Rational Constraint: McDowell's Mind and World Bob Brandom

John McDowell's Mind and World* takes on some of the same explanatory challenges as C.I. Lewis' 1929 classic, Mind and the World Order, published 65 years before. Both seek to explain how, in the Kantian terminology they appropriate, intellectual spontaneity -our capacity to deploy richly inferentially articulated concepts- is constained in empirical thought by sensuous receptivity, what Lewis called the Given element in experience. Though by the end of the book his concerns have expanded far beyond their beginnings, to encompass reconceptualizations of nature, selves, and moral sensibility, the focus of the first half of the book, which motivates the approach to those wider issues, is firmly centered on the nature of perception. That is the portion of his project I will discuss here. The prkcis of his views McDowell has offered here stays at a pretty high level of abstraction. He does not recount the specific features of his account of perception that I want to raise questions about. So I'll have to do a fair bit of exposition, with all its attendant dangers of misrepresentation. 'Harvard University Press, Cambridge, Mass., 1994. In what follows, numbers inside brackets refer t o pages in this book. [N. Ed.]

McDowell calls 'minimal empiricism' the view that "the world's verdict, to which thinking must be answerable at all if it is to be thinking a t all, is delivered in the guise of a pronouncement from 'the tribunal of experience"'. This is unexceptionable if it just means that perceptual experience is the medium through which our thought becomes answerable to how things are. No-one, I suppose, denies that noninferentially elicited perceptual judgments are the first tribunal rendering verdicts on our empirical thought --verdicts, to be sure, that can be appealed to a higher court, perhaps ultimately the court of overall explanatory power or our total belief system. But McDowell's view is more specific and more contentious. He thinks we need a notion of conscious experience that is prejudgmental, but nonetheless through and through conceptually contentful, and so capable not merely of prompting perceptual judgments, but actually of pronouncing on them. Experience as he conceives it is always propositionally contentful: experience that things are thus and so. As such it is the right shape to stand in literally evidential relations to the perceptual judgments it prompts, providing the perceiver's reason for the perceptual judgment. I a m not going to claim that this is wrong. For all I know, its where we ought to be at the end of the day. But there is a lot more to it than just minimal empiricism, and I want to suggest that McDowell's arguments in the book actually motivate only something much broader and weaker than the position he in fact adopts.

McDowell describes three broad ways in which our thinking about our thinking in this area can go wrong. His own recommended idiom then emerges from his diagnosis of their failings. The first way to go wrong is what he calls "bald naturalism". By this he does not mean "an unreflective scientism" [89n.], but a more sophisticated view that "aims to domesticate conceptual capacities within nature conceived as the realm of law" [73] in order to defuse apparent philosophical difficulties concerning the place of reason in nature. It's strategy is to "leave unquestioned the conception of nature that threatens to extrude reason from nature, but to reconceive reason in naturalistic terms" [log]. He is surely right to see naturalism so conceived as a widespread current in contemporary philosophical thought.

23. PERCEPTION A N D RATIONAL CONSTRAINT 243

McDowel17s objections to this sort of naturalism,' which he has told us he does not take to amount to a refutation, are harder to characterize briefly. Generically, they take the form of the accusation that bald naturalists ignore important features of spontaneity -precisely those features that make it hard to understand how such spontaneity or rationality could be incorporated into a naturalistic world picture. McDowell is concerned at once to acknowledge such features, and to begin to explain how nature must be reconceived so as to overcome the difficulties. Bald naturalism, he complains, "does not address the philosophical worries I have been considering, but simply refuses to feel them" [108]. It "opts out of this area of philosophy altogether, by denying that the spontaneity of the understanding is sui generis in the way suggested by the link to freedom" [67]. To understand the shortcomings McDowel1 sees in bald naturalism, then, we must understand this final dark phrase, by which he specifies the content of their mistaken denial -that is, what it is about spontaneity that such naturalism

overlook^.^ The second general way in which we can go wrong in our thinking about empirical thought is to fall into coherentism. By this McDowell means a view according to which "the world's impacts on our senses have nothing to do with justification" [15], so that "experience can be nothing but an extra-conceptual impact on sensibility" [14]. Here, by contrast to bald naturalism, the role of spontaneity is acknowledged, but receptivity is shorted by being disconnected from spontaneity. Specifically, such accounts (McDowell instances Davidson, Sellars, and Rorty) "depict our empirical thinking as engaged with no rational constraint, but only causal influence, from outside" [14]. Experience is taken not to justify belief, but only to cause it, be-

'

L'This sort" because the approach he endorses is also a sort of naturalism, "relaxed naturalism". It depends on reconceiving nature as extending beyond the realm of laws to encompass the sort of "second nature" human animals acquire by being enculturated -by coming to move in the space of reasons. 'That a particular idiom "opts out of this area of philosophy altogether" may seem a peculiar objection for a methodological admirer of the later Wittgenstein to put forward. (Or indeed, those of other orientations: I myself would not be chastened by such a description of my attitude toward Trinitarian controversies.) But McDowell's point is that there is substantial philosophical work that must be done in order to entitle oneself to ignore various apparent difficulties. Bald naturalism does not have that work behind it. It is "not a principled avoidance of unprofitable philosophy, but a way of thinking that does not explicitly appreciate what threatens to lead to it. Perhaps people who think like this should be congratulated on their immunity, but it ought not to be mistaken for an intellectual achievement". [89n.]

cause, as Davidson has it "nothing can count as a reason for holding a belief except another belief".3 Here the challenge is to understand what McDowell means by insisting that the world exert rational, and not merely causal constraint on thought by means of our exercise of our receptive capacities in perception, and why he thinks only that sort of constraint will do.4 T h e form of his objection is clear: unless our picture of perception includes such rational constraint from without, "we cannot make it intelligible to ourselves how exercises of spontaneity can represent the world at all" [17]. The issue is "whether the picture can accommodate the sort of bearing on reality that empirical content amounts to" [14]. Now everyone would surely admit that, as Lewis puts it "There is, in all experience, that element which we are aware that we do not create by thinking and cannot, in general, displace or alter" [MWO 481, that "If there be no datum given to the mind, then knowledge must be contentless and arbitrary; there would be nothing which it must be true to" [MWO 381. We must avoid commitment to a picture of thought as, in McDowell's image, "frictionless spinning in the void" .5 The question is rather why and in what sense the necessary constraint must be "in experience" or "given to the mind" in a way that is sensibly talked about in terms of "rational constraint" on thought. After all, perceptual judgments are the very paradigms of judgments that arise noninferentially. T h e third way in which we can go wrong in our thinking about empirical thought is to commit ourselves to the Myth of the Given. T h e Myth is what we get if, seeing what is wrong with bald naturalism and coherentism, we do aim "to cast experience as a rational constraint on empirical thinking" [67], but forget that "a bare pres3 ' L Coherence ~ Theory of Truth and Knowledge", reprinted in Ernest LePore, ed., Truth and Interpretation: Perspectives on the Philosophy of Donald Davidson (Basil Blackwell, Oxford, 1986), p. 310. 4 ~ h i isnsistence is the oriein of his central claim t h a t r e c e ~ t i v i t vmust be " understood as bringing into play t h e same conceptual capacities exercised in the rational activity of pure spontaneity. 5e.g. a t [18]. Talk of 'friction' is somewhat peculiar imagery here; for surely friction is a causal process if any is. And t h e whole point t h e image addresses is t h a t what is required is rational friction, not causal friction. But what is t h a t ? T h e image is of little help, and is prone t o mislead. (Notice t h a t this complaint is about the use of this specific image, not about the appeal t o philosophical imagery in general. It is true t h a t imagery is always in danger of misleading or being misunderstood, but it is a fantasy t o think one can eliminate that danger wholesale by making everything explicit a t once. Retail removal of particular dangers is the best either strategy is capable of.)


ence cannot be a ground for anything" [19].~Thus for instance when Quine says things like "The stimulation of his sensory receptors is all the evidence anybody has to go on, ultimately, in arriving at his picture of the world", and talks about these sensory stimulations as the "data from which we posit bodies and project physics", or that that "Roughly specifiable sequences of nerve hits can confirm us in statements about having breakfast, or there being a brick house on Elm Street.. . ".7 we can see a confusion of the matter-of-factual order of causation and the normative order of justification, evidence, and inference. The myth is the idea that one can coherently talk of experience both as not itself having a conceptual content and yet also as exercising the appropriate sort of rational constraint on empirical judgment.8

McDowell recommends his own view as the only way to avoid these three unsatisfactory alternatives. What makes them all unsatisfactory, he argues, is that they all fail to give an adequate account of the sense in which perceptual experience embodies the way the world imposes rational, and not merely causal constraints on thought. The key to his botanization of alternatives, and so to the motivation of his own approach, then, clearly lies in what might be called McDowell's "rational constraint constraint" on theories of perception (and so of empirical conceptual content). Meeting that constraint is a criterion of adequacy, he is claiming, for giving due weight both to the dimension of receptivity (in virtue of which thought must be constrained from without) and to the dimension of spontaneity (in virtue of which the constraint must be rational) that are essential aspects of the contents of empirical thoughts. It may help to focus the issue at this point to ask what, specifically, goes wrong on this line with, say, externalisms of the Davidsonian or reliabilist variety, which seek precisely to appeal to relations in the causal order to ground the contentfulness of empirical claims? Why exactly is it

or

this reason, McDowell takes the target of Wittgenstein's private language argument t o be a form of the Myth of the Given. [20-231. 7 6' Epistemology Naturalized", p. 75 in Ontological Relativity and Other Essays [Columbia University Press, New York, 19691; "Posits and Reality", p. 240 in Ways of Paradox and Other Essays [Random House, New York, 19691. Emphasis added. 'Gareth Evans is discussed as falling into this error, in Lecture 111.

that in such terms "we cannot make it intelligible to ourselves how exercises of spontaneity can represent the world a t all" [17]? One clue emerges in a discussion of Davidson's renunciation of rational control from independent reality. He thinks a merely causal, not rational linkage between thinking and independent reality will do, as an interpretation of the idea that empirical content requires friction against something external to thinking. But it will not do. Thoughts without intuitions would be empty, as Kant almost says; and if we are to avert the threat of emptiness, we need to see intuitions as standing in rational relations to what we should think, not just in causal relations to what we do think. Otherwise the very idea of what we think goes missing. [68, emphasis added]. T h a t is, in order to underwrite a n intelligible notion of empirical conceptual content, the constraint on our thought exercised by the world in perception must be normative: it must settle how it would be correct to apply the concepts in question, how they ought to be applied. The concept of conceptual or intentional content, like the concept of meaning, is a normative concept. A second point is then that not just any norms will do. Specifically conceptual norms govern our exercises of spontaneity in understanding ("'Spontaneity' here can be simply a label for the involvement of conceptual capacities" [9]). And "the topography of the conceptual sphere is constituted by rational relations. The space of concepts is a t least part of what Wilfrid Sellars calls 'the space of reasons"' [5]. The contents of concepts are articulated by the inferential relations among them in virtue of which one judgment can serve as a reason for or against another. The rational activity of giving and asking for reasons in turn makes sense only in a context in which we are responsible for our conceptually contentful commitments -a context in which the thinker's entitlement to those commitments is always in principle at issue. It is essential to conceptual capacities.. . that they can be exploited in active thinking, thinking that is open to reflection about its own rational credentials. When I say the content of experience is conceptual, that is what I mean by 'conceptual'. [47] Conceptual capacities.. . belong to a network of capacities for active thought, a network that rationally governs comprehension-seeking responses to the impacts of the world on sensibility. And part of the point of the idea that the understanding is a faculty of spontaneity -that conceptual capacities are capacities whose exercise is in the domain of

23. PERCEPTION AND RATIONAL CONSTRAINT 247

responsible freedom- is that the network, as an individual finds it governing her thinking, is not sacrosanct. Active empirical thinking takes place under a standing obligation to reflect about the credentials of the putatively rational linkages that govern it. [12]

A third claim of McDowell's is then that the representational norms that connect the correctness of our thought t o the facts, t o how things are with that bit of the world the thought therefore counts as being about, must be understood as aspects of the rational norms that govern this process of active critical reflection on credentials: In 'outer experience', a subject is passively saddled with conceptual contents, drawing into operation capacities seamlessly integrated into a conceptual repertoire that she employs in the continuing activity of adjusting her world-view, so as to enable it to pass a scrutiny of its rational credentials. It is this integration that makes it possible for us to conceive experience as awareness, or at least seeming awareness, of a reality independent of experience. [31] T h e connection is that subjecting the deliverances of experience t o rational scrutiny, engaging them critically with reasons for and against them drawn from all our beliefs, has the effect that: The object of experience is understood as integrated into a wider reality, in a way that mirrors how the relevant concepts are integrated into the repertoire of spontaneity at large ...This integration allows us to understand an experience as awareness of something independent of the experience itself.. . [32] Intuitions without concepts, that is, apart from their engagement in the rational enterprise of giving and asking for reasons, are blind. And To say that an experience is not blind is to say that it is intelligible to its subject as purporting to be awareness of a feature of objective reality: as a seeming glimpse of the world. [54] McDowell is covering a lot of ground with these three ideas. They are none of them self-evident, they all could be worked out in more than one way, and we are given very little in the way of details. At this level of his argument, McDowell contents himself with making his commitments explicit in passages such as those I have cited, without showing just how he would propose t o show himself entitled t o them. One would, I think, particularly like to know how the last point could be filled in -just how reference to what we are thinking

about precipitates out of (or is intelligible as expressing a n aspect of) rational relations among various things we might think (or beliefs we might have), as it turns out, about what we are thinking about. But I propose that we grant McDowell these three claims: about the normativeness of representational relations, about the rational or inferential articulation of concepts on which the critical assessment of their credentials depends as essential to the contents of those concepts, and about the dependence of the idea of a state's so much as seeming to be of or about some objective feature of the world (in the sense of answering to that feature for its correctness as a representing) -its glimpselikeness- on the liability of that state to critical examination of its rational credentials when confronted with other states related to it as reasons for or against it. I think they are all three both true and important, and I think that the promissory notes concerning our entitlement to them that McDowell is implicitly offering can be redeemed.g My present point concerns not the antecedents of these ideas, but their consequences. For together, they both give a definite sense to the rational constraint constraint, and set criteria of adequacy that any account must meet in order to count as satisfying that constraint. They entail that to qualify as offering even purported glimpses of how things are, perceptual experiences must have rational credentials that are available for critical scrutiny. If they are insulated from the facts they purport t o report, in the sense that how things actually are does not have a rational bearing on them, then they cannot qualify as being of or about anything (and so having empirical conceptual content) a t all.

. . .in the context of the insulation from spontaneity, the talk of concepts is mere word-play. The point of the claim that experience involves conceptual capacities is that it enables us to credit experiences with a rational bearing on empirical thinking. But the point of the strategy of insulation is that it confines spontaneity within a boundary that leaves experiences outside it. That means that the putatively rational relations between experiences, which this position does not conceive as operations of spontaneity, and judgments, which it does conceive as operations of spontaneity, cannot themselves be within the scope of spontaneity liable to revision, if that were to be what the self-scrutiny of active thinking recommends. And that means that we cannot genuinely recognize the relations as potentially reason-constituting. [52] 'I have explained in detail how I think it can be done, in Making It Ezplicit: Reasonang, Representing, and Discursive Commitment [Harvard University Press, 19941.


The idea seems to be that critical scrutiny of the rational credentials of perceptual judgments requires more than just that as a matter of fact they have arisen as the result of matter-of-factual dispositions linking them as responses to environing stimuli. Their status as putative glimpses of how things are outside and independently of the perceiver is a claim of a certain kind of authority for them: the authority to say how things actually are. But, McDowell is claiming, we can recognize that sort of cognitive authority only where its rational credentials as authority the judgment is entitled to are available for critical inspection. Perceptual judgment cannot in the end be warranted simply by its origin in a disposition; causal constraint is not enough.

But is it really the case that externalisms of the reliabilist and Davidsonian sorts cannot satisfy these requirements? Reliabilism in the epistemology of perception is a view that McDowell will classify as a bald naturalism, while Davidson's view, which permits only causal and not conceptual intermediaries between facts and judgments, he classifies as a coherentism. Yet it seems to me that each of them can be understood as meeting the rational constraint constraint, at least as unpacked above. Consider first reliabilism. It is important at the outset to distinguish between two different sorts of views that share that title. Weak, or epistemological reliabilism, claims that the role traditionally played by a justification condition on attributions of knowledge can be played by a reliability condition: to count as knowledge, true perceptual beliefs must be the outcome of reliable belief-forming processes, that is, processes that are (in specified circumstances that include the actual circumstances of belief formation) likely to lead to true beliefs. Perceptual beliefs so formed are ones the perceiver should count as entitled to, in the sense relevant to assessments of knowledge.10 Such a view takes for granted the contents of beliefs or judgments that are candidates for empirical knowledge, and undertakes theoretical commitments only regarding how it is possible for the believer to become entitled to those contents. The claim is that ''Of course there are many varieties of epistemological reliabilism, and many would not restrict the thesis to perceptual beliefs. But for our purposes here, it suffices to consider this committively minimal variety: perceptual epistemological reliabilism, as it might be called.

in at least some circumstances in which attributions of knowledge are made and assessed, the role traditionally played by the candidate knower's capacity to offer reasons, to cite other beliefs that stand in appropriate inferential relations to the one in question, can be played by the fact of being a reliable reporter or belief-former. Strong, or semantic reliabilism, by contrast, claims that the role traditionally played by justification, by inference, by rational relations among beliefs, in the articulation of the conceptual contents of beliefs can be understood instead entirely in terms of reliability. The inferential relation two beliefs stand in when one is a good reason for the other is just an abstraction from one species of reliable belief-formation. Giving and asking for reasons plays no essential role in the constitution of conceptual contents. This stronger view is clearly incompatible with McDowell's understanding of the spontaneity of the understanding. There are good reasons to think that one cannot distinguish what is conceptually contentful from what is not without appeal to the specifically inferential articulation of the conceptual. A parrot can be trained to respond to visibly present red things by saying "That's red", and we can suppose that its differential responsive dispositions to do so match those of a well-trained human reporter of red things. But the parrot will not be reporting the presence of red things, and will not be expressing a perceptual judgment, by its responsive sounding-off. For its reliably differentially elicited response is only to make the noise that corresponds to our word 'red', not to apply our concept &. It will not because although its response can mean something to US, it does not mean anything to the parrot. In particular, in making its response the parrot is not taking up a stance in the space of reasons, is not undertaking a commitment that can serve as a reason for or against others and whose entitlement can itself be brought into question. Only what is inferentially articulated, only what plays a suitable role in the game of giving and asking for reasons, can engage critically with other claims and beliefs in the way required for it to have specifically conceptual content t o be, in McDowell's sense, an exercise of spontaneity. Extruding practices of justification -the giving and asking for reasons- from the picture involves extruding the conceptual as well. Strong or semantic reliabilism debars itself from understanding specifically conceptual content. This realization is part of what lies behind McDowell's repeated insistence1' that the issue about rational constraint does not have to do just with our assessments of entitlement as they bear on 11

e.g. at pp. 31, 32, 42-43, 46.


attributions of knowledge, but on the very intelligibility of the idea of empirical conceptual content. Thus semantic reliabilism is false: one cannot replace the notion of inferential relations wholesale with that of reliable signalling, while retaining an intelligible notion of conceptual content. But semantic reliabilism is also not a consequence of epistemological reliabilism; the sins of the former should not be visited on the latter. One can perfectly well acknowledge that someone may count as entitled to a belief or judgment (in the sense that matters for assessments of knowledge) without being able to offer a reason or justification for it, while insisting nonetheless that it can be thought of as a belief or judgment at all only insofar as it can serve as both a premise and a conclusion of inferences, and so is liable to critical assessment on the basis of its relation to other beliefs and judgments. On this picture of the concept-use that defines exercises of spontaneity, there can be noninferentially elicited judgments, and they may express perceptual knowledge even if the knower cannot justify them, so long as the knower is sufficiently reliable (according to the one assessing the knowledge claim). But there could not be an autonomous discursive practice --one that could be that could be engaged in though no others were- that consisted entirely of noninferential reports. Beliefs, judgments, and claims that have noninferential uses must also have inferential uses in order to qualify as having conceptual content. And this point about the essential involvement of spontaneity in the deliverances of receptivity is entirely compatible with weak or epistemological reliabilism. It seems to me that epistemological reliabilism about perceptual beliefs fully meets McDowel17s demand for the rational criticizability of such beliefs. Assessing reliability is one way of assessing a believer's entitlement to beliefs she finds herself with perceptually. This is as normative a matter as can be -a matter of what one should think, and not just what one does think (to echo the passage from [68] quoted above). And it is evidently not an assessment that somehow takes place outside the space of reasons. On the contrary, taking someone to be a reliable reporter is just endorsing an inference of a characteristic pattern: taking it that the fact that the reporter is noninferentially disposed (in the right sort of circumstances) to acquire a belief with a given content provides a good reason for acquiring a belief with that content oneself (a reason the attributor of reliability could cite in justifying his own belief, even if the one whose belief it is could not). That reliability (in specified circumstances) as a reporter is likelihood of truth (in those circumstances) of reports ensures that the connection envisaged by reliabilists be-

tween reported facts and reports of them is not merely causal, but also rational. The same seems to me to hold for semantic externalism of the Davidsonian sort. The interpreter is responsible for characterizing the perceptible environing facts to which the one interpreted is taken to be responding, and is charged with attributing propositional contents to the sentential responses that make the one interpreted largely correct in her noninferential reports (subject, of course to such adjustment as is necessary in order to make sense also of other attributed noninferential and inferential dispositions and commitments). The interpreter assesses the adequacy of the content of the reports and perceptual beliefs that are noninferentially wrenched from a subject in virtue of her wiring and training, by comparing it with the facts that responsively elicited those reports, according to the norm that one ought to say of what perceptibly is that it is. Perceptual reports that do not measure up to this standard are criticizable precisely on these grounds. From the point of view of the interpreter, in other words, the relation between the facts and the reports or perceptual beliefs is not merely a causal one, but also one rationally assessable in terms of the truth of those reports or beliefs, relative to the independent facts.12

Keeping these observations in mind, we can then ask why it is that McDowell sees only causal, and not rational constraint as operative in these two cases? Why does he see the epistemological reliabilist as ignoring reasons (and spontaneity) entirely (a charge, admittedly, that can be leveled with some justification a t semantic reliabilists), and the Davidsonian externalist as a coherentist, doomed as such to leave unintelligible the crucial dimension of empirical conceptual content in virtue of which they are rationally answerable for their correctness to how things in fact are? The answer is not far to seek if we look a t the view that McDowell recommends as necessary and sufficient to avoid falling into one or another of the pits of oblivious naturalism, frictionless coherentism, or (what is not really at issue in the present discussion) the Myth of the Given. l21n t h e first half of Chapter Four of Making It Explicit ( o p . cit.) I present a detailed account of how I see semantic externalism and epistemological reliabilism as highlighting crucial features of t h e essentially social practices of giving and asking for reasons.

23. PERCEPTION AND RATIONAL CONSTRAINT 253 For there is a slide in the move from McDowell's diagnosis to his recommended therapy -a slide that seems symptomatic of a specific sort of blind spot in his thinking. What emerges from the diagnosis, I have been urging, is epitomized in such claims as these: The world itself must exert a rational constraint on our thinking. If we suppose that rational answerability lapses at some outermost point of the space of reasons, short of the world itself, our picture ceases to depict anything recognizable as empirical judgment. [43, emphasis added] I am trying to describe a way of maintaining that in experience the world exerts a rational influence on our thinking. [34, emphasis added]

The sort of rational constraint that really matters is constraint by the facts, by what is perceived. It is rational answerability for their correctness to these that is required to fund them as empirically conceptually contentful in the sense of presenting at least putative glimpses of how things are independently. And the point of rational criticism is precisely to fund this notion of answerability to how things in any case are. But in his positive suggestions, McDowell looks to rational constraint, not by the facts, but by experience of the facts. He offers a way of thinking about perceptual experiences as conceptually articulated, and hence able t o stand in rational, and not merely causal relations to perceptual judgments, while still distinguishing them from perceptual judgments themselves. I t is not my purpose to object to the account he gives of perceptual experiences. As far as I can see, that account is entirely coherent and workable. What I do want to question is whether the arguments he offers against the three alternatives canvassed above oblige us to adopt that view. Its most promising competitor, it seems to me, is a view along the lines laid out by Sellars and Davidson. According t o such a story, when we are properly wired up and trained, and favorable circumstances, the perceptible facts wring from us perceptual judgments. In order to explain how this is possible q u i t e a different enterprise from justzfying the resulting judgments- we postulate the existence of something like sense impressions,13 whose properties systematically covary with the contents of the judgments they causally elicit 1 3 ~ h i iss Sellars' terminology, in "Empiricism and the Philosophy of Mind", where the distinction between what is relevant t o epistemic justification (and so conferral of semantic content) is rigorously distinguished from what is relevant t o the quite different enterprise of explaining in scientific terms how it is possible for us t o respond differentially t o just the range of phenomena we turn out t o be capable of reporting. (Though t h a t enterprise cannot explain why our response counts as a report of something.)

from us. But these sense impressions are features of the physiology of perception. They are not something we are aware of, and they do not themselves have conceptual content. They merely occasion contentful judgments. A perceptual judgment is one that is noninferentially elicited as a response to a sense impression. Experience (globally, not necessarily in each particular case) has the two dimension of sense iinpressions and perceptual judgments, whose relation is rationally controlled by the facts perceived and reported. If a suitable story is told about the constitutive inferential engagement of these noninferentially elicited judgments with other judgments, then their essential liability t o rational criticism -and hence status as denizens of the space of reasons and products of s p o n t a n e i t y is secured. Bald naturalism is avoided. If a suitable story is told about how they are rationally criticizable by those who key their correctness to their correspondence to the facts reported, and their entitlement to the reliability of the noninferential process that elicits them -a matter of the assessors' rational willingness themselves to endorse respectively the claim in question and the inference from the reporter's making of it to that endorsement themselves- then rational constraint by and answerability of perceptual judgment to how things actually are is secured. Coherentism is avoided. And nothing nonconceptual -no mere presence lacking the shape of the fact, claim, judgment, or belief that things are thus and sol4- is every appealed to in order to justify or serve as a reason for a claim. Sense impressions play only a n explanatory, not a justificatory role, so the Myth of the Given is avoided. This is precisely a n account of how in experience (that is, perception) the world exerts a rational (criticizable) influence on our thinking, where rational answerability does not stop anywhere short of the world itself -just what is demanded in the last two passages cited. But McDowell does not see this possibility. . . . in order to escape from the oscillation we need to recognize that experiences themselves are states or occurrences that inextricably combine receptivity and spontaneity. We must not suppose that spontaneity first figures only in judgments. [24, emphasis added] These experiences are understood as themselves conceptually structured in a way that makes them fit to serve as just2fications of the perceptual judgments they elicit. When we trace the ground for an empirical judgment, the last step takes us to experiences. [lo]


And again: Suppose.. . there is an inclination to apply some concept in judgment. This inclination does not just inexplicably set in. If one does make a judgment, it is wrung from one by the experience, which serves as one's reason for the judgment. In a picture in which all there is behind the judgment is a disposition to make it, the experience itself goes missing. [61] This is indeed one way to escape the oscillation (between coherentism and the Myth of the Given, once we have committed ourselves to spontaneity being genuinely in the picture), but we do not need to go this way. The move from the need to escape the oscillation to the need for conceptually structured prejudgmental experiences that warrant our perceptual judgments is a non sequitur -the 'need' and 'must' of the first passage are unwarranted.15 What we are not told is what justifies the move from the need for rational constraint by the world, as invoked in the passage quoted three paragraphs back, to rational constraint by experience.

Now one might question whether the views I have considered here actually do secure rational constraint by the facts. Addressing that issue would take us too far afield for the current discussion. But it should be clear in any case that the objections McD has raised against bald naturalism, coherentism, and the Myth of the Given do not sufficiently motivate his own view. There are other ways in which those considerations could be met. Some further argument is needed that a notion of conceptually articulated but prejudgmental experience is required. Merely causally covarying sense impressions (part of the medium of our sensory awareness, not its objects or contents) will do, if rational constraint by the facts is all that is needed. There may be such an argument, and so a reason to prefer McDowell's approach. But he has not told us what it is. If this is right, then we should ask, in a quite McDowellian spirit: "What is it about his approach that systematically blinds the author of Mind and World to the possibility of the alternative approach indicated above?" Here again, I think, an answer is not far to seek. " ~ ~ a i nI ,am not claiming there is something wrong with the view McDowell recommends, only that the considerations he advances do not require us to adopt it.

McDowell systematically underplays the significance of the social dimension of the practice of giving and asking for reasons -justifying, challenging, criticizing, and revising our claims and beliefs- that is the concrete embodiment of the aspect of our activity he talks about abstractly under the rubric of 'spontaneity'. C.I. Lewis, the author of Mind and the World Order, used the singular 'mind' instead of the plural 'minds' in his title deliberately, and with his eyes wide open. His is a story of a single mind confronting an alien reality. Cooperation with its fellow minds is simply not in the picture. The author of Mind and World does not share those explicit commitments. The social nature of spontaneity and the space of reasons is acknowledged, but only belatedly, in the discussion of the need for knowers and agents to be properly brought up in order to be sensitive to various sorts of norms. I think that the slide I have accused McDowell of is a manifestation of his falling prey to the bad consequences of thinking only of Mind, instead of minds, as knowing and acting in the world. For in the case of Davidsonian and reliabilist externalism, rational criticizability of perceptual judgments may be available in the first instance only from a third-person perspective, from the point of view of someone who is assessing the extent to which a report is true to the facts reported, or the perceiver is reliable, and so entitled to the claims noninferentially wrenched from her by her perceptible environment. But so what? This keeps it from being "selfscrutiny" in the sense of the passage from [52] cited above only if we illegitimately treat the fact that the mind of McDowell's title is grammatically singular as not a mere facon de parler, but as a substantive thesis -as bringing with it the individualist commitments that his predecessor Lewis undoubtedly did undertake. Assessments of truth and reliability are not outside the practice of giving and asking for reasons, scrutinizing and criticizing the rational warrant for our commitments, but of its very essence. No more are they invoking 'merely causal' elements outside the space of reasons. McDowell's own formulation of the rational constraint constraint is that: What we wanted was a reassurance that when we use our concepts in judgment, our freedom --out spontaneity in the exercise of our understanding- is constrained from outside thought, and constrained in a way that we can appeal to in displaying our judgments as justified. [8] The question here is about what it means to insist that we can appeal to what constrains our judgment in assessing its warrant. Given the conclusions he draws, McDowell uses it as if it meant "each of us", so


that the one whose judgment is justified must be the same one who can appeal to the external contraint in justifying it. Only on that assumption is it correct to conclude that "When we trace the ground for an empirical judgment, the last step takes us to experiences", that is, only if the judger herself is the one who must trace the ground. But why assume that? If we take the we seriously in these passages, the one who can see the constraint as justifying the application of the concept may be someone else, someone other than the one who is applying the concept. Epistemological reliabilism and Davidsonian externalism, as I have sketched them, have this latter structure. They do not insist that when there is an inclination to apply some concept in judgment the inclination just inexplicably sets in or that there is nothing behind it that could be the source of its entitlement other than the mere disposition (as in the passage quoted above from [61]). But what stands behind it is not necessarily available at that moment to the one who is inclined noninferentially to make the judgment. One should ask: "Inexplicably to whom?" On no-one's account is it inexplicable to the third-person scientist who knows a sense impression was present. And the judgment can be wrung from one by the sense impression without the sense impression having for that reason to serve as one's reason for the judgment. (The passage continues: "If one does make a judgment, it is wrung from one by the experience, which serves as one's reason for the judgment". What is the argument for the implied relation between the final clause and the first part of this sentence?) Insofar as one has a reason, it is surely that things are thus and so, or, more carefully, that one can see that things are thus and so. 'Behind' here is ambiguous precisely between a causal and a normative sense. Sense impressions are 'behind' the judgments in a causal sense, and facts are behind then in a normative sense (as well as, in the favored cases, in a causal sense). What is the source of the insistence that there must also be some internal thing, the experience, that plays both these roles at once? Surely the most the arguments (from the requirement of making empirical representational content intelligible) settle is that something, what is reported, must play both roles (and not merely the causal one). There are some important premises missing here. It seems to me that they are individualist assumptions.

I have made two main claims about McDowel17sproject. First, there is a gap between his diagnosis and his recommended therapy -a

gap visible in the slide from a reading of the rational constraint constraint as insisting that the world have a rational, and not merely a causal bearing on perceptuals judgment to a reading of it as insisting that experience have a rational, and not merely a causal bearing on perceptual judgments. As a result of this slide, McDowell treats the considerations he has advanced in defense of the rational constraint constraint as obliging us to endorse a notion of conceptually articulated but prejudgmental perceptual experiences. But doing so involves overlooking other alternatives, alternatives that are not in fact ruled out by the considerations he advances. Second, I have suggested that the aetiology of this blindness to alternatives should be traced to a residual individualism in McDowell's thinking about reasoning -a systematic underestimation of the significance of the fact that talk of the space of reasons is an abstraction from concrete, essentially social practices of giving and asking for reasons.16 The first half of McDowell's book urges two large, controversial claims. First, perceptual judgments should be understood as both causally elicited and rationally warranted by conscious perceptual experiences that themselves already involve the application of concepts. ("That things are thus and so is the content of the experience, and it can also be the content of a judgment" [26].) Second, that the conceptual has no outer boundary: T h e conceptual is unbounded; there is nothing outside it. [44] Reality is not located outside a boundary that encloses the conceptual sphere. [41]

This is compatible with genuine external constraint on our thought, because: What we need is constraint from outside thinking and judging, our exercises of spontaneity. The constraint does not need t o be from outside

thinkable contents. [28]

I have not taken issue with either of these claims. Regarding the first, I have claimed only that though that view may be n~otivated by the arguments against the viability of bald naturalism, coherentism, and the Myth of the Given, it is not entailed, necessitated or established by those arguments. The second claim, McDowell's conceptualism, his attempt to "domesticate the rhetoric of Absolute 1 6 1 have explored this suspicion before in regard t o s a n e of McDowell's other writings. See "Knowledge and the Social Articulation of the Space of Reasons" forthcoming in Philosophy and Phenomenological Research.

23. ~ E R C E P T I O NAND RATIONAL CONSTRAINT 259

Idealism" [44, paraphrased] is far too large a topic to address here. I am not in fact disposed to take issue with it either. (This is a defining point of agreement among Pittsburgh neo-Hegelians.) The distinction I have insisted on between rational constraint by the facts and rational constraint by experience is an intramural distinction, neither side of which takes issue with this conceptualist commitment. I think, however, that in the end the way in which paying attention to the social articulation of the practices of giving and asking for reason makes room for the insights of semantic externalists and epistemogical reliabilists provides the key to seeing conceptualism as a sensible and natural idiom for talking about the relation between minds and the world, rather than the craziness it is likely to seem upon first exposure. Thus, although I cannot pursue the point here17, I think there are large and important claims essential to McDowell's overall project that become more, rather than less intelligible and plausible, in the light of the criticisms of his argument offered here.

171 do so in Making It Explicit. See especially t h e first half of Chapter Four, and Chapters Eight and Nine.


Spin Control Comment on John McDowell's M i n d a n d World* Alex Byrne

We have justified beliefs about the "external" world, and some of these are formed directly on the basis of perception. I may justifiably believe that a certain dog is in certain manger, and I may have this belief because I can see that the dog is in the manger. So far, so good. How is it that we can perceive and believe that things are thus and so and, moreoever, be justified in so believing? One of the main concerns of Mind and world' is to answer these questions.2 As might be expected, McDowell's answer is both subtle and provocative: if he 'For helpful discussion I am indebted t o Ned Block, Ned Hall, Jim Pryor, and Robert Stalnaker. 'Harvard University Press, 1994. All page references are t o this book, unless otherwise noted. go his is a rather rough, and perhaps somewhat misleading, guide to McDowell's project. He is not trying t o formulate a theory of intentionality, or justification, in the sense in which, for example, Fodor has a theory of intentionality and Goldman has a theory of justification. More accurate would be: to provide a picture of the mind and its relation t o the world so t h a t it does not seem impossible or mysterious that we perceive and justifiably believe things about the environment in which we live.

is right, then many orthodoxies are houses of cards, built on sand. In this comment, I want to focus on the tantalizing dialectic McDowell sets up at the beginning of the book. It is out of this that the synthesis of his own position emerges, one central strand of which is that experience has only conceptual content. More -although not enough- of this later. In the first lecture, McDowell sets out two answers to the question about justification that he finds unsatisfactory: the Myth of the Given, and a Davidsonian version of coherentism. The inadequacies of the Given, he argues, propel us into the open arms of coherentism, only to discover that its charms are illusory. And so we return shamefacedly, like unfaithful spouses. As is usually the case after infidelities, things are never quite the same, and so begins an "interminable oscillation1'. McDowell's own position is advertised as the way to "dismount from the seesaw" (9). However, I find it hard to understand why dismounting should be necessary, for I cannot see how the promised oscillation is supposed to get started. Both the Myth of the Given and coherentism are theories of justification, not of mental representation. However, McDowell repeatedly claims that either one makes a mystery of how thought can have "empirical ~ o n t e n t " . With ~ a partial exception for coherentism, I do not understand why McDowell thinks this. But for the present discussion, this won't much matter.

1 The Given Let us begin, as McDowell does, with the Given end of the seesaw. When McDowell first introduces the Given, he does so in the standard way. It is something "simply received in experience" (6) that serves as the ultimate epistemic ground for perceptual belief. And my receiving the Given is not my being in a state which has representational content. A fortiori, receiving the Given does not require exercising what McDowell calls "spontaneity" ("a label for the involvement of conceptual capacities" (9)). Now it turns out that this last fact is the really problematic one: that the Given is not inextricably tied to spontaneity. That is enough, McDowell thinks, to show that the idea of the Given is "useless for its purpose" (7). For, on McDowell's view, the only candidates for being justifiers are contentful mental states in which 3e.g. 6, 15, 25, 47, 68, 134, 142. This worry, about how thought can have any bearing on reality, has been a continuing theme in McDowell's writings.

conceptual capacities are "already operative1' (62). As he puts it, amending a similar claim of Davidson's, "nothing can count as a reason for holding a belief except something else that is also in the space of concepts" (143). Evidently the attack on the Given, if successful, will overwhelm a number of more fortified enemy positions, such as any reliabilist theory of justification. And any view that takes experience to be a justifier, but merely assigns to it "non-conceptual" content, succumbs to the same difficulty. In fact, Gareth Evan's view, which is of this latter kind, is classified in lecture I11 as "a version of the Myth of the Given" ( 5 1 ) . ~ According to McDowell, the thought motivating the Given, that impacts from outside must enter the justificatory story, is quite correct. In this he disagrees with Davidson and agrees with Evans. But, as just noted, he agrees with Davidson that all reasons are within the space of concepts. Putting those two together, we get McDowell's claim that experience must have conceptual content. In this first section, I want to question one of McDowell's arguments against the view that experience with non-conceptual content can justify belief. So a few sketchy remarks about the notions of conceptual and non-conceptual content are in order. Possessing the concept SQUARE is to have a capacity, in particular a capacity to judge (think, believe) that objects are square. (Take that as stipulative.) If someone judges that an object is square then she must possess the concept SQUARE: the content of the judgement is conceptual. Now take a visual experience that represents an object to be square -that is veridical only if the object is square. If the subject does not need to possess the concept SQUARE in order to enjoy this kind of visual experience, then the experience has nonconceptual content. This modal test only gives a sufficient condition for an experience to have non-conceptual content; there is little consensus on necessary and sufficient conditions, but we can make do the name, the distinction ~ here with a sufficient ~ o n d i t i o n .Despite between conceptual and non-conceptual content is not, in the first 4 ~ oEvans' r position, see T h e Varieties of Reference (Oxford University Press, 1982). An Evans-style view and reliabilism are of course perfectly compatible. 'Why not say this? The content of a perceptual experience t h a t presents t h e world as thus and so is (wholly) conceptual if and only if t h e capacities that constitute possession of the concepts required t o believe things are thus and so are also (partly) constitutive of t h e having of the experience. (This could only b e a first attempt, for there is no reason t o think that the total content of a perceptual experience could be t h e object of a single belief.) T h e suggestion seems t o be akin t o what McDowell has in mind when he writes:

instance, a distinction between contents -sets of possible worlds, structured propositions, or whatever. Rather, it is distinction between representational -that is, truth-conditional- mental states. It is not mandatory, for someone who believes that the content of perception is non-conceptual, to select different sorts of abstract objects to serve as the contents of perception and belief: the same kind may do duty for both.6 Returning to the Given, McDowell's first way of putting his objection is that such "a brute impact from the exterior" cannot be a reason: the Given merely allows us to be "exempt from blame" (8) for what we believe. Here the worry appears to be that the relation between the Given and belief is "merely causal" (71, fn. 2). That worry he appears to share with Davidson, who remarks in a similar context,

a

It is essential to the picture I am recommending that experience has its content by virtue of drawing into operation, in sensibility, of capacities that are genuinely elements in a faculty of spontaneity. The very same capacities must also be able t o be exercised in judgements, and that requires them to be rationally linked into a whole system of concepts and conceptions within which their possessor engages in a continuing activity of adjusting her thinking to experience (46-7). And: "capacities that belong to spontaneity [are] inextricably implicated in an operation of mere receptivity" (14-5). And: "conceptual capacities.. .are operative also in our perception of the world" (72). (For simplicity I am ignoring McDowell's further claim that conceptual capacities in his sense require the subject to be self-conscious (47, fn. I).) But now, regarding the first formulation, one would like to know how (the possession of) various capacities t o believe and judge such-and-such might constitute the having of the experience in question, if this is not supposed to be the simple modal claim that it is impossible to have the experience without having the capacities. And regarding the second, third and fourth formulations, one would like to know what "drawing into operation", "inextricably implicated", and "operative", amount to, exactly. If there is a language of thought, then we can state the thesis that the content of experience is wholly conceptual very easily: there is a visual experience "box", just as there is a belief box, and all the sentences in the former are written in the language of thought, the language that can appear in the belief box. But this formulation would not be acceptable to McDowell (see his "Putnam on Mind and Meaning", Philosophical Topics 20 (1992), 35-48, esp. section 8). 'Christopher Peacocke, in A S t u d y of Concepts (MIT Press, 1992; hereafter SC), develops the distinction between conceptual and non-conceptual content so that it is not only a distinction between representational mental states, but also between contents. According to Peacocke, conceptual contents are precisely those composed of "concepts" (i.e. Fregean senses). But as I said, this is an optional extra. The main arguments for non-conceptual content (e.g. that the content of perception seems only partially capturable in thought) are not arguments for distinctions between contents.

in his paper "A Coherence Theory of Truth and Knowledge" on the "difficulty of transmuting a cause into a reason" (CT 311). But their reasons for rejecting the Given are, it seems to me, ultimately different: McDowell's argument, unlike Davidson's, is supposed to work also against non-conceptual content. Davidson observes that "[tlhe relation between a sensation [a.k.a. the Given] and a belief cannot be logical, since sensations are not beliefs or other propositional attitudes" (CT, 311), which suggests that if they were, the problem of justification would not arise. So a view according to which perceptual experiences have content, albeit of the non-conceptual kind, would escape Davidson's worry. For the relation between the belief that an object is square and the visual experience with the non-conceptual content that an object is square evidently is, to use Davidson's terminology, "logical": these states both have content, in fact the same content. So why couldn't a causal relation between these states, like causal relations between belief, desire and action, or between beliefs and other beliefs, be within the "space of r e a ~ o n s " ? ~ ~ ~ McDowell directly addresses this objection in the postscript to his third lecture. He answers it in the course of criticizing Peacocke's explanation of how perceptual states with non-conceptual content can justify beliefs.'' (As it happens, Peacocke's explanation is part of a larger and extremely ambitious project, but we can ignore this for the present point.1') Peacocke claims, using the example of the perceptual judgement that something is square, that "[ilf the thinker's perceptual systems ' ~ nE. LePore, ed., Truth and Interpretation: perspectives o n the philosophy of Donald Davidson (Blackwell, 1986); hereafter CT. he phrase is Wilfrid Sellars'. See McDowell's p. 5. '~ndeed, a casual reading of McDowell's claim that "we cannot really understand t h e relations in virtue of which a judgement is warranted except as relations within the space of concepts: relations such as implication or probabilification, which hold between potential exercises of conceptual capacities" (7) might lead one to think that content is all that matters, not just conceptual content. For once we have content, we do have the relata of "implication or probabilification". 1°Peacocke, it should be noted, thinks t h a t the content of experience is only partly non-conceptual (SC, 88). "AS is familiar, this larger project is t o give possession conditions for concepts. Such a condition, for example for the concept SQUARE, may well employ this very concept. Only circularity with respect to possessing the concept is disallowed, not circularity with respect t o the concept itself. Thus the possession condition for the concept SQUARE may appeal to the content of experience, and may even use the concept SQUARE in specifying this content, but there will be no objectionable circularity if the content is non-conceptual.

are functioning properly, so that the non-conceptual representational content of his experience is correct, then when such experiences occur, the object will really be square" (SC, 80, quoted by McDowell at 163). The idea seems to be simply that if my perceptual state is veridical (in Peacocke's strange usage, if my perceptual systems are "functioning properly"), then a certain kind of experience is an infallible indication of a square object before me. Modulo a worry about whether my experience is veridical, I am thereby justified in believing, on the basis of this kind of experience, that there is a square object before me. This is not much of an explanation. But here we can simply take Peacocke to be baldly asserting that sometimes experience with nonconceptual content can be a reason for belief, and concentrate on McDowell's arguments why this cannot possibly be correct.12 Peacocke, McDowell complains, has severed "the tie between reasons for which a subject thinks as she does and reasons she can give for thinking that way. Reasons that the subject can give, in so far as they are articulable, must be within the space of concepts" (165). And evidently McDowell thinks that all genuine reasons must be "articulable", a t least "minimally" (165). He concedes that someone "in command of Peacocke's more or less abstruse conceptual apparatus for talking about non-conceptual content" (164) may well come to believe that something is square for the reason that her experience has a certain non-conceptual content. But, as he says, Peacocke's claim about reasons is supposed to apply to ordinary subjects. Here we have the heart of -to use a rather unhelpfully broad term- McDowell's internalism about justification.13 If all reasons must be minimally articulable -and here I take him to differ from Davidson, at least in emphasis- then the (traditional) Given and reliabilism are dead in the water, for ordinary subjects can hardly be credited with a grasp of the causal relations or the ineffable Given in virtue of which their perceptual beliefs are supposed to be justified.14 Even if McDowell is right about justification, it is a further claim that to lack such "internalist" reasons would be disastrous. So we ''Of course, McDowell himself has no better answer t o t h e question of how he knows t h a t his perceptual systems are "functioning properly" than Peacocke does. McDowell is not in t h e business of answering t h e sceptic, but rather ignoring him (113). '"or a penetrating discussion of similar "internalist" theses, see William P. Alston, "Internalism and Externalism in Epistemology", reprinted in his Epzstemic Justzfication (Cornell University Press, 1989). I4Perhaps the Given is in some sense accessible t o subjects. B u t articulating i t , as philosophers well know, is quite another matter.

don't have reasons for our perceptual beliefs, we have "schmeasons" instead, where having a schmeason is a matter of some reliable causal connection. What is so bad about merely having ~chmeasons?'~ Anyway, what I propose to do is to grant the requirement that reasons must be minimally articulable, and then argue that this does not, pace McDowell, impugn the claim that experiences with nonconceptual content may justify belief.16 Return to Peacocke's example of judging, on the basis of experience, that something is square. If we insist on the "minimal articulateness" requirement, then the subject has to be able to recognise her reason for judging that the object is square. She has to recognise, that is, that she is having an experience that is veridical only if the presented object is square. This seems enough -McDowell does emphasise "minimaln- and it certainly does not require that subject has read A Study of Concepts. So let us have the subject say, in reply to the request to give her reasons for thinking that the object is square, "Because it looks square". At this juncture it seems that McDowell has another objection, quite distinct from the one about the Folk lacking Peacocke7s "abstruse conceptual apparatus". McDowell will apparently insist that "the reason is articulable.. . so it must be no less conceptual than what it is a reason for" (166).17 Presumably this is supposed to show that the content of the experience is conceptual after all. But it doesn't. First, let's lay one possible confusion to rest. When the subject says "Because it looks square", she expresses her belief 151t's quite straightforward what McDowell's answer would be: merely having schmeasons for our perceptual beliefs would deprive us seeing how they could be about the external world. This comes out, for example, in the discussion of Davidson in the first part of the Afterword. According to Davidson, although we d o have reasons, experience can a t best be a schmeason. And this is enough t o lead t o the worry about content, on McDowell's view: "we should be suspicious of his [Davidson's] bland confidence that empirical content can intelligibly be in our picture even though we carefully stipulate that the world's impacts on our senses have nothing to do with justification" (15). But what I, a t any rate, find perfectly unstraightforward is why McDowell thinks this. I can see that a theory according to which we can think about dogs even though none has causally affected our senses might engender a worry -not necessarily insoluble- about how canine content "can intelligibly be in our picture". What I don't see is how the worry about content gets off the ground once we have "the world's impacts" firmly in place, whether or not we have an adequate story to tell about justification. 161 am not defending Peacocke's argument, because I am not confident I know what it is. I am simply objecting that McDowell's argument, that experience must have conceptual content in order t o be a justifier, is invalid. 17Cf. the last part of the preceding quotation: "Reasons that the subject can give, in so far as they are articulable, must be within the space of concepts" (165).

that the object looks square, and of course this belief is a mental state with conceptual content. Hence articulating her reason involves deploying the concept SQUARE. But that in no way shows that the experience her belief is about has conceptual content, which is what McDowell needs. Another way of arguing that the experience has conceptual content, also suggested by McDowell's remarks, is this. In order for the experience to be a reason for judging that something is square, we are assuming, the subject must able to articulate that the experience is veridical only if the object is square. Hence she must possess the concept SQUARE. Doesn't this show -without conflating the articulation of a reason with the reason articulated- that having the experience entails having the concept? And although this doesn't prove that the experience has conceptual content, it might be thought highly suggestive.18 But this second way is no better than the first. If the experience is to be a reason for the subject's judging that something is square, she must possess the concept. But this is guaranteed simply in virtue of the fact that the subject judges that the object is square. There is no reason to suppose it guaranteed simply in virtue of the subject's having the experience.lg "No more than highly suggestive, because passing the modal test is only a sufficient condition for non-conceptual content, not a necessary condition. Therefore failing the modal test does not show that the content in question is conceptual. "1 am somewhat hesitant about attributing either of the above arguments to McDowell, for they do not sit happily with the dialectic in his first objection to Peacocke. This first objection conceded that someone who cited Peacocke's theory as justification would be "someone who forms the belief, with its conceptual content, for a reason supplied by an experience, with its nonconceptual content" (164). (Then McDowell went on to make the point about "ordinary subjects".) But if either of the arguments above is really McDowell's, it seems he should have not made this concession. For here too "the reason is articulable.. .and so it must be no less conceptual that what it is a reason for" (166). Earlier in Mind and World McDowell has what appears to be another argument -for experience having conceptual content. He writes: In a particular experience in which one is not misled, what one takes in is that things are thus and so. That things are thus and so is the content of the experience, and it can also be the content of the judgement: it becomes the content of a judgement if the subject decides to take the experience at face value. So it is conceptual content (26). Here the premise appears to be that the content of the experience -the abstract object- is the same as the content of the belief endorsing it. Peacocke, for ex-

The upshot, then, is that even if we make McDowell a handsome gift of the minimal articulateness requirement, we may still resist his conclusion that sponteneity -"the involvement of conceptual capacities" (9)- extends all the way out to experience.

2

Coherentism

Let us now slide down the seesaw to the coherentist end. What has McDowell got against coherence theories of justification? The intuitive worry is that coherentism represents our conceptual scheme a s -in the best metaphor of the book- "a frictionless spinning in a void" (11). Fairy stories are coherent in an obvious sense, but that does not make them justified. Of course, it is hardly as if coherentists about justification (or truth, for that matter), have been oblivious to this worry. Davidson, who holds a coherence theory of justification, has not been, and McDowell criticises his position explicitly. It is this criticism of Davidson that puzzles me greatly, and which is the topic of this final section. Here is one place, as we'll see in a moment, where theories of justification become clearly entwined with problems about "empirical content". Now Davidson, of course, holds that as an a priori matter, one's beliefs are mostly true. For what one believes is what an ideal interpreter would hold one to believe, and an ideal interpreter must employ the principle of charity. Hence one's beliefs must be mostly true by the interpreter's lights, and since we may assume that the interpreter is not herself mistaken, one's beliefs are mostly true simpliciter, and are thereby justified. Evidently, Davidson's argument may be disputed at any number of points. His interpretative theory of the mind and the role in it of the principle of charity are both controversial, but even waiving these problems, one may question whether Davidson has answered the ample, would deny this (see fn. 6 above). In any event, pace McDowell, nothing obviously follows from this about whether the experience has conceptual content, for the distinction between conceptual and non-conceptual content is not in the first instance a distinction between contents (see the main text, section 1). Someone may hold that the content of an experience is identical to the content of the corresponding belief, and also hold that the experience may be enjoyed without possessing the concepts in question. Such a theorist would hold, with apparent consistency, that although the content of experience is belief content, the content of experience is nonetheless non-conceptual. Too bad for the terminology, admittedly.

question of how individual beliefs are justified, as opposed to sirnply answering the question of how systems of beliefs are j ~ s t i f i e d . ~ ' Further, one might have thought that McDowell has a particular complaint against Davidson, stemming from the requirement that one's reasons be minimally articulable. For the theory about ideal interpreters and so forth is only available to -borrowing a phrase from McDowell- "those who are philosophically speaking in the know" (164).~' But none of these is the difficulty that McDowell presses. Consider, he says, the familiar case of a brain in a vat. What is Davidson going t o say about that? Well, the brain (or, more exactly, the person whose brain it is) is supposed to have mostly true beliefs, and given a little more of the interpretative story, these will be "beliefs about the brain's electronic environment" (17). McDowell then argues as follows: But is that the reassurance we need if we are to be immunized against the attractions of the Given? The argument was supposed to start with the body of beliefs to which we are supposed to be confined, in our active efforts to suit our thinking to the available justifications. It was supposed make the confinement imagery unthreatening by reassuring us that those beliefs are mostly true. But the response to the brain-in-avat worry works the wrong way round. The response does not calm the fear that our picture leaves our thinking possibly out of touch with the world outside us. It just gives us a dizzying sense that our grip on what it is that we believe is not as firm as we thought (17). Now the last line of this quotation might appear to be raising a worry concerning our beliefs about what we believe. Maybe our first-order beliefs are mostly true, but perhaps, for all Davidson says, our second-order beliefs are not as fortunate. However, this 'O~avidson himself addresses this t o some extent, claiming only t o have shown t h a t individual beliefs are justified in t h e sense t h a t they have a "presumption in favour of their t r u t h (319). He certainly concedes t h a t there is more t o be said on t h e topic ("some [beliefs] may not be justified enough, or in t h e right way, t o constitute knowledge" (319)). But one may reasonably wonder just where the misssing part of t h e story is going t o come from, if it is not t o reintroduce the foundationalism t h a t Davidson abjures. 21Davidson apparently would demur: he claims t h a t "there is a pretty strong sense in which we can be said t o know t h a t there is a presumption in favour of t h e overall truthfulness of anyone's beliefs, including our own" (314). But it seems a pretty weak sense t o me, meaning something like "available t o informed philosophical reflection", which does not differentiate Davidson's account from Peacocke's.

is both a non-problem and a misinterpretation of McDowell, as the footnote appended to the quotation makes clear. There McDowell points out that Davidson's interpretative view easily allows for the desired harmony of second- and first-order beliefs. So what is the problem? It appears to be given in the last part of the footnote, which runs as follows: The problem [with Davidson's argument that belief is mostly veridical] is that.. .we ring changes on the actual environment (as seen by the interpreter and brought into the interpretation) without changing how things strike the believer, even while the interpretation is supposed to capture how the believer is in touch with her world. This strikes me as making it impossible to claim that the argument traffics in any genuine idea of being in touch with something in particular. The objects that the interpreter sees the subject's beliefs as being about become, as it were, merely noumenal as far as the subject is concerned (17, fn. 14). If I understand this passage correctly, it raises a genuine difficulty for Davidson's official position (at any rate, I'll assume it's Davidson's). However, Davidson's view can easily be amended, in either of two ways, to sidestep McDowell's argument. Moreover, at least one of these ways appears congenial to McDowell himself. Hence McDowell does not succeed in raising a serious problem for Davidson-style coherentism. Here's what I take to be McDowell's argument. Consider a brain in a vat -an exact duplicate of your brain, say. What is it like to be this brain (the person whose brain it is)? Well, what it's like for the brain is just the same as what it's like for you. In other words, you and the brain share the same phenomenology -in one sense of this phrase, things seem the same to you as they do to the brain. But now, the brain is alleged by Davidson to have mostly true beliefs about its world, the world of electronic impulses and computer states. Suppose that you see a dog in a manger, and you think "that dog is in a manger". How the world is, that that dog is in a manger, is something that your experience reveals to you. But the brain is having exactly the same experience, characterized phenomenologically. It's just that the brain is experiencing that that electronic impulse is in state such-and-such. But if this very experience you're having, phenomenologically characterised, could be a n experience as of electronic impulses, a thick veil of ideas has suddenly descended between you and the objects. T h e dog is behind it -a merely noumenal dog, as McDowell says. This is a serious problem. McDowell sometimes puts it this way: that Davidson makes a mystery of how beliefs can have "empirical

content". I do not see that: your belief is about a dog in manger, and Davidson has explained, a t least to some extent, how this is so. But the other way -admittedly metaphorical- McDowell uses seems to me apposite: that for Davidson sense impressions are not "transparent" .22 We will have a solution if we can amend Davidson's interpretative theory of the mind so that you and the envatted brain do not enjoy the same phenomenology: either the brain's phenomenology is quite unlike yours, or the brain has no phenomenology a t all. This will imply, of course, that phenomenology is not intrinsic -that it does not solely supervene on what's in the head. Recently, intentionalism has been gaining adherents: the claim that the phenomenal aspects of an experience are determined solely by the representational content of that experience.23 And Davidson is, of course, an externalist about content. A theorist who holds both these views has ample reason to deny that phenomenology is locally "The difficulty t h a t I a m taking McDowell t o raise for Davidson is similar t o what Mark Johnston calls t h e "problem of acquaintance, t h e problem of how we could be acquainted with anything given t h e nature of information transmission" ("How t o Speak of the Colors", Philosophical Studies 68 (1992), 221-263, a t p. 256). The way Johnston has of setting t h e problem u p seems t o me crucially t o depend on t h e assumption t h a t there are sense data, or something of the sort (e.g. Peacocke's "sensational properties": see his Sense and Content (Oxford University Press, 1983), ch. 1). "[S]ensory experience", Johnston writes, "is unsatisfyingly like morse code transmission; both involve interpretable effects a t t h e end of an information-bearing process or signal. But the intrinsic natures of the originators of the signal are not manifest in t h e signal. This is a very depressing comparison" (257). Clearly t h e assumption is t h a t we are aware, when enjoying a visual experience of a dog, of not only the dog, but also of the uzsual effects of t h e dog. And then t h e difficulty is t h a t our awareness of the dog is indirect, going via awareness of its effects. This problem, it seems t o me, goes away if we deny t h a t we are aware of any such effects, as is typically denied by proponents of intentionalism (see the second paragraph below in the text). (For more on Johnston's problem of acquaintance, see his contribution t o this volume.) 2 3 ~ e ee.g., , Gilbert Harman, "The Intrinsic Quality of Experience", Philosophical Perspectives 4 (1990), 31-52; Sydney Shoemaker, "Phenomenal Character", 28 (1994), 21-38; Michael Tye, Ten Puzzles of Consciousness (MIT Press, forthcomine): and William G. Lvcan and Fred Dretske's contributions t o this volume. There is a slight complication here. Davidson does not think of experience as possessing content a t all, or a t any rate does not use t h e word 'experience' t o mark a representational state (as McDowell notes a t p. 140, fn. 13). So the standard way of putting intentionalism would perhaps not be welcome. But we could avoid this difficulty by equipping Davidson with a version of intentionalism according t o which phenomenology supervenes on the contents of t h e subject's beliefs, or intentional states generally.

~~u~

" 1 ,

supervenient. Externalism, together with the assumption that the brain is interpretable, implies that the content of the brain's experiences is radically different from yours. Adding intentionalism to the mix makes it easy to hold that the brain's phenomenology is radically different from yours.24 Now McDowell's problem does not arise. That is the first solution. The second solution begins by denying that the brain is any more interpretable than a pickled walnut, which I would have thought fits extremely well with many of Davidson's writings on the topic. To complete the second solution, we need the weak assumption that without intentional content, there can be no phenomenology. Putting those two together, we get the result that there is nothing it is like to be a brain in a vat. Again, McDowell's problem does not arise. Now many will protest that there must be something it's like to be a brain in a vat. But McDowell can hardly make this complaint. As he has written elsewhere, "the sheer fact that a brain is going through the motions that an embodied brain goes through when a person thinks or experiences is by itself no ground a t all for supposing that there is a mind in there".25 So the seesaw, I think, does not oscillate as it should. To end on a cautionary note: my spade has only scratched the surface of McDowell's rich and difficult text -bedrock is far away.

241ntentionalism does not entail t h a t the brain's phenomenology is different from yours: intentionalism merely says t h a t there can b e no phenomenological difference without an intentional difference, and this is consistent with (to take an extreme example) there being no phenomenological differences a t all. 25 L'TheContent of Perceptual Experience", Philosophical Quarterly 44 (1994), 190-205, a t p. 201. Incidentally, there's much reason t o suppose t h a t McDowell would be sympathetic with the intentionalist assumption employed in the first solution.


McDowell's Direct Realism and Platonic Naturalism Roger F. Gibson

1 Introduction In Mind and World* (Cambridge, Mass.: Harvard Univeristy Press, 1994). Professor McDowell's main concern is to answer the venerable question of the relation of mind and world. He formulates this question in the somewhat quaint Kantian vernacular of spontaneity and receptivity: What's the relation between spontaneity and recep tivity? According to McDowell, two opposite and equally unsatisfactory answers to this query have dominated traditional epistemology. First, foundationalists have declared that receptivity provides the Given which furnishes both the content and the grounding for spontaneity's empirical thoughts and judgements. On the other hand, coherentists have declared that the Given, being non-conceptual, is useless for the purpose of providing the content or the grounding for those empirical thoughts and judgements. Foundationalists retort that coherence, in the absence of constraints originating outside of thought, is incapable of explaining how empirical thoughts and judgements bear on the world. McDowell aptly pictures this exchange between advocates of the Given and advocates of uncon'Numbers inside parentheses refer to pages in this book.

strained coherentism as a perpetual seesaw: struck by the uselessness of the Given, one recoils to coherentism, but then, struck by the defects of coherentism, one recoils back to the Given, and on, and on. While paying homage to both Wittgensteinian therapy and Kantian theory, McDowell proffers his plan for dismounting the seesaw. The key, he explains, is to give up thinking of experience in terms of the Given and to learn to think of experience as necessarily imbued with conceptual content. According to McDowell, experience has conceptual content because, in delivering forth experience, the passive faculty of receptivity draws on the conceptual capacities of spontaneity. In other words, contrary to Kant, receptivity makes not even a notionally separate contribution to experience. This is McDowell's key insight for dismounting the seesaw: since one's experience is both passively acquired and (unlike the Given) conceptual, receptivity may intelligibly be said to provide empirical constraints on one's active faculty of spontaneity, the faculty by which one exercises one's conceptual capacities in thinking and judging. McDowell's plan for dismounting the seesaw is a very sophisticated piece of philosophical reasoning. As such, however, I think it owes more to Kantian theory than to Wittgensteinian therapy (or quietism). For example, there is a great deal of theory construction afoot in McDowell's treatments of receptivity and spontaneity, experience, direct realism and the world, and platonic naturalism. The reason for this is that McDowell unabashedly accepts the traditional problematic of giving an account of how individual minds relate to the world. This sanguine attitude toward the traditional problematic is unbecoming a philosopher intent primarily upon relieving conceptual cramps. Thus, McDowell's plan for dismounting the seesaw ought to be viewed as a new entrant into the field of competing theories of mind and world rather than as a technique for showing flies out of bottles. McDowell's general position regarding the relation of mind and world which emerges from his book could be called a coherent form of foundationalism (see, for example, 29). It is a form of foundationalism since it allows that certain judgements are warranted by experience; it is a coherent form of foundationalism since it maintains that experience can warrant certain judgements because experience is irreducibly conceptual. Though I am genearally sympathetic to much of what I take to be McDowell's general position, some key parts of his position remain opaque to me. Thus, my remarks below are proffered more in the spirt of a request for elaborations than as criticism.

2 Receptivity and Realism As we have noted, for McDowell, receptivity is a passive faculty; in it's openness, it "takes in" the world. Such "takings in", or experiences, are irreducibly conceptual because receptivity in operation draws on the conceptual capacities of spontaneity in delivering forth experience; thus, there is no Given. This is the insight which allows one to dismount the seesaw of traditional epistemology. Furthermore, McDowell explains that "[iln a particular experience in which one is not misled, what one takes in is that things are thus and so. That things are thus and so is the content of the experience, and it can also be the content of a judgement. .." (26). Moreover, "that very same thing, that things are thus and so, is also a perceptible fact, an aspect of the perceptible world" (26). Presumably, McDowell intends 'is' in the immediately preceding quotation to be understood as the 'is' of identity: a non-misleading experience is identical to a particular aspect of the world. This interpretation of McDowell's remarks raises two questions: (1) how does he picture the external world? and (2) how does he handle the problem of error? McDowell maintains that aspects of the external world (i.e., the world of "outer experience"), including those aspects which can be "taken in" via the openness of receptivity, exist independently of all instances of experiencing, or, more generally, of all instances of thinking. So, if there were no experiencers, no thinkers, there would still be an external world. However, despite this thoroughgoing realism, McDowell maintains that no aspect of the world lies essentially beyond the scope of thought. Combining these two features, McDowell pictures the world as a totality of potential contents of (true?) thoughts. Such potential contents are facts, according to McDowell, so the world divides into the totality of facts. Thus, McDowell's world is more akin to the world of the Tractatus than to some Kantian Ideal of pure reason. However, as McDowell notes, his picture of the world appears to have a certain arrogant anthropocentric aspect: "a baseless confidence that the world is completely within reach of our powers of thinking" (39-40). He attempts to counter this charge by pointing out that the faculty of spontaneity carries with it a standing obligation to reflect on the credentials of the putatively rational linkages that, at any time, one takes to govern the active business of adjusting one's world-view in response to experience. Ensuring that our empirical concepts and conceptions pass muster is ongoing and arduous work for the understanding.

It requires patience and something like humility. There is no guarantee that the world is completely within the reach of a system of concepts and conceptions as it stands at some particular moment in its historical development. Exactly not; that is why the obligation to reflect is perpetual. (40) However, merely pointing out that spontaneity is under a perpetual obligation of self-criticism fails to meet the heart of the objection. At best, that response justifies taking McDowell's picture of the world as a heuristic, a Kantian Ideal of pure reason. But as we have seen, McDowell's picture of the world is metaphysically substantial. Thus, meeting the anthropocentric objection calls for a justification for identifying the world with the totality of potential contents of (true?) thoughts in the first place. I a m not sure that McDowell has provided such a justification. What, now, of the problem of error? If the content of a n experience and of a judgement regarding that content is that things are thus and so, and if that very same thing, that things are thus and so, is an aspect of the world, then how is error possible? How is a misleading experience possible? Just what gets "taken in" in such instances? Surely not an aspect of the world. So far as I can tell, McDowell does not discuss the problem error (the problems of how we can get things wrong), though he does discuss the correlative problem of scepticism (the problem of how we can get things right). He puts the sceptical question as follows: "how can one know that what. one is enjoying a t any time is a genuine glimpse of the world, rather than something that merely seems t o be that" (112)? McDowell's short answer is t o concede that we cannot tell. But [he adds] that is beside the point. It would matter if it showed that the very idea of openness to facts is unintelligible, and it does not show that. For my present purposes, the sheer intelligibility of the idea is enough. If the idea is intelligible, the sceptical questions lack a kind of urgency that is essential to their troubling us, an urgency that derives from their seeming to point up an unnerving fact: that however good a subject's cognitive position is, it cannot constitute her having a state of affairs directly manifest to her. There is no such fact. The aim here is not to answer sceptical questions, but to begin to see how it might be intellectually respectable to ignore them, to treat them as unreal, in the way that common sense has always wanted to. (113) McDowell is surely correct when he says that the sceptic's questions do not show that the idea of openness t o facts is unintelligible, but by

the same token, I do not think the idea will become fully intelligible until he explains his resolution of the problem of error.

3

Spontaneity and Platonic Naturalism

As we have seen, according to McDowell spontaneity is under "a standing obligation to reflect on the credentials of the putatively rational linkages that, at any time, one takes to govern the active business of adjusting one's world-view in response to experience" (40). This activity of spontaneity appears to be sui generis (i.e., beyond causal laws). Thus, even if McDowell's account of receptivity is accepted, he still owes an account of the sui generis character of spontaneity. McDowell's discussion of spontaneity is centered around three concepts: nature, the realm of law, and the realm of reason. The kind of intelligibility associated with the realm of law is characteristic of science; the kind of intelligibility associated with the realm of reason is characteristic of meaning. McDowell places spontaneity squarely within the realm of reason. He then proceeds to discuss four conceptions of how spontaneity might relate to nature; they are: (1) Bald Naturalism, (2) Davidsonian Naturalism, (3) Rampant Platonism, and (4) Platonic Naturalism. Bald Naturalism equates the realm of law with nature and then tries to show how spontaneity and the realm of reason generally can be constructed entirely from the resources of the realm of law. This approach t o relating spontaneity to nature is tantamount to denying that spontaneity is sui generis. McDowell dismisses Bald Naturalism as pie in the sky. Davidsonian naturalism also equates the realm of law with nature, but denies that spontaneity can be explained naturalistically. Thus, unlike Bald Naturalism, Davidsonian Naturalism accepts that spontaneity is sui generis. But even though Davidsonian Naturalism recognizes two sets of conceptual capacities, it tries to get by with an ontology confined to the realm of law. Though McDowell is not among those who have a taste for desert landscapes, he dismisses Davidsonian Naturalism primarily on the grounds that because it identifies the realm of law with nature, it disallows his conception of experience -a conception involving a receptivity imbued with concepts. (This criticism applies to Bald Naturalism and to Rampant Platonism as well.) Rampant Platonism equates the realm of law with nature and the realm of reason with the extra-natural. On this view, humans are

partly in nature and partly outside of nature. McDowell dismisses Rampant Platonism on the grounds that it makes the human capacity to respond to reasons look like an occult power. "What we wanted", McDowell explains, "was a naturalism that makes room for meaning, but this is no kind of naturalism a t all" (77-78). Finally, there is the naturalism that McDowell favors, Platonic Naturalism. Unlike its three competitors, Platonic Naturalism refuses t o equate the realm of law with nature. Rather, McDowell expands the concept of nature so that it encompasses both the realm of law and the realm of reason. However, unlike Bald Naturalism, Platonic Naturalism eschews any attempt to assimilate the realm of reason t o the realm of law. Accordingly, nature is two-tiered: there is "first nature" which is comprehendible within the realm of law, and there is "second nature" which is comprehendible within the realm of reason. Furthermore, unlike Rampant Platonism, Platonic Naturalism construes the realm of reason as a natural outgrowth of ordinary human development. In this regard, McDowell's basic idea seems to be that people are, during their normal course of maturation, initiated into particular conceptual capacities, including responsiveness t o the demands of reason: Our nature is largely second nature, and our second nature is the way it is not just because of the potentialities we were born with, but also because of our upbringing, our Bildung. Given the notion of second nature, we can say that the way our lives are shaped by reason is natural, even while we deny that the structure of the space of reasons can be integrated into the layout of the realm of law. (87-88) McDowell seems to espouse a form of realism, in this connection, when he adds that "[tlhe demands of reason are essentially such that a human upbringing can open a human being's eyes to them" (92), but they exist whether anyone ever comes to know them or not. But, if this really is a form of realism it differs radically from the realism which McDowell espouses in connection with the denizens of outer experience, for he says that they would exist even if there were no thinkers a t all. Not so, the demands of reason; for if there were no thinkers then there would be no second natures either. In conclusion, I a m left wondering how McDowell might respond to the following three clusters of questions -answers t o which might help readers to better understand McDowell's book; they are: (a) Is it McDowell's view that the world consists of the totality of potential contents of (true) thoughts? If so, does he have a reason for holding this view which is independent of the circumstance that it facilitates

dismounting the seesaw? In particular, does he have a reason for equating the world with the potentially thinkable, when that view is taken to be a substantive metaphysical claim as opposed to a Kantian heuristic? (b) Does McDowell view the problem of error (as opposed to the problem of scepticism) as a problem for his direct realism? If so, how does he think that problem should be addressed? (c) Is McDowell really a realist regarding the demands of reason? Is he, perhaps, a metaphysical realist regarding the world of outer experience, but an internal realist regarding the demans of reason?


Reply to Gibson, Byrne, and Brandom John McDowell

I am grateful to my commentators for the pains they have taken with my book. I am sorry that I cannot deal properly with all the challenging issues they have raised.

1 Gibson Gibson says I "unabashedly" accept "the traditional problematic of giving an account of how individual minds relate to the world". He is right that'such a stance would be "unbecoming a philosopher intent primarily on relieving conceptual cramps". But I cannot see why he thinks the stance is mine. My concern is with why the questions that are raised within the traditional problematic seem to be real ones. It is true that I use conceptual apparatus from philosophy's past, for instance "the somewhat quaint Kantian vernacular of spontaneity and receptivity". But this does not justify Gibson's claim that I accept the traditional problematic. Discussing Rorty, I urge that we should not take just any occurrence of the language of traditional philosophy to reveal a stance within the tra-

ditional problematic. Traditional language can be exploited with a view to exorcizing the traditional anxieties, rather than answering the questions that give expression to them. (See Mznd and World, p. 155.) Gibson describes my outlook as a form of foundationalism. What he means is that I credit experience with a warranting role, and he is right about that. But the image of foundations is potentially misleading as a way of capturing my picture; this is a counterpart to a point made by Sellars.' There is indeed a relation of rational dependence, of what (if this were the whole story) we might be tempted to call "superstructure" on what we might be tempted to call "foundations". But just because concepts are involved in experience, and the conceptual realm is a seamless web of rational interconnections, there is also a rational dependence (of a different sort) in the opposite direction. We would have to say that, in respect of this other dimension of rational dependence, the "foundations" are partly held and that makes the image of founin place by the LLsuperstructure", dations unhappy. Gibson says I dismiss bald naturalism "as pie in the sky". That is not right. My point against bald naturalism (at least in Mznd and World) is that it does not concede what I can concede as an insight, in the thinking that makes the traditional problematic seem compulsory. Hence it yields a less satisfying exorcism of philosophical anxiety than my alternative. Gibson ends with three "clusters of questions" that summarize the main concerns he expresses. The best way to respond to his concerns will be to try to answer his questions. (1) Is it my view that the world consists of the totality of potential contents of (true) thoughts? Yes. (We can just say "the totality of true thoughts", if we use "thoughts" in the sense of thinkables rather than episodes of thinking.) Do I have a reason for holding this view, apart from its helpfulness in getting out from under philosophical anxieties? No; do I need one? Gibson suggests I do, on the ground that my picture of the world is "metaphysically substantial". I am not sure what he means by this. The world is everything that is the case; that is, everything that can be truly thought to be the case. There is a permanent possibility of having to decide we were wrong, '"Empiricism and t h e Philosophy of Mind", p. 300: "the metaphor of 'foundation' is misleading in t h a t it keeps us from seeing t h a t if there is a logical dimension in which other empirical propositions rest on observation reports, there is another logical dimension in which t h e latter rest on the former." A counterpart only, because Sellars's picture has "observation reports" where I have experience.

26. REPLYTO GIBSON, BYRNE,AND BRANDOM 285

and that is enough to ensure that the world so conceived does not degenerate into a shadow or reflection of the norms that, at any time, we take to govern our thinking -a junior partner in the interaction of mind and world. That is as much "realism" about the world as I want; as far as I can see, it is as much "realism" about the world as it is sensible for anyone to want. (2) Do I view the problem of error as a problem for my direct realism? No. In fact I am not sure what Gibson means by "the problem of error". Earlier he formulates it like this: "How is a misleading experience possible?" But I see no difficulty for me here. How is it possible that straight sticks look bent when partly immersed in water? As I understand this question, it does not express a conceptual difficulty; the question can be answered by pointing to facts about how light travels through air and water. The task of answering such questions in detail is not mine in particular. No doubt there is a different difficulty about the possibility of error, a conceptual difficulty, confronting people who aim to construct a notion of the content of experience out of, for instance, an idea of what the senses are for, parallel to the idea that what the heart is for is to circulate the blood. But I have no such project.2 (3) Am I really a realist regarding the demands of reason? Yes -as much a realist as I am regarding anything (compare (1) above). Gibson suggests I am not entitled to say that, on the ground that "if there were no thinkers there would be no second natures either". But to suggest this opens a gap between my attitude to non-normative facts and my attitude to the demands of reason is to imply that the demands of reason figure in my picture as projections from, or reflections of, second natures. That ignores the fact that my naturalized platonism is a platonism. In the picture I recommend, acquiring a second nature brings the demands of reason into view (a view that may be skewed or inaccurate in some respects). It does not bring them into being.

' w h e n he talks about my discussion of "the correlative problem of scepticism", Gibson says my response t o t h e sceptical question ("How can one know t h a t what one is enjoying a t any time is a glimpse of t h e world, rather t h a n something t h a t merely seems t o be that?") is "to concede t h a t we cannot tell". W h a t I concede is t h a t we cannot tell to the satisfaction of the sceptic. This makes a difference. "We cannot tell" (period) risks giving t h e sceptic everything he wants, whereas I think I a m entitled t o claim t o know, as I sit a t my desk writing these words, t h a t what I a m enjoying is glimpses of t h e world.

2

Byrne

Leading into his discussion of the way I try to exploit an oscillation between the Myth of the Given and coherentism, Byrne says he "cannot see how the promised oscillation is supposed to get started". In particular, he does not understand why I think what is unsatisfactory about the positions that I suggest we are prone to oscillate between is that they make a mystery of how thought can have empirical s on tent.^ I hope the role I have given to "minimal empiricism" in my prkcis makes this clearer (if not more convincing). The oscillation gets started (this is the beginning of the diagnosis I offer in my fourth lecture) because we find ourselves conceiving experience in such a way that the idea of experience (understood in terms of impressions, the world's impacts on us as possessors of sensory capacities) would have to function in a logical space that is alien to the logical space in which concepts like that of answerability function. Seen clearly, this conception of experience makes minimal empiricism unavailable. That leaves empirical content mysterious, if we find ourselves unable to abandon the minimal empiricist thought that empirical content is intelligible only in terms of answerability to the tribunal of experience. And I think no one would dream of abandoning that thought, if abandoning it did not seem to be required by a powerful philosophical argument. In the context of how the natural character of sentience pushes us towards taking the idea of experience to belong in a logical space that is alien to the logical space of reasons, the undimmed attractiveness of minimal empiricism entices us, intelligibly, into the Myth of the Given. To embrace the Myth of the Given is to suppose our thinking can be responsive to both those forces: both the conception of experience as natural, in a sense that would in fact imply that the idea of experience is alien to the space of reasons, and minimal empiricism. But this is self-deceptive. It requires us to misconstrue the implications of conceiving experience as natural in the relevant sense; it requires us to think we can conceive experience in that way, but still take it that experiences themselves, not just beliefs about experiences, can be what beliefs and judgements are based on. (This is what figures in my first lecture as the thought that the space of 3 ~ makes e "a partial exception for coherentism". This looks forward t o a passage near the end of his comments, where he accepts t h a t I point t o something t h a t really is a problem for Davidson's coherentism. But there too he objects t o my formulating the problem by saying Davidson's position makes a mystery of how thought can have empirical content. I shall come t o this below.

26. REPLYT O GIBSON, BYRNE,A N D BRANDOM 287

reasons extends more widely than the space of concepts.) If we unmask this self-deception, and accept that conceiving experience as natural in the relevant sense implies that the idea of experience is alien to the logical space of reasons, then, so long as we stick to that conception of experience, there is no option but Davidsonian coherentism, the renunciation of even minimal empiricism. That is the oscillation I exploit, between a self-deceptive conviction that we can still have minimal empiricism in the context of a conception of experience that is naturalistic in a certain sense, on the one hand, and Davidson's clear-sighted realization that we cannot, on the other. I picture it as an oscillation, rather than as a structure of thought that simply forces us into the only position so far in view that is not self-deceptive, precisely because the putatively forced move is to renounce minimal empiricism. Davidson does nothing to undermine the plausibility of the thought that empirical content is intelligible only in terms of what it now appears we cannot have, the idea of answerability to the tribunal of experience. (Nor, so far as I know, does anyone else.) So the putatively forced move leaves the possibility of empirical content mysterious; it denies something that gives all the appearance of being a necessary condition for empirical content not to be mysterious -and not by explaining away the appearance that it is a necessary condition, but just by arguing that it cannot be true. If we look a t the putatively forced move from this angle, we need not be surprised that anyone is tempted into the Myth of the Given. That way, at least one accepts that apparently necessary condition for empirical content not to be mysterious. Of course this does not suffice for making empirical content unmysterious, and Davidson and Sellars are right that there is self-deception in thinking the job has been done. This is where we came in; the seesaw tilts again. My problem about Davidson's coherentism is, as I said, that Davidson does nothing t o undermine the thought that minimal empiricism, which he rejects, is a necessary condition for empirical content to be anything but mysterious. In the framework Davidson recommends, "How is empirical content possible?" ought to be an urgent question. And Davidson's only move in this area -the argument, from the nature of interpretation, that most beliefs are truedoes not address that question. It presupposes that we have beliefs, equipped with empirical content, as they must be to be beliefs a t all. That means the argument comes too late to deal with the problem that is in fact posed, I claim, by rejecting minimal empiricism.4 4 0 f course I a m not suggesting it is in doubt t h a t we have beliefs (which of course have empirical content). But rejecting minimal empiricism saddles us with

When I discuss the alarming thought that one might be a brain in a vat, my aim is just to make this dissatisfaction with Davidson vivid. According to Rorty, Davidson once suggested that the argument from interpretation applies to the mere brain that, in the alarming thought, one worries that one might be; the argument shows that even on that supposition, one has mostly true beliefs. In my first lecture, I urge that if the argument from interpretation works like that, it is a poor reassurance that one is in touch with the world. It makes the objects that one's beliefs are mostly true about, in the view of a good interpreter, merely noumenal in relation to one's own point of view. My idea was that this brings out vividly that the materials Davidson exploits do not yield a satisfactory conception of how empirical content is possible a t all, however well they may do in answering (or silencing) an ordinary scepticism. Now Byrne grants that I do here pose a problem for Davidson, a t least as Rorty reports him.5 But he suggests that the Davidsonian position can be amended, so as not to have t o credit a brain in a vat with beliefs a t As he remarks, this "fits extremely t h e question how it can be t h a t we do. (A comparison: suppose a philosopher tries t o combine saying "Of course it is not in doubt t h a t we perceive objective states of affairs" with t h e thesis t h a t our intake in perception is restricted t o something common between veridical perception and illusion. In t h e context of t h a t thesis, there is no "of course" about our perceptual grasp on reality.) 'Byrne says he does not see why I put t h e point by claiming Davidson makes empirical content mysterious (see above). B u t surely it is obvious t h a t if we have a story, putatively about content, t h a t would succeed a t best in displaying t h e mind as directed towards noumenal objects, we cannot have provided for anything recognizable as empirical content. (And we could p u t a Kantian thought by saying t h a t a notion of directedness towards noumenal objects is not really a n intelligible notion of content a t all.) 61 a m focusing on t h e second of t h e two moves Byrne offers. I a m less sure t h a t t h e first is effective. If I have a sceptical worry t h a t I want t o express by saying "Maybe I a m a brain in a vat", how would it help t o hit me with a doctrine t o t h e effect t h a t a mere brain's phenomenology would be different from mine? My worry is t h a t thzs phenomenology (mine) may be t h a t of a mere brain. It is not as if t h e intentionalist doctrine Byrne appeals t o would allow me t o say this: if I were a brain in a vat, t h e difference in my phenomenology would reveal t h a t t o me; so, since it does not, I can conclude t h a t I am not. Byrne's second move permits t h e Davidsonian response t o scepticism, purportedly made urgent in terms of t h e spectre of t h e brain in a vat, t o be put in this arguably telling way: success in t h e attempt t o suppose t h a t my beliefs are not mostly true, if such a thing were possible, would undermine my entitlement t o think I can get my mind around anythzng (since it would undermine t h e idea t h a t I a m interpretable) including t h e supposed thought t h a t I a m a brain in a vat. (Compare Putnam's treatment of "brain in a vat" scepticism, in chap. 2 of Reason, Truth and Hzstory (Cambridge University Press, Cambridge, 1982).)

26. REPLYTO GIBSON,BYRNE,AND BRANDOM 289

well with many of Davidson's writings on the topic". Perhaps it is not an amendment at all; perhaps the extempore remark cited by Rorty was just a mistake. If that is right, I lose a means to make my dissatisfaction with Davidson's easy rejection of minimal empiricism vivid. But this leaves the dissatisfaction unallayed. (Whereas Byrne implies that if I cannot use the brain in a vat as a complaint against Davidson, that brings my oscillation to a stop.) The fact remains that Davidson does nothing to dislodge the plausibility of the minimal empiricist thought, that empirical content is intelligible only in terms of answerability to the tribunal of experience. Just arguing that it must be false does not explain away the appearance that it must be true. I shall be in a better position to say something about the detail of the first half of Byrne's comments after I have responded to Brandom. At this point let me say something general about reliability. From what Byrne says, one might gather that my picture has no use at all for the idea of reliability, but that is not so. For one thing, reliability must be present in the conceptual surroundings of the very idea that someone has a capacity to see (for instance) that things are thus and so. That seems obvious to me, even though I had no For another thing, reliable linkoccasion to mention it in my ages within the organismic constitution of perceiving and thinking animals would surely figure in executions of the good project, scientific in an intelligible sense, that, as I point out in the prkcis, my position in Mind and World leaves room for (though it is no part of my concern in the book): the project of describing the material constitution of perceivers, and more generally beings with minds, so as to make it unsurprising that their interactions with the world are such as to reveal their being percipient, and more generally to reveal their possession of minds. One more remark about that project, not specifically aimed at Byrne (though his resistance to my book may stem, at least in part, from the suspicion that I leave no room for some clearly worthwhile scientific activities). Engaging in the project can be seen as addressing questions such as "How is it possible that organisms have minds?", and more specific counterparts, on an interpretation that distinguishes them from the sort of question, formulable in the same 7 ~ h i point s is in the vicinity of Brandom's suggestion that I have no ground to reject weak, or epistemological, reliabilism. But what Brandom means by "epistemological reliabilism" is something more committal (and by my lights more sinister) than the thought I here acknowledge to be obvious. I shall say something about this below.

words, that I aim to exorcize (rather than answer). On the interpretation that is my concern, a "How possible?" question expresses a way of thinking that gives the appearance of showing that its putative topic is conceptually impossible. If a "How possible?" question is asked in that spirit, a response in, so to speak, engineering terms a perspicuous description of the appropriate material constitutionwould be beside the point. Perhaps such a response might happen to suggest a diagnosis of what goes wrong in the way of thinking that finds expression in a "How possible?" question of the sort that is my concern, but that would be incidental to its aim. I can insist, on these lines, that what is needed to relieve a conceptual cramp is something different, without in any way debunking inquiry into the machinery of mindedness.

3

Brandom

It would take an inordinate space for me to touch on everything I would like t o deal with in Brandom's comments. I shall have to pass over a lot that I would need to take issue with if I were to attempt a complete engagement with Brandom's remarks -which would need to include a complete engagement with Brandom's own recent book.8 The third of the three commitments into which Brandom articulates my basic stance is "that the representational norms that connect the correctness of our thought to the facts, to how things are with that bit of the world the thought therefore counts as being about, must be understood as aspects of the rational norms that govern [the] process of active critical reflection on credentials". Brandom remarks in a general way that I do not undertake to vindicate my entitlement to the three commitments. About that one in particular, he suggests that I need a story about ''just how reference to what we are thinking about precipitates out of (or is intelligible as expressing an aspect of) rational relations among various things we might think (or beliefs we might have), as it turns out, about what we are thinking about". His book contains just such a story, and his point here is to suggest that, in so far as I do anything worth doing in this area, my book stands to his as promissory note to execution of the promise. Churlishly enough, I reject this offer of help. My book is not a sketch (with some of the outlines in the wrong places) for Bran' ~ a k i it n ~Explicit: Reasoning, Representing, and Discursive C o m m i t m e n t (Harvard University Press, Cambridge, Mass., 1994).

26. REPLYTO GIBSON,B YRNE,AND BRANDOM 291

dom's finished painting. For my purpose, I do not need Brandom's story about how representational directedness to facts, construed as involving reference to objects, precipitates out of something in which it is not yet explicitly in view. My aim is to exorcize a general conceptual difficulty, which shows up in a variety of specific forms. I can put the general difficulty like this. Suppose one pictures the structure of what is a reason for what -the topography of the space of reasons- in a potentially alarming way, as the layout of Plato's heaven. (Remember that my naturalized platonism is a platonism.) If one starts from there, one can be forgiven for finding it problematic how there could be such a thing as responsiveness to rational constraints as such. That would require that those relations, which constitute the topography of the ideal, enter into determining some of what happens -occurrences in the realm of (as one will be tempted to say if one feels the difficulty in this form) the merely actual, in which actual happenings occur as a result of actual circumstances and other actual happenings. How can the ideal interact with the merely actual in the way it must if it is to make a difference to what happens? To extrapolate from what I say in my book (where the difficulty does not figure in these terms): this felt difficulty can be diagnosed as issuing from a restrictively naturalistic conception of what figures, when the difficulty is felt in this shape, as "the merely actual". Remember -I say- that second nature, which brings the ideal into our view by opening our eyes to the demands of reason, is nature too, so it can figure in making intelligible, as natural phenomena, the bits of actuality that constitute elements in our lives. Then the felt difficulty evaporates. This difficulty could arise specifically about the idea of rational responsiveness, in forming beliefs, to other things one believes, or might believe. What Brandom describes as "rational relations among various things we might think (or beliefs we might have)" would give the shape such responsiveness would ideally have. If my exorcism works, it dissolves the difficulty in that specific application. But the difficulty can also arise directly in connection with the idea that deciding whether to believe that things are thus and so must be rationally responsive to the fact that things are thus and so, or the fact that things are not thus and so (as the case may be).9 As empiricists, we want to distinguish such responsiveness into immediate 'That things are thus and so is not another thing one might believe, but the very thing one is deciding whether t o believe; the rational relation here is not "among various things we might think (or beliefs we might have)".

responsiveness, where the beliefs are observational, and mediated responsiveness, where the beliefs are theoretical. It is the difficulty in this form, as a difficulty about rational responsiveness to facts, that, by (in effect) bringing minimal empiricism into question, threatens the very idea of empirical content -representational directedness towards the empirical world- in the way I aim to deal with. The undermining of minimal empiricism poses the threat directly where the responsiveness would have to be immediate, and indirectly where it would have to be mediated. And if my exorcism works, I can apply it at once, without preliminaries, to the difficulty in this form. I can understand, and (unlike proponents of bald naturalism) sympathize, if someone finds a threat here to the very idea that judgements or beliefs are representationally directed towards the facts; but my invocation of second nature disarms the threat. In doing so, it frees the idea of representational directedness towards the facts from the intelligible but misguided conceptual difficulty that is my sole concern about it. For this philosophical exercise of mine, I have no need to proceed stepwise, in the way Brandom suggests I must. I have no need to start by vindicating an employment of the idea of rational responsiveness in which rational responsiveness to the facts is not yet explicitly in view, and only then to proceed to a question about how the latter precipitates out of the former. I can go straight to the latter.'' I turn now to Brandom's main charge against me. The charge is this: I claim that my picture is compulsory, but my purported basis for that -the "rational constraint constraint", on which Brandom represents himself as agreeing with me-- is equally satisfied by a different position, in which an externalist reliabilism does the work I ' O ~ h ecriticism I a m responding t o here reflects how Brandom misses t h e significance of a fact t h a t I can p u t like this: t h e idea of active critical reflection on credentials figures in my book not, as it does in his, as a building-block in a constructive and (so t o speak) context-free account of mindedness (including t h e directedness of thought a t t h e world), but as a n ingredient in a train of thought t h a t poses a threat, which I aim t o exorcize. This is not t h e place t o comment in detail on Brandom's different project and his execution of it. (To avoid being drawn into t h a t , I have ignored t h e way in which, when Brandom complains t h a t I d o not display the precipitating out t h a t I should, he equates representational directedness towards t h e facts, on the one hand, a n d , on the other, the relation of thinking or speaking t o objects, in virtue of which it is how things are in respect of those objects t h a t determines whether t h e thinking or speaking is true. To deal with this, I would need t o discuss Brandom's project in his chap. 6, with which I have some sympathy, and his project in his chap. 8, which seems t o me t o be misbegotten for reasons I shall get close to, b u t not formulate, in a n argument I deploy below in self-defence.)


assign to the idea of rational responsiveness to experience -the work of ensuring that our thinking is depicted as rationally responsive to the facts themselves, in the way that is required for its possession of empirical content to be intelligible.'' It is just another way of putting this to accuse me of a slide, from the requirement on which Brandom thinks he and I agree, that our empirically contentful thinking be rationally constrained by the empirical facts themselves, to the requirement Brandom questions, that the thinking be rationally constrained by experience. What this accusation brings out is that Brandom is wrong in thinking he agrees with me on the "rational constraint constraint". When I suggest that the very idea of empirical content requires that our thinking be rationally constrained by the facts, I mean that to be a case of what I described, above, as "responsiveness to rational constraints as such". And just because of its externalism, Brandom's alternative to my exploitation of experience does not give us responsiveness to rational constraints as such. That would require the status as a rational constraint of what is responded to to be in the view of the responder. But as Brandom makes clear, the rational relation between features of the environment and what he wants to be entitled to regard as observational reports of them -the rational relation that, according to him, constitutes the fact that the supposed reports are answerable to those features, and hence intelligible as reports of them- comes into his story only from the point of view of an interpreter of the supposed reports. That is the position's externalism. (He writes: "From the point of view of the interpreter,. . . the relation between the facts and the reports or perceptual beliefs is not merely a causal one, but also one rationally assessable in terms of the truth of those reports or beliefs, relative to the independent facts.") This means that Brandom's alternative does not, as he claims, fully meet my demand that perceptual beliefs be answerable to the perceptible facts. I need no slide to obliterate Brandom's alternative from my view. I a m responding to Brandom's accusation that there is a gap between the thought that motivates my picture -the "rational constraint constraintn-- and the picture itself, in what is the obvious way to respond to such an accusation; I a m insisting that the motivating thought is stronger than Brandom thinks. Does this merely shift the bump in the rug? Does the complaint now tell against me under an obvious reformulation, to the effect that my stronger ''In his book Brandom leaves room for conceptually contentful thinking without empirical content; see, e.g., p. 234. I need not go into t h a t here.

motivating thought is itself unmotivated, by comparison with the weaker reading of the "rational constraint constraint" that Brandom thought he and I agreed on? Is the weaker reading enough for empirical content not to be mysterious? I think not. Brandom's less demanding reading of the constraint does not yield anything genuinely recogriizable as a rational vulnerability of thinking to the world. In Brandom's alternative picture, the rational constraints that he claims he can represent the world as affording are not responded to, by the person who responds with what Brandom wants to be entitled to count as observational reports, as the rational constraints Brandom wants to be entitled to suppose they are. Their supposed status as rationally related to the supposed reports comes on the scene for the interpreter, not for the responder. If the presence of a feature of the environment reveals that a response, of a sort whose members Brandom wants to be entitled to see as reports, satisfies a rational constraint on reports as to whether or not things are thus and so, that fact does not come into view for the subject who supposedly responds to the feature of the environment with such a report (at any rate it is not in her view qua responder to the feature). But how does this differ from saying that according to this picture, the fact that is supposedly reported, in what is supposed to be an observational report, is not in the view of the subject who responds with the supposed report? From the point of view of the responder, the response Brandom wants to be entitled to see as an observational report degenerates, just because of the picture's externalism, into a blind reaction to she knows not what. How can it be recognizable as a report of an observation, when the fact it is supposed to report -which just is the required rational constraint, from outside the activity of thinking, on reports or judgements as to whether or not it obtains- is placed in the story in such a way that it is not even available, as the rational constraint it is, to the supposed reporter? If it is not available to the responder as the rational constraint it is, how can it intelligibly be available to her as the fact it is? Those are not two distinct characters it has. So far I have left unquestioned the assumption that an interpreter can observe environmental circumstances and a responder's responses to them -so that the interpreter can assess the relation between circumstances and responses in terms of concepts such as that of reliability. But I have urged that in the context of Brandom's externalism this appeal to reliability does not help to make it intelligible that the responses are reports of the circumstances. Their status as reports is rendered positively unintelligible, because the

26. REPLYTO GIBSON, BYRNE,AND BRANDOM 295

facts supposedly reported are excluded, in their capacity as external rational constraints, from the view of the responder qua responder. If it were clear how a response so pictured could be a report, no doubt there would be no problem about an interpreter's assessing it for truth. But the picture's externalism undermines the antecedent of that conditional. It would not help with the problem I am posing for Brandom to note that the two personae, those of responder and interpreter of responses, can coincide in a single individual. The assumption that we can credit the people Brandom wants to see as interpreters with observational capabilities was only provisional. The difficulty arises again about how Brandom can be entitled to the idea that an interpreter (or, better, the persona he wants to be entitled to see as that of an interpreter) is in touch with the relevant aspects of reality (the environmental circumstances supposedly reported on, and, now in addition, the responses that Brandom wants to be entitled to see as reports). It can only be qua responder -qua capable, herself, of responses that are intelligible as reports- that a person who is s u p posed to be an interpreter comes into OUT view (if she does, which I am bringing into doubt) as able to observe the relevant aspects of reality: able to adopt, as a result of suitable impacts from the environment, mental postures with the appropriate empirical content. And here as before, Brandom's picture leaves the fact itself, as the external rational constraint that it is on the activity of deciding what to believe, out of her view qua responder. That means that the supposed interpreter's observational hold on reality is in turn made unintelligible by the picture's externalism. Let us acknowledge, by all means, that the persona that is supposed to be that of an interpreter can coincide with the persona Brandom is trying to entitle himself to see as that of a maker of observational reports. This cannot help with my difficulty about the latter, if the persona that is supposed to be that of an interpreter is not itself intelligibly in observational touch with any aspects of reality. It should be apparent that I am not impressed with Brandom's suggestion, in a diagnostic vein, that the element in my picture that he finds unmotivated reflects a residual individualism. The role I assign to experience (as I conceive experience) is nothing but an expression of the "rational constraint constraint" as I understand it; there is no slide that would need diagnosis. My outlook is, I would have thought, rather strikingly non-individualist, at a different point from the one where Brandom deplores the fact that I do not exploit a divergence of perspectives: I suggest that the very idea of a thinker is unintelligible except in the context of the idea of initiation into

a shared language, conceived as a repository of tradition. When Brandom accuses me of individualism, his point can only be that I do not undertake to make the bearing of thought and speech on the objective world intelligible in the way he does, by appealing to an interplay between interpreters and interpretees. That is the structural idea that, as applied to observational reports and judgements, seems to Brandom to make room for externalist reliabilism as a competitor to my invocation of experience, in satisfying the "rational constraint constraint". It seems strange to suggest that the fact that I have no truck with this kind of thing can be diagnosed as reflecting the absence of this kind of thing from my picture. But more seriously, I have tried to make it plausible that there is nothing to deplore in this absence. As applied to observational reports, Brandom's play with a distinction between what he wants to be entitled to conceive as the standpoints of interpreter and interpretee gives at best a deceptive appearance of making immediately empirical content intelligible. There is no sound basis here from which one could proceed to make mediately empirical content intelligible, by exploiting inferential linkages to observational reports. So Brandom's alternative to my exploitation of experience is not, as he thinks, a serviceable first move towards a satisfactory picture of empirical content in general. I have not objected to the very idea that a notion of justification might get into our picture of some aspect of a subject's life only from an external perspective. In my book (p. 163) I mention a case, in the sphere of practice. A competent cyclist's adjustments of posture and so forth are not merely triggered by cues to alterations in road camber and the like, but are justified in the light of those alterations. So there is an intelligible sense in which what the cues are cues to constitute reasons for the adjustments. But in the normal case a cyclist does not respond to such things as the reasons they nevertheless intelligibly are. They stand revealed as such only from an external perspective, involving explicit knowledge -not usually possessed by competent cyclists, and not acted on even by those who do possess it- about the mechanics of balance and controlled forward motion on a bicycle. Here we have a division of perspectives, structurally parallel to the one that figures in Brandom's picture of observational reports and their justification. For an analogue in the realm of judgement, consider the chickensexers of philosophical folklore, on one version of the story about them. I mean the version in which the ability is, in a certain sense, permanently mysterious to its possessor. He looks at a chick and says whichever of "It's male" or "It's female" he finds himself more inclined to say. When the sayings are tested against, for instance,

26. REPLYTO GIBSON, BYRNE, A N D BRANDOM 297

whether the chick grows up to lay eggs, nearly all of them prove to be true, taken at face value as claims in English. But from the subject's own point of view, the inclinations to say one thing rather than another appear as brute facts, rationally unaccountable. It does not seem to him that he is responding to a pair of distinctive looks. No doubt we have to suppose he is (in a different sense) responding to a pair of distinctive cues, or complexes of cues. But, in this version of the story, the difference at the level of cues is not reflected in a pair of concepts at the subject's disposal, connected with the concepts of being male and being female in such a way that he could express them with such phrases as "looks male" and "looks female". Perhaps he can even learn the true theory about cues; but the cues function below the threshold of his conscious awareness, so the inclinations to say "It's male" or "It's female" are still, from his perspective as a responder to the environment, not equipped with reasons. That is what I meant by saying the ability is permanently mysterious to him; even if he knows completely how it works, that need not carry over into there being reasons for the sayings in his perspective as a responder to the environment. These sayings of "It's male" or "It's female" would relate to the facts about the sex of the chicks he looks at in just the way Brandom says observational reports relate to the facts they report. I can see no reason why these sayings would not count as observational reports by Brandom's lights. Now I would not object if it were claimed that their conformity to Brandom's structure contributes towards making it intelligible that they can be interpreted as claims to the effect that certain chicks are male or female. (To recognize them as claims, we might require them to be put forward with confidence; that might not be intelligible except on the basis of information about reliability, amassed from an interpreter's perspective that the sayer can also occupy.) But it seems plain to me that as I have told the story, these sayings are not intelligible as reports of observation. And I would urge that they are intelligible as claims at all only against a background we easily assume, in which some other sayings the chicken-sexers might go in for (e.g. "It's a chick", "It's laying an egg") would be genuinely intelligible as reports of observation. I would urge that that requires them to be rationally responsive to the reported facts as the rational constraints on judging and reporting that they are, or (it comes to the same thing) rationally responsive to the relevant experiences, as I conceive rational responsiveness to experiences. I am attacking the idea that conformity to Brandom's structure (of course against a suitable background of inferential linkages to

potential non-observational claims) is enough for performances to be intelligible as observational reports. Perhaps it will be urged, in defence of that idea, that I am wrong to suggest concepts expressible by "looks male" and "looks female" (as applicable to chicks) are irredeemably absent from the conceptual repertoire of the chickensexers as I am imagining them. Suppose it is urged that, just because of the conformity to Brandom's structure, it must be possible for such a subject to form and express such concepts. This seems to me to be deeply wrong about the concept, or concept-forming device, expressible by "looks.. .". The concept of something's looking thus and so to one is not the concept of the thing's being such that on looking at it one becomes inclined to say it is thus and so. It is not even the concept of the thing's being such that on looking at it one becomes inclined to say it is thus and so with (whether one knows it or not) a right to confidence that if one does say that, and the circumstances are of a kind that figures in a reliable connection between inclinations of this sort and how things are (a connection whose reliability gives one the right to this confidence), one will be speaking truly. That is indeed a specification of a possible concept, exploiting the fact that one can adopt an external assessor's perspective on one's own perspective as a responder to the environment. But no variation on this theme will amount to the concept of looking thus and so. The applicability of such a perspectivally complex concept does not exclude the possibility that, from one's perspective as a responder to the environment, the inclinations to say that something is thus and so that figure in the content of the concept may be a brute unaccountable fact, as with the chicken-sexers as I have imagined them. Whereas when the concept of something's looking thus and so finds application, one's inclination to say that the thing is thus and so is not rationally naked, from one's perspective as a responder to the environment. The performance one is inclined towards is equipped with a reason, which is in view as a reason from one's perspective as a responder to the environment, in the fact that the thing in question looks thus and so. I undertood to return to Byrne. With some discomfort, he credits me with internalism about justification, and my last few paragraphs should have made it clear that that is not quite right. I am not hostile to the idea of applications of a notion of justification that are made only from an external perspective. The internalism that Byrne is right to find in me is not about justification a . such, but specifically about the rational vulnerability of thinking to reality that figures in the "rational constraint constraint". I hope making that clear will disarm some, at least, of Byrne's suspicion of my moves


against Evans and Peacocke, on the question whether deliverances of perceptual states or occurrences, taken to have non-conceptual content, could innocuously play the role of the Given. I do not object to the very idea that beliefs, say, might be displayed to be in good shape from a standpoint of rational assessment (e.g. displayed to be likely to be true) in the light of circumstances their relation to which may have nothing to do with meeting the "rational constraint constraint". I am quite willing to believe non-conceptual content may figure usefully in doing such a thing. I hope that is enough for what Byrne wants. It seems to me to be quite another matter if our topic is what perceptual beliefs are, to use a phrase Byrne sometimes uses, based on (not just displayable to be in good shape in the light of, from a standpoint of rational assessment). I do not concede, as Byrne says, that someone equipped with Peacocke's theory would be in a position to base her perceptual beliefs on experiences with non-conceptual content. The concession is only that a story with perceivers in possession of the theory would (if it were true, which I do not believe) a t least have the right shape to be relevant to the question what their perceptual beliefs are based on. If asked what one's belief that something is square is based on, one might say "It looks square". I make much of the fact that such a response is articulate, as a ground for supposing that the offered basis is conceptual. Byrne objects that this fact about the response "in no way shows that the experience her belief is about has conceptual content". But a thought with this shape seems to me to cut loose from the project I was concerned to reject, that of making out that the belief about the object is based on the experience. On the view Byrne is defending, the belief about the shape of the object is based on a belief about the experience. According to me, in a picture in which the experience's content is non-conceptual, this second belief is not intelligibly based on the experience. I do not claim that the belief about the experience is therefore doomed to being pictured as merely triggered by the experience. I cannot deny this position the externalistic notion of justification that I have been discussing. So the story can be that the belief is in good shape in the light of the experience. Since that is so, according to the story, in the case of the belief about the experience, it can be so (by an obvious transitivity) in the case of the belief about the object; in being based on the belief about the experience, it is in good shape in the light of the experience. But we have not been shown how to make sense of the belief about the object as based on the experience, taken to have non-conceptual content. It is represented as based on a belief about

the experience. Perhaps that is all that is supposed to be meant by saying it is based on the experience; but in that sense beliefs can obviously be based on anything at all, and there is nothing special about experience. If this is the way the story goes, we are no longer in the business of trying to satisfy the motivation for the idea of the Given.


Molyneux's Question John Campbell

1 Phenomenal Experience of Shape in Sight and in Touch Molyneux's Question concerns a man born blind, who can tell by touch which things are spheres and which are cubes, and who now gains the use of his sight. The question is whether he will be able to tell by vision which things are spheres and which things are cubes. The question is about the relations between percep tions of shape in different sensory modalities. Some of the issues which Molyneux was raising have been settled (Brand Bolton 1994, Rock 1984), but there are some problems which remain. I think that one question is whether there is a difference between the phenomenal characters of shape experience in sight and in touch. Of course there are many differences between visual and tactual experience. We are aware of different properties of things, such as colours or textures, through sight and through touch. And there it may be that there is some more general, indefinable but demonstrable difference between visual and tactual experience, a difference in the character of the experience which we use in practice to determine by which sense we are perceiving something; this seems to have been Grice's (1962) view. But we can for the moment set these points aside and ask whether there is any difference between visual and tactual experience of shape itself; whether the shape per-

ception itself has a different phenomenal character in sight than in touch. What makes a perceptual experience an experience of shape? One possibility is that experiences have their own intrinsic geometry which exists independently of the perceiver's relations to the environment, that there is a phenomenal space in which sensations are configured (Strawson 1966). So there would be spatial properties and relations such as 'is a phenomenal straight line', 'is phenomenally to the left of', and so on. On this internalist view, the existence of a phenomenal geometry in our awareness of objects would be a primitive datum, not explained in terms of anything more fundamental. This view leaves it open that primitive consciousness of shape might be quite different in the different sensory modalities, which will all have their own phenomenal geometries. An alternative possibility is that what makes one's consciousness consciousness of shape is the fact that one is using a neural system whose role is t o pick up the shape properties of the objects in one's environment. The geometrical aspects of one's experience of objects will then be constituted by the geometry of the objects in one's surroundings. This is quite a radical externalism, and to bring this out we can contrast it with a more moderate view. On the more moderate view, sensations of shape, and indeed all perceptual experiences, are stratified into similarity classes prior to any environmental circumstances coming into play: they are intrinsically more or less like one another in this or that respect, such as experiential shape or colour. On this more moderate view, though, external considerations may ineliminably come into play when one tries t o say which similarity class of sensations one is identifying: perhaps one can say which sort of sensation one is talking about only by saying that it is the sort of sensation produced by a cube in good light. The basis of this view is the need t o allow for the possibility of illusion. Since I can have an experience of a cube which is not in fact produced by a cube, I need to be able t o say that it is an experience of the sort which is characteristically produced by a cube in good light, and this presupposes a stratification of experiences into sorts which is prior to the question how they are produced, a stratification which is prior to any environmental considerations coming into play. So although this moderate view is externalist about how in practice we identify classes of perceptual experiences -we do it by reference to their characteristic causes- it presupposes a prior, internalist stratification of experiences into similarity classes. The alternative, radical externalism holds that the initial stratification of experiences into similarity classes already demands the

involvement of the environment, that experiences viewed purely internally would be an amorphous mass. The basis of this view is the point that the idea of a purely internal stratification of experiences gives us no way of understanding what it would be for the experiences of different people a t the same time, or the experiences of a single person a t different times, t o be in the same similarity class. This was the point made in Wittgenstein's remarks on the idea of a private language. The alternative is to suppose that the sorting of sensations into similarity classes constitutively demands an appeal to the environment; and then, since the environment is shared by different perceivers, or by the same perceiver at different times, we have a commensurability between the experiences of different perceivers or the same perceiver a t different times. The problem for this radical externalism is the possibility of illusion, since we cannot appeal directly to environmental factors to explain the sense in which an illusion of shape, for example, is experientially similar to a veridical perception of shape. I gestured a t a solution t o this difficult problem above, when I said that what makes one's consciousness consciousness of shape is the fact that one is using a neural system whose role is to pick up the shape properties of the objects in one's environment; so we would capture that similarity by appealing to the sameness of the underlying neural system, externally individuated. But that is only a gesture. On this radically externalist picture, shape perception will be amodal, since the different senses will be picking up the very same properties of the objects around one. Insofar as we are externalist about shape perception, we have to think of it as amodal. For insofar as we are externalist about shape perception, we have t o think of experience of shape as a single phenomenon, in whatever sense-modality it occurs, individuated by the external geometrical property. For it is in fact the very same properties that are being perceived by sight as by touch. One way to put this point is to say that if there is any question for the perceiver about whether it is the same property that is being perceived through vision as through touch, it must be possible to explain what the basis of this doubt is. And insofar as we are externalist about the character of shape perception, then there is nothing in the character of the experience itself t o ground a doubt as to whether it is the same properties that are being perceived through vision as through touch. This issue affects the way in which we interpret the primitive recognition of sameness of shape perceived in different modalities that we find even in very young children and in animals. On one view, we are simply exploring the ad hoc functional connections between differ-

ent perceptions. There is no rationality about, for example, primitive transference of what one has learned about a shape as presented in one modality t o a presentation of that shape in another modality: sameness of the property perceived through sight and touch is not transparent to the perceiver. It may be true that visual perception of a shape and tactual perception of that shape have the same functional connections t o behaviour and imagery, but that could be so without it being apparent to the perceiver that it is so. If we are internalist about the phenomenal character of shape perception, that is the obvious interpretation of the data. For the radical externalist, however, there is no difference in the phenomenal character of shape experience in sight and in touch. The sameness of property perceived in sight and touch is transparent to the subject, and crossmodal transfer is a rational phenomenon.

2

What Shape Perception Is

This gives one way of setting up the question of the sameness or difference of shape perception in sight and touch. But you might object that this is quite the wrong way in which t o set up the problem. You might object that we have to think of shape properties in terms of their causal significance, and that shape perception is perception of this causal significance. In this section I will develop this causal approach to shape perception. Then I will argue that the approach is incomplete, and that there is still a crucial issue that is addressed by externalism about phenomenal experience of shape. Before setting out the causal view of shape perception, let us look a t the causal view of shape properties. One model for the description of a property in terms of its causal significance is provided by the familiar notion of a power of an object. A power is an input-output state of a thing, a disposition that it has to behave in a particular way in particular circumstances. For example, whether something is brittle is a matter of it tending to break when pressure is applied, for something to be striking is for it to be liable to draw notice, for someone t o be drowsy is for him to be liable to fall asleep. These are input-output states. The canonical description of a power is: In trigger circumstances C , the object responds in way R. When we describe the property in this way, we say what it is. If we can say what two properties A and B are in this format, then A is identical to B if the replacements for C and R are the same in the two descriptions. In the examples I have given so far, saying

what the property is might be taken to be a matter of defining the predicate we ordinarily use in ascribing the property to an object. But in general, saying what a property is need not be a matter of defining a term in ordinary use. Saying what magnetism is, or what gravity is, need not be a matter of defining a term as it is used in everyday English. But magnetism and gravity might nonetheless be powers. Or, if we want to keep the notion of definition, we might say that here we have 'real definitions', and that the above is the canonical form of the 'real definition' of an input-output property. Shapes are not input-output states in this sense. The shape of an object, as spherical or cubical, is not something that affects the behaviour of the object in a uniform way. The shape of an object is, of course, causally relevant to how it behaves. But exactly how it affects the behaviour of the object depends on what other properties the object has. For example, something chair-shaped has the capacity t o support a seated human, but only if it is the right size, and only if it is made of rigid rather than thoroughly flexible materials. But being chair-shaped matters for being capable of supporting a seated human, being about the right size and made of rigid materials is not enough. This suggests that we might think of a shape property as something which affects the causal powers of the object that has it, but does so in a way that depends upon just which other properties the object has. For example, being spherical together with being hollow and made of a resilient material will mean that the thing is relatively easy to propel through the air, whereas if it is spherical, solid and made of a dense material it will be harder to get moving but travel for longer. We could put this more formally by saying that a 'conditional power' of an object is something which means that the object will have a certain power, if it has certain further properties. The canonical description of a conditional power is: If the object has further properties Q1,. .. ,Q,, then the object has power P. The property of being cubical cannot be identified with any one conditional power. Being cubical involves having many conditional powers. Nevertheless, we could still identify being cubical with possession of a cluster of conditional powers. This is Shoemaker's (1984) theory of properties in general. Shoemaker holds that a property is a cluster of conditional powers. He does not hold the converse: that every cluster of conditional powers is a property. To constitute a property, a cluster of conditional powers must have a certain causal unity, not just any tossed-together collection of conditional powers will do. A cluster of conditional powers will constitute a property

only if it contains one or more conditional powers which are such that it is a consequence of causal laws that any object which has those conditional powers will also have all the other conditional powers in the cluster. In what follows I shall be developing this view, but going well beyond anything in Shoemaker's article. On this view, there can be no more to the perception of shape than the perception of an object as having a certain cluster of conditional powers. But this needs some explanation. On the face of it, we do not perceive the shape of a thing as a collection of unsubstantiated threats and promises as to which powers it will take on in various hypothetical circumstances. We perceive the substance behind the threats and promises. So we need some preliminary explanation of what it could mean to perceive an object as having a cluster of conditional powers, before we can understand the argument that that is all we perceive the object to have. We can begin on this by thinking about perception of simple powers. How does the fact that one perceives an object as having a simple power affect one's behavioural response to it? Suppose you perceive a cat as ready to pounce, for example. How you react will in part depend on what you make of the pouncing. For example, you might want to take advantage of the relative vulnerability of the cat to outside attack during the pounce. So then you will position oneself ready to take advantage, should the trigger to the pounce appear. I n the case of a simple power, then, the behavioural response is conditional on whether one takes the trigger conditions t o be met. It may involve acting to ensure that the trigger conditions are met, or to ensure that they are not met. Suppose now we consider how perception of an object as having a conditional power would affect one's behavioural response to it. What we need is a conditionalisation of the account just given of the behavioural response to perception of a simple power. Suppose that you perceive an object as having a power P, conditionally on its further possession of properties Q 1 , . . . , Q,. Then whether you have the behavioural response to the object which is appropriate to its having the simple power P will be conditional on whether you also know the object to have properties Q1, . . . , Q,. And depending on whether you want the thing t o have the power P, you may act to ensure, or to prevent the thing's having properties Q1,. . . , Q,. Finally, suppose we ask how perception of an object as having a cluster of conditional powers will affect one's behavioural response to it. This will be a matter of having a family of complex conditional behavioural responses of the type just indicated, one for each conditional power in the cluster. That one has grasped the causal unity

of the cluster will show up in the causal unity of one's behavioural responses. That is, there will be one or more of the family of responses which are such that if they are activated by perception, all the rest of the responses in the family are activated. There may be a single complex mechanism underlying all that family of behavioural responses. We could give an account of shape perception which has the same general type of structure but which focuses on imagery rather than on action and manipulation. That is, we could have an account which sees the perception as being located in a space of imagistic reasoning. Perception of an object as having a simple power would be a matter of the kinds of imagistic transformation to which one was disposed to subject the image. So, for example, perceiving the cat as ready to pounce would come to this: if you asked yourself, 'What will happen if a mouse appears?', by imaging the appearance of a mouse on the scene, you would follow that up by imagining the cat to leap. This transition would be internal to your imagistic reasoning, it would not be a matter of having your imagery controlled by some explicit verbal reasoning. How would perception of an object as having a conditional power affect the role of the perception in your imagistic reasoning? Here we need a conditionalisation of the account just given of perception of a simple power. Suppose that you perceive an object as having a power P, conditionally on its further possession of properties Q1,. . . , Q,. What this comes to is that if you perceive it also to have Q1,. . . ,Q,, then you will imagistically manipulate the perception in the way appropriate to its having power P . And even if you do not initially perceive the object to have Q1,. .. ,Q,, if you then make the imagistic transformation of imagining it to have Q1,. . . , Q,, you will then in further imagistic transformations proceed in the way a p propriate to the object's having power P . That is, if you imagine the trigger to appear, you will further imagine the response to have been produced. Finally, suppose we ask how perception of a cluster of conditional powers would affect the role of the perception in one's imagistic reasoning. This will be a matter of the perception being related in the way just indicated to a whole family of imagistic transformations, depending on just which further properties one perceived the object to have, or went on to imagine it as having. And the causal unity of the cluster of conditional powers would show up in the causal unity of this set of responses in one's imagistic reasoning. That is, if you located the perception thus and so with respect to one subset of the relevant family of imagistic transformations, it would automatically

be located appropriately with respect t o all the rest of the family of imagistic transformations. Whether we do it by looking a t the behavioural implications of the perception, or by looking at the role of the perception in imagistic reasoning, then, it does seem that we can make sense of the idea of perceiving a shape as a family of conditional powers. The problem now is not t o find what can be meant by saying that we perceive a shape as a family of conditional powers. There would only be a problem if we thought that perception of shape comes to any more than this; if we thought, for example, that perception of shape is perception of the categorical ground of a family of conditional powers, a point to which I now turn.

3

Shape Properties as Categorical

On the face of it, Shoemaker's account is analogous to functionalism about psychological properties. Suppose we take a particular shape property, and write down, in the above form, all the conditional powers which constitute it. This gives us a fragment of a theory, which says that there is a unique shape property such that.. . . In stating the theory, we will need to appeal to shape properties, in saying what powers the property may confer upon an object. For those powers will include interactions with other objects, and just what happens in those interactions will in turn depend in part upon the shapes of the other objects. Shoemaker's theory will in turn apply to those shape properties. Ultimately we will arrive a t a global theory of the physical significance of shape properties. We can then construct the Ramsey sentence of the theory, which says that there is a set of properties satisfying the open sentence which results from replacing the shape predicates of the theory by schematic letters. From this we could extract non-circular definitions of each of the shape properties. As Shoemaker presents it, this is a global theory covering all physical properties of objects, not just shapes, but I shall concentrate on shape. Once the causal view of properties is stated, there seems to be an obvious alternative. A conditional power is itself a power. As Shoemaker puts it at one point, a conditional power is a power to affect the powers of the object. It is a higher-order power. Now in general, we think of powers as having categorical grounds. A power of an object has to do with how the object is in various merely possible worlds. These truths about what goes on in various other possible worlds have to be grounded in the facts about the actual

world, otherwise we have let these other possible worlds take on lives of their own. So if an object has a power to affect its other powers, that power must itself have a categorical ground, a ground in what actually happens. So an alternative hypothesis to Shoemaker's about shape properties is that they are not clusters of conditional powers. Rather, they are the categorical grounds of clusters of conditional powers. Shape properties are the grounds of higher-order powers. This view is not as straightforward as it may seem. One standard example of the relation between a disposition and the ground of the disposition is the relation between the solubility of salt and those aspects of the molecular structure of salt in virtue of which it dissolves in water. But on the face of it, this is a relation between different powers. It is a relation between the dispositional property of the salt, its solubility, and the dispositional characteristics of its constituent particles which are triggered when the salt is put in water. This does not seem to be a helpful model for the relation between the higherorder powers that a thing has in virtue of being a certain shape, and the shape of the thing. One way to get at the intuitive thrust of the alternative to Shoemaker here is to recall the example of the square peg and the round hole. Why can't the peg get through the hole? As Putnam pointed out long ago, it is quite wrong to look for an explanation of this phenomenon at the level of quantum mechanics: a finely detailed story about the particles constituting the peg and the board is beside the point. We have to look for an explanation at the macroscopic level. And at the macroscopic level, the explanation is this. Because the peg and the board are both made of rigid material, and the length of the side of the peg is the same as the diameter of the hole, the squareness of the peg means that its diagonal is larger than the diameter of the hole, so the peg cannot get through. What, on Shoemaker's account, does the appeal to the squareness of the peg come to here? On this account, it is an appeal to a cluster of higher-order powers that the peg has. Now these will mostly be irrelevant to the problem at hand. But there is, in particular, the following: one of the cluster of higher-order powers is that if the object is rigid, then it will have the power to be unable to get through a hole with the same diameter as the length of its side. So since the object is rigid, it has the power to be unable to get through a hole with the same diameter as the length of its side. That is the explanation of why the peg cannot get through the hole. In effect, we have a virtus dormitiva explanation here. But there is a difference between this virtus dormitiva explanation and the original. Someone who explains the fact that opium puts people to sleep by saying that it has a dormitive power is at

any rate saying that the opium has this dispositional characteristic and holding open the possibility of explaining its possession of this disposition by appealing to microstructural features of the opium and human physiology. But in the case of the peg and the hole, we have, on Shoemaker's account, given a virtus dormitiva explanation and we know already that no deeper explanation in terms of microstructural properties will be forthcoming. That was Putnam's point. This is not the only level at which explanation seems to be missing. We could have a theory about how in general the shape of an object affects its behaviour. The particular shapes of individual objects will be clusters of higher-order powers. But these clusters of higherorder powers are all systematically related. For example, suppose that you have a medium-sized object that is just too heavy to lift, and you have to get it t o its destination by tipping it over and over along the ground. And suppose the object is flat, with a regular polygon as cross-section. If it is a triangle it will be quite hard to tip it over each time. If it is a square it will be easier, if a pentagon easier still. And if it has a thousand sides you can simply roll it. There is a generalisation to be had here, about the relation between the number of sides of the polygon and the ease with which the thing can be moved. On Shoemaker's view, the generalisation is simply a regularity holding between clusters of conditional powers. But that is not how we would ordinarily think of it. We would think that there are two levels of description here, one concerning the various shapes and the relations between them, and we would think of the level a t which we talk about the relations between the shapes as explaining what we have a t the second level of description, namely the relations between the conditional powers of the objects which have the shapes. On Shoemaker's account, however, we have only a single level of description, at which we describe the various clusters of conditional powers, and there is no explanation to be had of the relations between them. The alternative to Shoemaker is to suppose that the shape property is the ground of the higher-order powers, so that we are not giving a virtus dormitiva explanation a t all when we appeal to the shape of the peg in explaining why it does not get through the hole. We are appealing to the categorical property which explains why, at the macroscopic level, the thing behaves as it does. But this can hardly be thought to be the end of the story. The problems are just beginning at this stage. For now we have to explain what it means to say that the shape is a categorical property, how it can be that we have an explanation of the behaviour of the peg so readily.

From Shoemaker's perspective, the natural riposte is that all that has happened is that in explaining the behaviour of the peg we have appealed to a higher-order power of the peg, giving a virtus dormitiva explanation, and in saying that the property is 'categorical' we have declared ourselves to be quite happy with this state of affairs, and to have no intention of looking for a deeper microstructural explanation. Explanations come to an end somewhere, and all that is accomplished by labelling shape properties categorical is that we declare ourselves t o be content for explanations to end with appeals to shape. But that is quite consistent with holding that shapes are higher-order powers. What is Shoemaker's argument for his view? According to Shoemaker, the alternative to supposing that shape properties are clusters of conditional powers is to suppose that the identity of properties consists in something logically independent of their causal potentialities. This opens a number of possibilities. It ought to be possible for there to be properties which make no difference whatever to the behaviour of the things which possess them. There could be different properties that make, under all possible circumstances, exactly the same contribution to the causal powers of the things that have them. The potential of a particular property for contributing to the production of causal powers might change over time. Consequently, we could have no knowledge of the properties of a thing -since all we can know is the behaviour of the thing- and we would have no way of singling out a property in order to name it, and even if we did somehow manage to christen a property, there would be no way of there would be no way in which we could know that we had encountered the same property again. The crucial point in the argument is the idea that all we can know is the behaviour of a thing. The obvious riposte, from the proponent of the alternative view, is that we immediately perceive shape properties, and that what we perceive in perceiving shape properties are the grounds of higherorder powers. So there is nothing ineffable about shape properties so conceived; they are just what we ordinarily see and touch. The central issue is whether the connections between shape perception and action, or between shape perception and imagistic reasoning, exhaust the content of the shape perception. One way to put Shoemaker's view is as the view that these connections do exhaust the content of the shape perception. Then we have the functionalist account of our ordinary way of thinking about shape properties: shape properties are properties with a certain functional role, a functional role which is in effect specified by the pattern of connections between shape perception and action, or the pattern of connections

between images in imagistic reasoning. On the alternative view, these connections do not exhaust the content of shape perception. In addition to these connections, shape perception provides knowledge of the shape property as the ground of all the functional patterns that are specified. The problem then is to explain just what this further knowledge provided by shape perception is, and to explain its relation to these further connections and patterns of functional role. You might suggest that what makes it the case that we ordinarily perceive the categorical grounds of conditional powers, rather than families of conditional powers is that there is more to perception of shape than the kinds of functional connections indicated so far in that for ordinary humans, perceptions of shape can be input to explicit explanations of the behaviour of the objects around one, for example in explaining the inability of the peg t o get through the hole. But one can certainly perceive shapes without being able to engage in these explanations. You could show the ability to discriminate shapes perceptually, respond behaviourally to shapes in appropriate ways, and engage in imagistic reasoning about shapes. You could do all that, and still have no capacity to appeal to shape in giving explanations of the phenomena around you. An animal could display all the phenomena here, and still the question of how things are to be explained might never be an issue: for the animal, the question of explanation simply does not arise. You might suggest that this is a difference between animal and human perception: that animals perceive shapes as clusters of conditional powers, whereas humans perceive them as the grounds of clusters of conditional powers. But it seems worth pursuing the question whether an account at the purely perceptual level can explain how it could be that we are perceiving shapes as categorical.

4

Primitive Consciousness of Shape

One way to approach the question is to ask whether we can make anything of the idea of a primitive consciousness of shape, prior to grasp of the causa! role of the shape of an object. This would have to be perception of shape as categorical, since there is no grasp of causal role. There would be little room in such a use of shape predicates for the possibility of correction of error in judgements about the presence or absence of a particular shape. There can be illusions of shape -the straight stick that looks bent in water- but not much sense could be made of the possibility of illusion at this primitive level. We can check whether something really is a cube

by measuring it, we can check whether something is approximately spherical by rolling it, but these procedures would not be available at the primitive level. In using such tests we are exploiting the fact that the shape of an object typically results in possession by the object which has it of various causal characteristics, and that is precisely what we trying not to take on board here, at the primitive level. But perhaps we can nonetheless talk about a primitive experience of shape even in the absence of an ability to recognise illusions. For example, we might train an animal to respond in a particular way to a presentation of the shape 'A', and it might succeed in the tasks we set it even if it was quite unable to distinguish between veridical and illusory presentations of an 'A': we would still have here a primitive kind of shape perception. The animal would have some grasp of the contingency arbitrarily set up by the experimenter -for example, to press the bar on the left when an 'A' is present- but this contingency could not be regarded as one of a cluster of conditional powers constituting the shape property. If we are to talk about a primitive consciousness of shape as categorical, we have to keep in mind the systematic character of shape perception. That is, perception of various shapes is not a matter of exercising a set of disjoint perceptual capacities; there has to be a background geometry for perceptions of various different shapes. It is instructive to compare shape predicates with natural kind terms. In the case of 'gold', for example, you might first be conscious of the characteristic appearance of gold, and then later learn which conditional powers gold has. So there certainly is such a thing as primitive consciousness of gold, even though at this level one would have little understanding of the possibility of perceptual illusions of gold. One disanalogy between natural kinds and the case of shape predicates is that there is no analogue of the background geometry in the case of natural kinds: the various appearances of the various natural kinds really are disjoint, whereas the appearances of the various shapes are all systematically related. Another disanalogy is that there does not have to be any very apparent connection between the appearance of gold and the nature of the conditional powers which something has in virtue of being gold. In contrast, in the case of squareness, for example, the appearance of the thing, the way in which one is primitively conscious of the property, does seem to be closely related to the conditional powers a thing has when it is square. There are four singularities, the corners, in the appearance of the square, and there are four singularities, the corners, in the causal behaviour of the square. What these disanalogies suggest is that the assignation of causal significance to shapes is not done on a shape-by-shape basis:

it is done by giving physical significance to the background geometry of primitive consciousness of shape. The relation between a particular experienced shape and a particular set of conditional powers would then be a consequence of a more general physical geometry. The kind of primitive consciousness of shape I a m describing is, though, extremely primitive. There would be none of the complex of functional connections to action and imagistic reasoning which I described earlier. That means that we have to consider the question what makes it so that this consciousness is consciousness of shapes. Recall the radical externalism which I began by describing. On this view, what makes one's consciousness consciousness of shape is the fact that one is using a neural system whose role is to pick up the shape properties of the objects in one's environment. The geometrical aspects of your experience of objects are constituted by the geometry of the objects in your surroundings. On this view, the character of experience as perception of shape can be secured without having to consider the relations of the perception to imagery and action. You could have the right perceptual relations to the geometry of the objects around you even though your perceptions did not have the kinds of relation to action and imagistic reasoning that I described earlier, and in that case you would have a primitive perception of shape as categorical. This might be the case of, for example, the animal which can be trained to respond in a particular way to the presence of an 'A' shape, even though it may be incapable of much in the way of complex behavioural responses or imagistic reasoning involving that shape. Earlier I also remarked the possibility of an internalist account of shape experience, on which sensations are configured in a sensational space, and we have properties and relations such as 'is a phenomenal curve', 'is phenomenally parallel to', and so on. This would also suggest the possibility of a kind of primitive awareness of shape, on which it could be prior to the complex of functional connections to action and imagistic reasoning described above. And this would be an awareness of shape as categorical. But these sensational characteristics are supposed to be known from one's own case, it is your own experience of shape that gives you knowledge of what these sensational characteristics are. So this view has no way of explaining what it is for sensational experiences of shape to be the same or different in different observers, or for a single observer over a period of time. The point about both of these accounts, though, is that they suggest how there could be more to the content of shape perception than the complex of functional connections to action and imagistic

reasoning that I described in $2. If the content of shape perception is exhausted by this complex of functional connections, then it is hard to see how perception of shape could be perception of shape as categorical, rather than perception of shape as the cluster of conditional powers specified by the complex of functional connections. If, on the other hand, there is a further dimension to shape perception, what I have been calling primitive consciousness of shape, then this could only be perception of the shape property as categorical, and this dimension of shape perception will be what explains our having the conception of shape properties as categorical. As I said in discussing the kinds of causal explanations in which we appeal to the shapes of things, we do take ourselves to have the conception of shape as categorical. One reason for the appeal of the idea of primitive consciousness of shape is that it explains how we could have come by that conception. Both the externalist and internalist aim to give an account of this primitive consciousness of shape, the further dimension of shape perception. I want finally to look at the implications of these views for the ability to recognise sameness of shape perceived in sight and in touch.

5 Cross-Modal Equivalences I said earlier that the internalist view of primitive consciousness of shape leaves it open that the sensation of shape may be quite different in different sensory modalities. But the reason for this is that the internalist has no general account of sameness of difference of sensation of shape. That is obvious when we reflect on the question whether different people, or a single person at different times, have the same sensations of shape. On the internalist account, we simply have no way of answering this question. The obvious way of providing an account is to move in the direction of the externalist, drawing in environmental considerations, so that sameness or difference of experience depends on the sameness or difference of the geometrical properties of the objects perceived. But then, as I said earlier, this suggests that there will be a sameness in primitive consciousness of shape across the sensory modalities, since it is the very same geometrical properties that are being perceived through sight and through touch. Sameness of shape experience across person, time and sensemodality all hang together. Recognition of sameness of shape across sensory modality is a very primitive phenomenon; to give some sense of it, it may help to give

a brief review of some of the ways in which it can show up (for further discussion, see Streri 1993). There are transfer-of-learning paradigms, in which an animal is trained to give a certain response when a particular shape is presented in one modality. In the test phase, the animal is presented with that shape in a different sensory modality, and transfer has occurred if it produces the response. There are paradigms in which the same or different shapes are simultaneously presented in both modalities, and the subject is interrogated about the relation between them. For example, an ape may be able to see one object through a window while at the same time manipulating two objects, one in each hand, one of which is the same shape as the seen object and the other of which is a different shape to the seen object. The task set for the animal is to pull on the felt object which is the same shape a s the seen object. The animals are trained over a period of weeks to give the same-shape response for a particular pair of shapes, and then in the test phase they are given a new set of shapes. Success is managing to give the correct 'same shape' response for the new shapes. And animals do manage this. The same kind of set-up can be used where the matching is touch to vision rather than vision to touch. Here the animal is given one object it can grasp but not see, and simultaneously shown two objects behind a transparent window. The animal has to indicate which seen object is the same shape as the felt object, by pressing on the appropriate window. Studies on children tend to use habituation. For example, the child may be simultaneously presented with an object it can see but not touch and an object it can touch but not see. Children have their attention drawn to these objects. They lose interest more quickly if the objects are the same shape than if they are different shapes. Alternatively, in one study using a complex of mirrors, children were presented with a seen object of one shape at the place where they could feel an object of a different shape to be. They registered their appreciation of this by surprise or distress reactions. Finally, there are paradigms in which the same or different shapes are presented successively in different sensory modalities. For example, one set-up exploited the fact that infants prefer to grasp and manipulate objects which make a noise when shaken than those which do not. First the infants were shown two objects visually, out of their reach. Then one of the objects, which made a noise when shaken, was presented to the child, behind a screen where it could manipulate it but not see it. Then the child was shown the two initial objects again, and its preference noted. There is a simpler family of approaches in which the child is allowed to manipulate an object it cannot see, then visually presented with the object together with

a new object. If children systematically prefer to look at one rather than the other of the new and old objects, that shows cross-modal transfer. And the experiment may test matching in the reverse direction, so that we consider whether it identifies the sameness of shape of one of a pair of grasped objects to an object that was seen earlier. On the radically externalist view of primitive consciousness of shape, the phenomenal experience of shape is the same in sight and in touch. It will be in consequence of this amodal character of shape perception that this cross-modal transfer occurs, and cross-modal transfer will be a rational phenomenon. No ground for doubt as to whether one is perceiving the same shape properties through sight as through touch will be provided by an intrinsic experiential difference between the geometry of vision and the geometry of touch, for there is no such intrinsic, phenomenal difference. There may still be further phenomenal differences between and touch, which may tip us off as to which sense we are using, but these will be extrinsic to the geometrical characteristics of the perceptions. And it will be possible for different geometrical descriptions to be given of the very same shapes in sight than in touch; indeed, two different visual perceptions of the same shape may give different geometrical descriptions of it, as when one object is a rotated version of another, similarly shaped thing. In this case it may still be informative to be told that the shapes are the same; so if vision and touch give different geometrical descriptions of the same shape, it may still be informative to be told that it is the same shape one is seeing as touching. But given the unity of the underlying, externally constituted geometry of the two senses, it will be possible for the perceiver to determine a priori that it is the same shape that is in question. There is no return here to the incommensurability of shape perception in the two modalities which the internalist tries to ground in a root difference in the sensational character of the experiences of sight and touch. On this radically externalist view of primitive perception of shape, grasp of the causal significance of an object's having a particular shape will be a derivative phenomenon, dependent on one's giving a physical interpretation to the perceptual geometry. And it will be in consequence of this amodality of primitive perception of shape that we in practice assign the same causal significance to shapes perceived through any modality.'

'Earlier versions were given a s talks a t Edinburgh, St. Andrews and Stirling Universities, a t a seminar in Oxford and t o t h e Oriel discussion group. This talk was prepared with t h e support of t h e British Academy.

Brandt Bolton, Martha. 1994. 'The Real Molyneux Question and the Basis of Locke's Answer'. In G.A.J. Rogers (ed.), Locke's Philosophy: Content and Context. Oxford: Oxford University Press. Grice, H.P. 1962. 'Some Remarks About the Senses'. In R.J. Butler (ed.), Analytical Philosophy. Oxford: Blackwell. Rock, Irving. 1984. The Logic of Perception. New York: W.H. Freeman. Shoemaker, Sydney. 1984. 'Properties and Causality'. In Sydney Shoemaker, Identity, Cause and Mind. Cambridge: Cambridge University Press. Strawson, P.F. 1966. 'Kant's Theory of Geometry'. In P.F. Strawson, The Bounds of Sense. London: Methuen. Streri, Arlette. 1993. Seeing, Reaching, Touching. Cambridge: MIT Press.


Comments on John Campbell, "Molyneux's Question" Brian Loar

1 What Is Radical Externalism? John Campbell proposes a radical externalist thesis about shape perceptions and their phenomenal character. The central issue is "whether there is any difference between visual and tactual experience of shape itself; whether the shape perception itself has a different phenomenal character in sight than in touch". Campbell's thesis is that shape perception is amodal: "we have to think of experience of shape as a single phenomenon, in whatever modality it occurs, individuated by the external geometrical property". Now from these remarks alone one does not get a firm grip on what is being individuated. When should we count two perceptions as instances of a single phenomenon or as having the same phenomenal character? It is difficult to evaluate radical externalism merely as a phenomenological thesis about how to type-individuate experiences of shape or phenomenal character. For, as Campbell notes, even on a radical externalist view "there may still be further phenomenal differences between sight and touch", and the question is why these differences do not matter in determining sameness of experience or phenomenal character. Campbell's point is that those differences do not introduce elements of doubt into inferences between visual and

tactual perceptions of the same shape. They do not interfere with the transparency of such inferences. So radical externalism about individuating experiences and phenomenal character rests on an epistemic criterion, one that is, on the face of it, Fregean. This rather suggests that radical externalism can be formulated as a thesis about certain visually and tactually based concepts of shape, and their content-equivalence. Some wish to distinguish sharply perceptual content from any sort of conceptual content. But there seems to be a thesis about concept-individuation that is equivalent to Campbell's epistemic thesis about experience and phenomenal character. Perhaps he will regard the following as capturing the flavor of his view. We can think of recognitional concepts, of shapes, of colors, of natural kinds, as type-demonstrative concepts that are grounded in perception; they point from certain perceptual perspectives to certain recurring properties: "that sort of thing". The question of the rationality of cross-modal transfer of shape perception becomes the question of the a priori rationality of certain judgments involving recognitional concepts: this shape is identical with that shape, where the former concept-token arises from touch and the latter from sight. The radical externalist thesis would then be that those two recognitional concept-tokens are the same concept just in case they refer to the same shape. And so Campbell's remark that radical externalism implies that shape perception is amodal would then imply that recognitional concepts of shape are amodal; they are not distinguished by the perceptual modalities via which their references -certain shapes- have as it happens been determined. Any two recognitional concept-tokens that pick out squares are tokens of the same concept. Why take a radical externalist view of shape perception? It seems that Campbell offers two rather distinct motivations. a) We cannot individuate perceptual states without mention of references; this is common to all forms of externalism. And no further internal phenomenal factors can be required in individuating perceptual states, because of Wittgensteinian scepticism about the interpersonal ascription of internally determined phenomenal states. Campbell does not tell us how generally radical externalism applies: is the thesis restricted to concepts of shape? The Wittgensteinian things he says in rejecting internalism and moderate externalism apparently imply that it cannot be so restricted. We will return to this. b) Radical externalism explains the rationality of cross-modal transferences.

2 The Rationality of Cross-Modal Transference Campbell considers two possible views of the rationality of crossmodal transference. On one, there are visual and tactual functional equivalences, "but that could be so without it being apparent to the perceiver that it is so". "On the alternative view , the sameness of property perceived in sight and touch is transparent to the subject, and cross-modal transfer is a rational phenomenon". Now this is a rather misleading pair of alternatives. You do not have to be a radical externalist to hold the second view. The mere fact that you are aware of your cross-modal inferential skills, and that your knowledge is in some way a priori, does not on its own imply a radical externalist individuation of shape perceptions or recognitional concepts. The picture of the amodal transparency of the identity of what we perceive when we see a triangle, and when we explore one by touch, is in some ways an apt metaphor, but it is not obvious that we should take it fully seriously. Intuitively, seeing and touching something triangular are of distinct and conceivably separable experience-types, and no amount of sceptical Wittgensteinian argument about phenomenal states can make this unintuitive. There are moreover general epistemological models of rationality that, when applied to innate cross-modal inferential tendencies, would count them as a priori rational without our supposing that this rationality consists in an amodal grasp of, say, an object's triangularity-in-itself. Innate cross-modal inferential tendencies would have as much claim to a priori rationality as, but no more than, any other innate conceptual or inferential dispositions. If they are supposed to be a priori in the strong sense that we cannot imagine such inferences breaking down, then the view goes too far. So I find it somewhat obscure what distinctive advantage radical externalism is meant to have (if we put aside the apparently independent Wittgensteinian point) in explaining intuitions about the rationality of cross modal inferences.

3 Categorical Shape Properties It is not clear (to me) what role categorical shape properties, and the primitive consciousness of shape, play in Campbell's overall argument for radical externalism. Both moderate externalism and internalism can (as he points out) also acknowledge that shape properties are categorical. Those positions are supposedly meant to be refuted by independent Wittgensteinian objections, and the categorical nature of shape properties is on the face of it beside that point.

Now radical externalism is most naturally interpreted in terms of a causal theory of shape concepts, and such a theory is at its most straightforward if shape properties are categorical. But this will be equally convenient for any proponent of a causal account of the reference of shape concepts. And that, clearly, is compatible with supposing that concepts that draw on sight and touch are individuated more fine-grainedly than are their references. The discussion of categorical shape properties serves then to motivate a certain convenient presupposition of radical externalism that does not distinguish it from other accounts of concept and perception individuation. Perhaps that is all it is intended to do.

4

The Viability of Radical Externalism

While radical externalism does imply and hence would explain the rationality of cross-modal transference, it seems to me to imply too much, and hence not to give a correct explanation of the rationality of cross-modal transferences. I will raise a few worries about radical externalism, which are not particularly surprising. I proceed still in some doubt as to whether Campbell's concern with individuating shape-perception or experience is subject to the Fregean constraints on individuating perceptually based recognitional concepts that I will invoke. The objections are independent of internalism and compatible with the neo-Fregean view that recognitional concepts are individuated in part but not entirely by the properties they pick out. Campbell rejects moderate externalism because of Wittgensteinian scepticism about the interpersonal ascribability of internally determined phenomenal states. No further internal phenomenal factors can apparently contribute to individuating perceptual states or the corresponding recognitional concepts. (He does not consider the possibility that perception-types might be individuated in part by external factors over and above the properties they discriminate, factors such as perspective and sensory modality.) But this justification of radical externalism for shape perceptions surely would equally justify radical externalism for all recognitional concepts. And, interestingly, the Wittgensteinian justification would imply radical externalism quite independently of empirical evidence for cross-modal transference -which suggests that there is something badly wrong in the Wittgensteinian point. There are two rather important consequences of a consequent extension of the radical externalist thesis. (i) Any pair of recognitional concept tokens that are not a priori coinferable but that pick out the same property (regardless of whether

it is a shape property) are tokens of the same recognitional concept. This holds, mutatis mutandis, for perception or experience types. (ii) Given Campbell's observation that radical externalism implies the amodal status of shape-perception, we would then have to conclude from radical externalism about recognitional concepts in general that -contrary to the assumption of the previous pointthere cannot after all be pairs of recognitional concepts that pick out the same property and are not a priori co-inferable. Consider this example. You see animals of a certain kind up close and form a recognitional concept "that kind", which is grounded in a visual discriminative ability. You also see a kind of animal a t a distance and form a corresponding recognitional concept. Now as it happens those animal-kinds are one and the same. Still, you cannot tell whether this so; tokens of the one recognitional concept are not a priori co-inferable with tokens of the other recognitional concept. Yet the Wittgensteinian consideration would imply that these are the same concept because they pick out the same kind; and the thesis that radical externalism implies amodality implies that these concepts are hence co-inferable. So even if Campbell's principal thesis is meant to be restricted to shape concepts, his additional observations seem to justify generalizing the thesis to this sort of case. Suppose the thesis can be kept from generalizing, and applies only to shapes. Even so it seems to imply too much. Here are two cases. (i) Consider visual and tactual concepts that pick out a geometrically complex kind, e.g. things the shape, size and texture of siamese cats. Such concepts are, I take it, not guaranteed to be cognitively co-inferable. But Campbell's principles appear to imply that they must be so, that is, given enough a priori reflection. They apparently imply this quite independently of the empirical testing of crossmodal identifications of Siamese cats. (ii) Our visual facial recognitional abilities are stupendously refined. One doubts, however, that our tactual facial recognitional abilities are equally refined, still less cross-modally co-inferable with visual facial recognitions. What are we to say about these perceptions and recognitional abilities? That radical externalism does not hold here? If the principles behind Campbell's argument do imply radical externalism for all spatial concepts, they are then doubtful. If they do not, for some reason, extend to this case, then some perceptions and recognitional concepts are fine-grainedly individuated, and the Wittgensteinian argument should be abandoned. Now suppose we try to restrict things further: radical externalism is merely to fit the empirical cross-modal data; it is to hold only for

pairs of perception-tokens or recognitional concepts that have been established as cross-modally accessible. Campbell's thesis appears to be far bolder than this. But, still, consider the more cautious thesis. Even it does not appear well motivated. The sort of rational a priori co-inferability that cross-modal experiments establish apparently does not imply a radical externalist individuation for relevant shape perceptions and recognitional concepts. For nothing that we have seen makes it a t all implausible that, although certain of our visual and tactual shape perceptions and recognitional abilities are innately tied, this is just a bit of contingent wiring that we can conceive to be interrupted. Could there not be creatures the relation between whose visual and tactual capacities are as Berkeley imagined them to be? If that is a conceptual possibility, it apparently indicates that there is more conceptual structure here than Campbell allows, despite our de facto hard wiring. We have seen independent reasons to suppose that tactual and visual recognitional concepts of the same complex shapes might be conceptually independent. This strongly suggests that even in these favorable cross-modal cases, visual and tactual perceptions and recognitional concepts ought to be typed distinctly. There are epistemological views, to repeat an earlier point, on which inferences connecting innately tied concepts count -in a somewhat unexciting sense perhaps- as a priori rational.'

'An alternative to Campbell's picture is this. Perceptual shape concepts are conceptual amalgams. Conceptual amalgams function as single concepts, but have separable conceptual components. There are weak amalgams and strong amalgams. For strong amalgams, because of innate connections, separability is only separability in imagination. If some of our visual and tactual recognitions of shape are innately cross-modal, those modality-specific conceptual shape factors are components of strong amalgams. The conceptual components of weak amalgams --e.g. a name and a f a c e - can be pulled apart by ordinary contingencies. But conceptual amalgams are typically unified self-consciously, the strong ones having a sort of a priori unity, at least in a weak sense of 'a priori'.


Shape Properties and Percept ion Kirk Ludwig

1 Introduction We can perceive shapes visually and tactilely, and the information we gain about shapes through both sensory modalities is integrated smoothly into and functions in the same way in our behavior independently of whether we gain it by sight or touch. There seems to be no reason in principle we couldn't perceive shapes through other sensory modalities as well, although as a matter of fact we do not. While we can identify shapes through other sensory modalities -e.g., I may know by smell (the scent of mango) that the object causing my sensory experience is round- this is not perceiving an object as shaped, but rather inferring from the character of one's sensory experience and collateral information that an object of a certain shape caused it. That it is possible to perceive shape through other sensory modalities, however, is suggested by the case of bats and aquatic mammals like dolphins which navigate through their environment by a form of sonar. It is plausible that they have some form of auditory representation of space, and so of shape. These facts about shape perception raise important questions about the relation between those features of perceptual experience which are intrinsic to different sensory modalities and the nature of our per-

ceptual representation of shapes, and, more generally, of the space within which we perceive shaped objects to be located. John Campbell's paper, "Molyneux's Problem" (see above), raises a number of interesting and important questions about the nature of our perception of shape properties, particularly the cross-modal nature of shape perception, and ties them to more general questions about the nature both of perceptual content -whether it should be understood externalistically- and of shape properties --whether they should be understood categorically or as analyzable ultimately in terms of dispositions of objects. The question Campbell starts with is Q1. "whether there is a difference between the phenomenal characters of shape experience in sight and touch" (p. 301). This is one of the questions prompted by Molyneux's Question: Suppose a man born blind, and now adult, and taught by his touch to distinguish between a cube and a sphere of the same metal, and nighly of the same bigness, so as to tell, when he felt one and the other, which is the cube and which is the sphere. Suppose then the cube and the sphere placed on a table, and the blind man be made to see: Quaere: whether by his sight, before he touched them, could he distinguish and tell which was the globe and which the cube? (Locke 1975, 11, ix,8) If there is no phenomenal difference between perception of shape in sight and touch, then a man blind from birth whose sight is fully restored as an adult should, it seems, have no trouble visually distinguishing the globe from the cube.' Clearly, the nature of perceptual representation of shapes is important to answering Q1, and, in particular, whether perceptual representation of shapes is constructed in some way out of different phenomenal or subjective features in tactile and visual sensory experience. While Campbell does not explicitly provide an answer to Q1, I believe the answer he favors is that there is no difference between the phenomenal characters of shape experience in sight and touch. However, the main aim of Campbell's paper is not to provide and argue for an answer to Q1, but rather to connect its answer to the answers to a number of other questions about the nature of perceptual experience of shape properties. The first of these is, 'I raise some doubts in section 5 about t h e importance of there being no phenomenal difference between visual and tactile experience of shape for whether someone can know t h a t he sees t h e same shape t h a t he feels.

29. SHAPEPROPERTIES A N D PERCEPTION327

Q2. Ought we to be externalists2 about shape perception? Again, Campbell does not explicitly argue for externalism about perceptual representation of shape, but it seems clear that he is sympathetic to it. This question is in turn linked to the question, Q3. Are shape properties categorical properties or not?

We will explore more fully below what being categorical comes to, but in the first instance the contrast is with properties that are dispositional properties or conceptually analyzable in terms of dispositional properties. It is not clear that Campbell gives a definite answer to this question either, though it seems that he is in sympathy with the view that shape properties are categorical properties. The two issues raised by Q2 and Q3 are in turn linked to the answer to another question, namely, Q4. Is the cross-modal transfer of information about shape rational or not?

Cross-modal transfer of information about shape is the coordination of perception of shapes through different modalities in the production of behavior, in the sense that the information gained is integrated into and functions in the same way in the production of behavior independently of the sensory modality through which it is received. The cross-modal transfer of information is rational iff it is the result of the subject's access to the representation of shape in the experience itself. The alternative is that the cross-modal transfer of information is simply hardwired into us, a brute fact about our functional organization. On this view, there would be a physical explanation of it, and perhaps an evolutionary one, but no explanation which involved claims about the subject's recognizing that his perceptual experience was itself of an object of a certain shape. If we assume, as I think we should, that if something is in the content of one's perceptual experience, it is accessible to one in consciousness, then the 'Campbell distinguishes between radical and modest externalism about shape perception. It is not clear to me how to understand 'modest externalism', but since Campbell's concern is clearly with what he calls 'radical externalism', I will concentrate on that. Radical externalism appears to be the view that the representational elements of perception of shape are entirely determined by relational properties of the perceiver, and, in particular, relations to objects possessing the properties the individual represents. (What I mean by 'relational property' is the following: A property P is a relational property iffdf necessarily, for all x, x has P iff there is a y such that y # x or any part of x and y is a contingent existent.) Henceforth, by 'externalism', I will mean this view.

second view reduces to the view that one's perceptual experiences of shape do not contain any representations of shape. The beliefs that would arise from perceptual experience about shapes would not, then, have their contents derived from the contents of the perceptual experiences; rather, certain perceptual experiences that do not represent shapes would cause in us beliefs about shapes3 It is unclear to me, however, whether Campbell accepts the assumption on which this reduction depends. Again, while Campbell does not say outright that he thinks that the cross-modal transfer of information about shape is rational, he seems to be in sympathy with the view that it is. These questions give rise to a fourth question, which I believe is the main focus of Campbell's paper, namely, Q5. What is the relation between the answers to questions Q1-Q4? It is not as clear as it could be how Campbell wants to answer this question, but I think at least the following theses can be discerned. [TI] If shape properties are categorical, then perception of shapes is externalist . [T2] If perception of shape properties is externalist, then there is no difference between the phenomenal character of shape experience in sight and touch. [T3] If externalism about shape perception is correct, then crossmodal transfer of information is rational. The path to these conclusions is complicated, and reconstruction of arguments for them requires some detective work. I am by no means confident that I have identified the arguments or theses correctly. But these theses, if correct, are important results, particularly [T2], which would establish a surprising connection between the categoricity of shape properties and how perceptual representation of shape is determined. With respect to the primary questions, Q1-Q4, my own view is that there is a phenomenal difference between perception of shapes 3 ~ h i also s raises a question about whether what would occur in this case would count as perceiving shape, as opposed t o merely identifying it through sight or touch, as, in our example above, we might, through smell, identify a n object as round. If there is this connection between perception of a n object's having a property and t h e object's being represented in the experience as having it, then there is cross-modal transfer of information in perception only if it is rational. I will leave aside this worry for t h e rest of t h e discussion.

29. SHAPE PROPERTIES A N D PERCEPTION 329

in sight and touch,* that externalism about perception of shape is incorrect, that shape properties are categorical, and that the crossmodal transfer of information in normal adult human beings is rational. With respect to the relations among the answers to these questions, I think that they are by and large independent of one another, and, in particular, that none of [TI]-[T3] can be supported by showing that the antecedent entails the consequent. I will not argue in this paper that there is a phenomenal difference between perception of shape in sight and touch, or that externalism about shape perception is false. I will argue that shape properties are categorical, and that the cross-modal transfer of information is rational. In part, my aim here is to show that the questions are most straightforwardly answered independently of the issues Campbell links them to. Finally, I will argue that Campbell has given us no reason to think that there is an entailment between the antecedents of [TI]-[T3] and their consequents. The plan of the paper is as follows. In section 2, I consider question Q3, whether shape properties are categorical. In section 3, I take up question Q4, whether cross-modal transfer of information is rational. In section 4, I take up the relation between the categoricity of shape properties and internalism and externalism [TI]. That there is an important connection must be a major thesis of Campbell's paper, for roughly half of the paper is taken up with a discussion of the nature of shape properties. But I can find no interesting connection between the question whether shape properties are categorical or not and the question whether perceptual content of shape properties is externally determined. In section 5, I take up, briefly, [T2] and [T3], and conclude in section 6. Each of sections 2-5 can be read independently of the others. Thus, e.g., a reader primarily interested in the discussion of Campbell's arguments for [T2] could skip straight to section 4. 4 ~ e (Martin e 1992) for a n interesting recent discussion of what this difference might come to. Martin argues t h a t it is not, as O'Shaughnessy (1989) has urged, t h a t there is no sense field in tactile experience of t h e sort there is in visual experience, but rather t h a t visual experience is "experience of objects external t o one as arranged in physical space," while tactile experience is "experience of objects as they come into contact with one's body"; he continues, one is aware of one's body and its limits and so aware of objects coming into contact with one's body as they discernibly affect those limits. Normal visual experience is essentially experience of objects as they fall within t h e visual field; tactual experience is essentially experience of objects as they press from t h e outside onto the limits of a felt sensory field. (p. 210)

2 Are Shape Properties Categorical or Not? Categorical properties are contrasted with powers, that is, with dispositional properties, or properties which are definable by appeal to only logic and powers. The mark of a power, according to Campbell, is that it can be defined by the following formula, [PI X has P iff if circumstances C obtain, then X responds in way R (where the specification of way R does not make appeal to the property P being defined). In addition to powers, Campbell says there are conditional powers, which can be defined as follows,

[CP] X has P iff if X has properties Q1,. . . ,Q,, then in circumstances C, X responds in way R. Campbell goes on to entertain the suggestion that shape properties might be, not powers, or conditional powers, but clusters of conditional powers, i.e., that shapes can be defined by the following form, where the [CPil stand for conditional powers: [CL] X has P iff X has CP1,. . . ,CP,. Campbell identifies this as Shoemaker's proposal (1984 pp. 206233) that all properties of spatio-temporal objects are clusters of powers as applied to shape properties in particular. This is not, I think, exactly right. The most important divergence between the way Campbell lays out the view he wishes to examine and Shoemaker's more general thesis is that Shoemaker's thesis is about what we might call the 'metaphysical' nature of properties, and is not about how our concepts of properties are analyzed.5 Since Campbell characterizes the view he is concerned with as being about how to analyze the concept of a shape property, it is not Shoemaker's view as applied to shape properties. The divergence is important, because the proper method for deciding issues about conceptual and metaphysical necessity are (thought to be) different. It is not, however, crucial to Campbell's line of argument that the view he considers be 5I a m skeptical t h a t there are any so-called metaphysical necessities which are not reducible t o conceptual necessities. T h e examples which motivate the distinction, such a s t h e putatively a posteriori discovery of t h e necessary t r u t h t h a t water consists of aggregates of H20 (and its isotopes) in t h e liquid state, are better understood as examples of t h e discovery of what concept 'water' expresses rather t h a n of a new kind of necessity not grounded in our concepts. But I cannot pursue this issue further here.

29. SHAPE PROPERTIES AND PERCEPTION 331

Shoemaker's view. In what follows, the view I intend to be evaluating is the one presented in Campbell paper, whether or not it is Shoemaker's view.6 I take this view to hold that two properties are identical iff the concepts which pick them out are identical, and that the proper method for investigating the nature of properties is by analyzing the concepts which pick them out. Before considering more closely the proposal that shape properties can be analyzed in the form [CL], it is worth asking whether conditional powers really are distinct from simple powers. It appears that they are not, for any conditional power defined as in [CP] can be recharacterized in as in [PI. To see this, suppose we have a conditional power defined as above. Now let circumstances C* = C and X's having properties Q1,. . . , Q,. Then we have [RE] X has P iff if circumstances C* obtain, X reacts in way R. Thus, we could simply take the suggestion that shapes are clusters of conditional powers as the suggestion that they are clusters of powers, and simplify the discussion. To determine whether shape properties are categorical properties or not, we need to know more about what a categorical property is. Categorical properties are to be understood by what they are contrasted with. Categorical properties are properties that cannot be defined as in [PI, [CP], or [CL]. Thus, we do not have to worry about their roles in explanations to explain what they are (see section 3 of Campbell's paper). They will, however, have a special role to play in explanations, in virtue of categorical properties being the ground of dispositional ones (though there's no reason to suppose all 'shoemaker also makes some terminological distinctions that Campbell does not observe, and which I will not observe either. For example, Shoemaker draws a distinction between 'is dispositional' and 'is a power'. The former he treats as a predicate of predicates; the latter of properties. It should be noted also that Shoemaker is concerned with what he calls genuine properties, as opposed to (mere) Cambridge properties; according to Shoemaker, not every predicate expresses a property. Part of the motivation of his account is to provide a way of distinguishing between genuine properties and mere Cambridge properties. Genuine properties, however, as Shoemaker conceives them, are only properties which are involved in (genuine) changes; a third class of properties, neither genuine nor mere Cambridge properties, are those of abstract objects such as numbers, e.g., being prime or being divisible by two. This last point in particular suggests, given the epistemological nature of Shoemaker's argument (see below), that his conclusion should really be recast as the claim that the only properties we could know objects to have are one's characterizable essentially in terms of powers, not that properties of spatio-temporal objects must be individuated by their causal powers.

categorical properties ground dispositional powers). What is it for a categorical property to ground a dispositional one? [GI P is the categorical ground of a disposition to R when C iff P is a categorical property and it is a causal law that if X has P and circumstances C obtain, then X responds in way R. To say it is a causal law that if X is P and circumstances C obtain, then X responds in way R, is not to give a definition or conceptual analysis of the predicate 'has P', but to make a remark about the nomic role the property picked out has in physically possible worlds. This is a contingent fact, and things could have been otherwise, and it must be established by a posteriori investigation. It is evident that there must be categorical properties, if there are dispositional ones. For, first, there must be properties which figure in the fundamental laws which explain why objects have the dispositions that they do. And, second, if every property were spelled out dispositionally, so that our understanding of the concept which picked out the property were expressed in a counterfactual conditional, then there would be no properties for which we had concepts. Powers are defined in terms of conditionals. The antecedents and consequents must employ predicates that express properties. The conditional has no content until we specify what those properties are. If for each of those predicates we had to substitute a conditional, and then for each predicate in the substituted conditional another conditional, ad infiniturn, then clearly no properties would be fixed at all. Now we are in a position to answer the question whether shape properties are categorical properties. It is sufficient (though perhaps not necessary) for a property not to be a categorical property that no biconditional of the form [PI, [CP], or [CL] be necessarily true with a predicate expressing the property in the place of 'has P'. Representing the universally quantified biconditionals of these forms as predicates of properties as follows '[PI (P)', '[CP](p)', and '[CL](p)', where p is a variable for the property being characterized, we can put the condition as follows: [C] (p)(p is a categorical property provided that it is not necessary that [Pl(p) or [CP](p)or [CLl(p)). Are shape properties then categorical? It will help to ask a bit more particularly about what we have in mind in talking about shape properties. When we talk about an object's having a shape, e.g., of an object's being spherical, we ascribe a property to an object located


in space. Now consider the space which that object occupies. Can we speak of the shape of that region of space? Yes. For remember that we also represent the regions of space between and around the objects which we represent in it, and that this is essential to our representation of the shapes of those objects. But now in talking of the shapes of regions of space, we are not talking about an object in the usual sense, something which occupies space, and can move about within it, though we are clearly employing the same concepts as those we apply to objects. The two uses are of course connected: an object has a certain shape just in case the region of space it occupies (at a time) has the same shape. Let's take 'shape' as applied to regions of space first. Since space is thought of as causally neutral, the stage where the play takes place, rather than a participant in the action, it is not clear that there are any biconditionals of the form (3) true as matter of fact of shape properties of regions of space, let alone necessarily true. Our concept of shape properties then cannot be captured by our concept of any cluster of powers. This is the central difficulty in trying to analyze shape properties in terms of powers. Powers are properties of objects. If they are objects which have shapes, they have spatial locations. Their shapes, however, are understood in terms of the geometry of the regions of the space in which they are located. Our conception of that space and its geometry is independent of that of the objects located within it.7 Thus, our understanding of the shape properties of objects could not be reduced to powers which they possess, since this leaves no room for an independent conception of the geometry of the space in which they are located and in terms of which their shapes are understood. Now consider 'shape' as applied to objects. We understand what it is for an object to have a shape in terms of the shape of the region of space it occupies. Does the shape of an object necessarily have any consequences for what it will do in various circumstances? This depends, I think, more on our concept of an object than on our concept of shape. To see this, consider the contrast between rigid and non-rigid objects. A rigid object is one that retains its shape (and size) under all conditions (that is, the distance between any two points in it remains constant through all forces exerted upon it). A non-rigid object is one that does not. This is not a contingent fact about rigid objects; nothing counts as a rigid object of which 7This assumes, as I think correct, that spatial relations cannot be reduced to relations among objects; arguments to the contrary from a principle of sufficient reason or verificationism rest on doubtful premises.

this is not true. (In point of fact, there are no rigid objects; rigid objects are an idealization in physics, like a frictionless plane.) A rigid and non-rigid object could have the same shape. But they will do different things in different circumstances. For example, the rigid square peg will not go through a hole in a rigid board whose diameter is the same as the length of the peg's sides; a non-rigid peg may well do so e . g . , a square peg made of clay. It is not the shape and size of the peg relative to that of the hole in the board that determines whether or not it will pass through it, but rather the dispositions of the peg and board to retain their shapes in various circumstances. Since we understand the contrast between rigid and non-rigid objects in terms of their dispositions to retain or not retain their shapes and sizes in various circumstances, and this distinction is crucial to saying what an object of a certain shape will do in various circumstances, it is clear that shape properties cannot be defined in terms of clusters of powers of objects. For t o try to understand it that way, we would have to appeal to the concept of rigidity, which presupposes a prior grasp of the concept of shape. The problem becomes painfully obvious when we consider inserting a characterization of rigidity into a characterization of shape in terms of rigidity. [RI] X is rigid iff for any shape S, and Size 2, if X has S and 2, then no matter what the circumstances C, X has S and 2. Consider now how one might try to turn this around into a partial definition of shape in terms of what an object will do if it is rigid: [RV] X has S only if, if stances, X has S.

X is rigid, then no matter what the circum-

There are really two problems here. First, of course, we have seen that 'rigid7 is defined in terms of shape. Second, even apart from that, we see that the particular shape we a.re attempting to define here has to be mentioned in the consequent of the conditional. Clearly, no progress has been made. Before leaving this section, we should take a brief look at two further matters. The first is the functional definition of shape properties which might be proposed to avoid the circularity problem (Campbell, p. 311). The second is Shoemaker's arguments, as represented by Campbell, that properties have to be understood as dispositional all the way down. To get around the charge of circularity, Campbell suggests we can introduce the shape property by way of a description of it as the

29. SIIAPE PROPERTIES AND PERCEPTION 335

property which has a certain role in a theory of shape properties obtained in the following way. Consider a particular shape property S. Suppose it is definable as a cluster of conditional powers (or simply powers),

[To] ( x)(x has So iff if x has Q, then if C , x will R

and if x has Q1, then if C1, x will R1

and if x has Q2, then if C2, x will R2

...

and if x has Q,, then if C,, x will R,). Note that in the [Cils we will have to mention shape properties. Now, summarize the open sentence on the right hand side of the bi. . ,Sk)', where the [Sil represent the terms conditional a s 'TO(x,S O,. referring to shape properties in the specification of the circumstances. Now we represent the definition of So as follows. [Def,-So] (x)(x has SOiff To(x,S O , . . , Sk)). This is a circular definition (since we will need to say how objects of the same shape interact in characterizing the dispositions of an object of any shape). Now, take the conjunction of all the true biconditionals of the form of [Def,-So], and represent this as a theory: Now replace each of the terms referring to shape properties with distinct variables and existentially quantify over each of the variables to get: This is the Ramsey sentence for [TI]. Now, call the predicate obtained from [T2]by removing the existential quantifier over the i-th variable CT;(S~)]. Now, we can define each property Si as follows: [Def-Si] (x)(x has Si iff x has (isi) ~ ; ( s i ) ) . ~

I want to note two points here. First, it is obvious that this is not how we understand shape properties, since there are an infinite number of shape properties, so that above the values of 'k' and 'r' will be infinite, and we clearly could not grasp the resulting proposition. Second, and quite apart from this, [Def-Si] does not define CSil in terms of powers, for it is not strictly a definition. For all that [DefSi] says CSi] could denote a categorical property. This is because 'Where as usual '(is;)' is interpreted as 'the si such that'.

[Def-Si] fixes the property picked out by [Si] by description. It is whatever unique property plays the appropriate role in the theory. That property, for all that has been said, may well be a categorical property. [Def-Si] fixes the meaning of [has Sil in the way one fixes the referent of a name by a description such as 'the first child born in the 21st century'. The description does not give the meaning. Thus, there is no reason to think that in fixing what property ISi] denotes in this way we have determined it to denote a property definable in terms of clusters of powers. Thus, this is not a response to the objection that definitions of shape properties in the form [CL] are going to be circular, and so fail as definition^.^ In conversation, Shoemaker has told me that this is not part of his aim, and that the content of his thesis is exhausted by the claim that the above biconditional is necessarily true. Now let's turn to the arguments adapted by Campbell from Shoemaker. As I noted above, I think there are some important differences between the position that Campbell characterizes and Shoemaker's. I will be considering the arguments advanced below as they bear on the position that shape properties and properties in general are understandable in terms of powers as it is characterized by Campbell. It is clear that, in at least some cases, the arguments are not applicable to Shoemaker's own position. The overall form of the argument is the following: 'I would like t o insert a brief query about Shoemaker's own official proposal, which is represented as a criterion for individuating properties. Modified t o meet a n objection of Richard Boyd's (Shoemaker 1984, p. 233), it goes as follows, PI = P2 iff necessarily, for all causal potentialities CP, PI has C P iff Pz has C P a n d for all circumstances C , C is sufficient t o bring about a thing's having PI iff C is sufficient t o bring about a thing's having Pz. P u t this way, one can see t h a t there is a striking parallel between Shoemaker's proposal for individuating properties and Davidson's for individuating events (Davidson 1980). Thus, it looks as if it suffers from t h e same defect Quine (1985 p. 166) pointed out in Davidson's a t t e m p t t o individuate events in terms of their causes and effects. Since causal potentialities and circumstances quantified over on t h e right hand side of t h e biconditional themselves involve properties, t h a t is, they are partly individuated in terms of what properties they involve, t h e right hand side of t h e biconditional above presupposes t h e individuation of properties, and so cannot explain in what their individuation consists. MThether this criticism applies t o Shoemaker's account depends in part on what he means when he says t h a t he is giving individuation conditions for properties. Quine's criticism presupposes t h a t in giving individuation conditions for a kind of entity, t h e conditions one gives d o not presuppose one already understands how t o individuate t h e kind of entity in question.


[A] 1. If properties are categorical, then we can never know that any object has them. 2. Whatever properties there are, we can know that some object has them. 3. Therefore, there are no categorical properties. A first remark to make is that since there have to be categorical properties if there are powers, a further conclusion can be drawn from 3, namely, 4. There are no properties, i.e., no object has any property.

Since there are objects only if they have properties, it follows, 5. There isn't anything1' This is a conclusion sufficiently absurd to lead us to suppose that something has gone wrong with the argument for it. First, there is no reason to think 2 is true. There may be many properties which objects have which we don't and can't know that they have because of the nature of the properties, or because of the nature of ourselves. Second, none of the reasons offered for 1 are any good. As represented by Campbell, these are: [Rl] If there are categorical properties, then it 'ought to be possible for there to be properties which make no difference whatever to the behaviour of the things which possess them.' (p. 311) [R2] If there are categorical properties, then there 'could be different properties that make, under all possible circumstances, exactly the same contribution to the causal powers of the things that have them.' (p. 311) [R3] If there are categorical properties, then the 'potential of a particular property for contributing to the production of causal powers might change over time.' (p. 311) Consequently, [C] '[Wle could have no knowledge of the properties of a thing since all we can know is the behaviour of the thing- and we would have no way of singling out a property in order to name it, and even if we did somehow manage to christen a property, there would be no way.. . in which we could know that we had encountered the same property again.' (p. 311) ''Or, rather, anything except abstract objects, in line with the qualification in footnote 6.

There is nothing in the concept of a categorical property which licenses any of [Rll-[R3]. What is required for properties to be categorical is that there be no biconditionals of the form [CL] (or [PI or [CP]) which define them. With respect to [Rl], it is compatible with this being so for shape properties that it is necessary that shape properties make a difference to the causal powers of the objects that have them, for that does not require that they make the same difference in every possible world. With respect to [R2], it is compatible with this being so that different shape properties necessarily make a different difference to an object's causal powers. With respect to [R3], it is compatible with this that shape properties necessarily make the same difference to an object's causal powers at every time. Furthermore, even if [Rll-[R3]were true, it is not clear that it would follow that we could have no knowledge of the properties of a thing. It is clear that the proponent of this argument supposes that we can have knowledge of dispositional properties, and since not all properties of objects can be inferred from antecedent knowledge of the objects' having other properties, on pain of an infinite regress, we must be capable of knowing for some properties and some objects that the objects have the properties without inferring that they do. The epistemic problem that the objector to categorical properties envisages rests on the assumption that we must infer that objects have categorical properties from the effects of their having those properties. Their effects will be manifested as properties of some objects that we can know the objects have non-inferentially. Clearly, to be consistent, the proponent of the argument must suppose those properties are powers or definable in terms of powers. Thus, the proponent of the current argument must hold that we can know directly that objects have dispositional properties, but that there is some special obstacle that prevents us from knowing directly that objects have categorical properties. But what is that obstacle? If we could know directly that objects have dispositoinal properties, why could we not know directly that objects have categorical properties? Indeed, the objector's assumption that we would have to infer what properties objects have from direct observation of (some at least of) their dispositional properties seems to get things upside down. If there is a priority of knowledge of the one sort of property over knowledge of the other sort, knowledge of the dispositional properties of objects should rest on knowledge of categorical properties of those objects. For we only know what powers a thing has by inductive reasoning. To discover a thing's powers, we must see what it does in various circumstances, i.e., we must have access to the manifestations of its powers. But these manifestations can't (ultimately) be themselves


powers if our inductive reasoning is to have materials to work on to begin with. Thus, it seems that even if we grant [Rll-[R3], we have no reason to suppose we could not know that objects had categorical properties, and we have reason to think that unless we could know that objects had categorical properties, we could not know that they had dispositional properties or properties definable in terms of dispositional properties. To summarize, we have seen that shape properties are categorical properties, that there is no response to the circularity objection in appeal to definition by way of Ramsey sentences, and that none of the arguments against shape properties being categorical properties we have examined are any good. Furthermore, our treatment of these issues has not required us to make any commitments about how the contents of our perceptual representations of shape are determined.

3 Is Cross-Modal Transfer of Information

Rational?

We described the phenomenon of cross-modal transfer of information in a way that was neutral with respect to the two interpretations of it. Both of the interpretations are prima facie conceptually possible. In particular, it seems possible for the brute functional organization thesis to be true of some creatures who have concepts of shape.'' It also seems conceptually possible for perceptual experiences to contain in their contents representations of space and of shaped objects within that space (to which the perceiver has access). If both options are conceptually possible, then if we are to answer the question whether cross-modal transfer of information is rational, we cannot do so for all possible perceivers. We must relativize the question to particular groups of perceivers. Clearly, the thesis that cross-modal transfer of information is rational, since it involves a claim about the content of perceptual experience and access to that content by the perceiver, can't be answered on the basis of behavioral evidence. The behavioral evidence will be compatible with either alternative. Therefore, where we can have no insight into the character of the experience from the point of view of its subject, we must be skeptics about the answer to the question. lllf there is some difficulty about this, it would have to be because of some important connection between the ability t o possess the concept of space and the ability to experientially represent space. It is tempting to suppose that there is such a connection, but I cannot see how to make good on the claim.

FIGURE 1. Necker cube.

FIGURE 2. Penrose triangle.

This is plausibly the case with animals, and with infants and young (enough) children. I propose then to reformulate the question so that it is specifically about the phenomenon as it appears in normal adult human beings. Henceforth I will understand the question in this way. Given that we have access to the contents of our conscious percep tual experiences, the question whether cross-modal transfer of information is rational comes down to the question whether we are aware of representations of shaped objects as part of our visual and tactile experiences of our environment. What is the answer to this question? I think we obviously do have access to such representational information in the content of our perceptual experiences. When I look out my window, I have a visual experience of, as I would put it, a number of heads, on a number of shoulders; that is a remark about the content of the experience, and in particular conveys some (not very precise) information about the shapes of objects which my visual experience represents. Likewise, as I hold this pen in my hand, I have a tactile experience of (and as of) a cylinder, as I would put it, which conveys information about the representational content of the tactile experience, and, in particular, about shapes of objects which it represents. As I look a t the pen, my visual experience also represents it as cylindrical. These representational features of my experience are transparent to me, for my experiences phenomenally seem to put me in direct contact with the world that they represent, and I am representing a three dimensional spatial world containing objects of various shapes a t various distances from me. It does no good to protest that this representational content is not part of the content of my perceptual experiences, but rather just of my beliefs, for even if I suspend belief about my environment, my perceptual experiences still represent it as being the same way. This is the possibility upon which hallucinations and illusions which persist through


discovery depends, such as a 3-dimensional Necker cube (Figure I ) , or Ames's distorted room (see Gregory, pp. 85-7), and the perceptual representation of objects recognized to be impossible (such as the Penrose triangle (Figure 2)). The question whether cross-modal transfer of information is rational in our case, then, seems to admit of a straightforward and incontrovertible answer: it is. Before moving on, let us ask whether we find any reason to be skeptical of this result in the case of the man who is the subject of Molyneux's question, a man born blind, who can recognize by touch a sphere and a cube, whose sight is restored as an adult. Molyneux's question is whether he would be able to recognize by sight the sphere and the cube, that is, to know that the object which he sees is of the same shape as the object which he feels, when he sees the sphere, or cube, and when he touches it. How might the results of this experiment be used against the above claim? Suppose that the man could not at first recognize the object identified as the sphere by touch as that object by sight, but later comes to be able to do so. Then it might be urged that the connection is purely a functional one, since the content of his visual experience has not changed, and he could not originally make the identification, and surely at the end of the process his experience will be the same as ours. There have been actual cases like the one imagined by Molyneux, in which an individual blind from birth because of opacity of the cornea, or because of cataracts, has had his sight restored by an operation. The empirical facts are not easy to interpret, however. In some of these cases, individuals required a long period of training to recognize and name even simple objects or shapes by sight. In others, they were able to see well almost immediately. The cases are difficult to interpret, because it is difficult to sort out by behavior cases in which physiological function has been restored completely from those in which it has been restored partially at first, and fully only slowly if at all. There does seem to be some suggestion that where the operation, as in cornea transplants, results immediately in a clear image on the retina, the individual is able almost immediately to recognize by sight objects which he is familiar with through touch. (See Gregory 1990, pp. 201ff.) However, whatever the empirical facts, the answer to Molyneux's question does not have the power to upset the conclusions above. The actual empirical facts are limited to whether or not the subject is able immediately to recognize shapes by sight or not, and then becomes able to do so later on. The objection considered above depends on making two further assumptions, which amount to an explanation of the empirical facts. The two assumptions, as we saw, are that the perceiver's experience

does not change its character, and that his experience when he can identify shapes by sight is the same as ours. The empirical evidence is compatible with denying either of these. Thus, e.g., in coming to identify shapes as we do, our imagined subject may well have subjective experience of the sort we do, but as a result of the character of his perceptual experiences changing from the time when his sight was initially restored. This would be much the most likely explanation, in light of the fact that the representational content of visual experiences is a feature of its phenomenal content. Contrast auditory perception of an uttered sentence when one understands it and when one does not, or perception of script which one understands with perception of script which one does not. Or, again, consider the difference between different ways of seeing an ambiguous figure, such as the duck-rabbit, or a Necker cube. In all these cases, changes or differences in the representational content of the perceptual experience clearly affect what it is like to have them. Even if we did have good reason to think, however, that the subject's subjective experiences had not changed, there would still be explanations of the data compatible with the result reached above. Thus, e.g., we might think his experience was not in fact similar to ours, that in his case there was only a functional connection. It is hard to see, in fact, how the (third person) empirical evidence could overturn our first person authority over what the contents of our perceptual experiences are like. Therefore, even if a man blind from birth whose sight is restored would not be able identify the object he sees as the sphere he touches as opposed to the square that he touches, it would not show that cross-modal transfer of information is not rational in our case. Thus, the answer to the question whether cross-modal transfer of information is rational is 'yes', and the answer can be given independently of any considerations that have to do with externalism or the nature of shape perception.

4 How Are the Issues of the Categoricity of Shape Properties and Externalism about Perception of Shape Properties R,elated? It is not easy to see why Campbell thinks that the question whether shape properties are categorical is relevant to the question whether perceptual representation of shape properties is externalist. The question of the nature of shape properties enters into the discussion in the following passage:

29. SHAPEPROPERTIES A N D PERCEPTION 343

This gives one way of setting up the question of the sameness or difference of shape perception in sight and touch. But you might object that this is quite the wrong way in which to set up the problem. You might object that we have to think of shape properties in terms of their causal significance, and that shape perception is perception of this causal significance. In this section I will develop this causal approach to shape perception. Then I will argue that the approach is incomplete, and that there is still a crucial issue that is addressed by externalism about phenomenal experience of shape. (p. 304) T h e 'sameness or difference of shape perception in sight and touch' refers here to whether there are phenomenal differences between perception of shape in sight and touch. I t is not clear exactly what Campbell has in mind by 'this way of setting up the problem': the problem does not require setting up, it is expressed in a simple question, Q1. So it is mysterious what objection we are to imagine is being raised by the thought that shape properties are to be understood in terms of their causal significance (as clusters of powers (or conditional powers)). I t is evident, though, from the last line of this passage, that Campbell thinks that this approach to understanding shape properties fails -'is incomplete'- and that externalism is required in some way t o address some 'crucial issue'. What is that crucial issue? How do we work our way around from a discussion of the nature of shape properties to whether the representation of those properties in perception is externally determined or not? T h e transition takes place in two steps. First, we are invited to reflect on what perceiving shapes would come to on the view of shape properties as definable as clusters of conditional powers. Campbell represents this as a pressing need for the proponent of the cluster of powers view of shape properties: On the face of it, we do not perceive the shape of a thing as a collection of unsubstantiated threats and promises as to which powers it will take on in various hypothetical circumstances. We perceive the substance behind the threats and promises. So we need some preliminary explanation of what it could mean to perceive an object as having a cluster of conditional powers, before we can understand the argument that that is all we perceive the object to have.'' (p. 306) 1 2 ~ a m p b e ldoes l not explain how he is using the expression 'perceiving x as

. . . ', but I will take this t o mean that the representational content of the percep-

tual state involved represents x as falling under the concept expressed by what goes in for '. . . '.

Campbell spends some time saying what it could mean, but the question and time spent are not well-motivated. What it means is as clear as the thesis that shape properties are definable as clusters of conditional powers. It is not a motivation for asking after the meaning of some claim that it is in conflict with common sense. At most that is a reason for inquiring after its truth. What it means is just that in perceiving something as having a certain shape, we perceive it as being such that in such and such conditions, it behaves in such and such ways, which conditions and ways depending on the shape in question. Of course, this has implications for our behavior, given that we are rational agents, as Campbell points out a t length, but adding what those implications are, given our desires, does not shed any further light on what the claim is. In any case, this exercise sets up the next transition, which is the observation that if we want to deny that shape properties are wholly analyzable as clusters of conditional powers, then we must suppose that perception of shape properties involves "more" than perception of clusters of conditional powers. On the alternative view, these connections [between having a shape and reacting in such and such ways in such and such conditions] do not exhaust the content of shape perception. In addition to these connections, shape perception provides knowledge of the shape property as the ground of all the functional patterns that are specified. The problem then is to explain just what this further knowledge provided by shape perception is, and to explain its relation to these further connections and patterns of functional role. (p. 312) It should be noted that if shape properties are categorical properties, then it is understating the case to say that 'these connections do not exhaust the content of shape perception'. They are not part of the content of shape perception as such a t all. But we may grant the point that if shape properties are categorical, one must accept that perception of shape properties is perception of categorical properties. It is this point that apparently brings us back to the question of whether externalism about shape perception is correct. Campbell calls perception of shape as categorical 'primitive consciousness of shape'. (I cannot see that 'primitive consciousness of shape' comes to more than this.) Now, how do externalism and internalism enter into the picture? The transition takes place in the following passage: The kind of primitive consciousness of shape I am describing is.. . extremely primitive. There would be none of the complex of functional connections to action and imagistic reasoning which I described earlier.


That means that we have to consider the question what makes it so that this consciousness is consciousness of shapes. (p. 314) It is to respond to the question posed in the last sentence here that Campbell turns to discuss externalism and internalism. The question is to be motivated by the first 2 sentences, tying it to the discussion of the nature of shape properties. It is difficult, however, to see the bearing of the first two sentences on the last. Primitive consciousness of shape just is perception of shapes as categorical, i.e., perception of shapes, if shape properties are categorical properties. Then of course it follows that such perception of shapes is not perception as such of clusters of conditional powers, and as such does not enter into deliberations about what to do in the way that perception of clusters of conditional powers would. Why does that mean that we have to consider the question "what makes it so that this consciousness is consciousness of shapes?" In any case, internalism and externalism enter at this point. Externalism is supposed to provide the following answer to the question: since the "geometrical aspects of your experience of objects are constituted by the geometry of the objects in your surroundings," ". . . the character of experience as perception of shape can be secured without having to consider the relations of the perception to imagery and action" (p. 314). Internalism, too, is supposed to be prima facie in a position to answer the question: "an internalist account of shape experience, on which sensations are configured in a sensational space.. . would also suggest the possibility of a kind of primitive awareness of shape, on which it could be prior to the complex of functional connections to action and imagistic reasoning described above" (p. 314). However, it turns out that internalism faces a problem that externalism does not, and it seems that this is what provides finally a motivation for preferring externalism to internalism. Before taking that up, however, we should consider whether up to this point we have been provided with a reason to think that the issue of whether shape properties are categorical is specially connected with the debate between internalism and externalism. The only reason provided appears to be the 'need' to raise the question what makes perception of shape as categorical perception of shape. Given the kind of answer offered to this question, it is clear that 'what makes it so' is being interpreted as 'what facts other than the fact that the perception is perception of shape are sufficient for it to be so'. I cannot see any reason why this question needs to be raised at all, or any reason to think that if it must be raised, it

must be raised specifically in the case that shape properties are categorical. If it must be raised at all, why should it not need to be raised equally if shape properties are clusters of conditional powers? There is a hint of a reason in Campbell's paper in his discussion of what it would mean for perception of shapes to be perceptions of shapes as clusters of conditional powers. For it may be that Campbell thinks of that question as the same question as what makes it the case that perception of shape is perception of shape if shape properties are clusters of conditional powers. Then he may suppose he has given an answer independently of the internalism/externalism debate, so that that issue comes up only when we turn to the hypothesis that shape properties are categorical. But (i) the questions are not the same, and (ii) the answer Campbell earlier gives does not shed light on either of the questions, since the observation that we behave in ways appropriate to how we represent the world is not an answer to the question what it means to perceive shapes as clusters of conditional powers or to the question what makes representations of shapes representations of shapes. The question of the nature of shape properties, then, has nothing special to do with the question whether shape perception is externalist or not. The discussion of it is a red herring. As far as the debate between internalism and externalism goes, then, we could as well have started with the problem, alluded to above, that internalism is supposed to face that externalism does not. The problem is supposed to be that these sensational characteristics [which form a phenomenal space for representing geometry] are supposed to be known from one's own case, it is your own experience of shape that gives you knowledge of what these sensational characteristics are. So this view has no way of explaining what it is for sensational experiences of shape to be the same or different in different observers, or for a single observer over a period of time. ( P 314)

In contrast, The obvious way of providing an account is to move in the direction of the externalist, drawing in environmental considerations, so that sameness or difference of experience depends on the sameness or difference of the geometrical properties of the objects perceived. (p. 315)

What is the problem here? Campbell does not explain. It is unclear whether Campbell is worried about how the internalist explains what it is for two individuals both to be having a perception of, say, a


round object, or whether his concern is with how the internalist can explain what it is for two individuals both to be having, say, a visual perception of a round object. But in neither case is it obvious what the problem is supposed to be. What is it for A and B both to have a perceptual state which represents there being a round object in front of them? It is for both A and B to have a perceptual state which represents a round object in front of them. What is it for A and B both to have a visual perceptual state, as opposed to tactile, which represents there being a round object in front of them? It is for it for A and B both to have a visual perceptual state which represents there being a round object in front of them. Why is this answer not available to the internalist? Why should the fact that the internalist claims (if he does) to know what experience of shape is like from his own case raise any difficulty?13 Until more argument is forthcoming, we should conclude that there isn't even an apparent problem for the internalist in saying what it is for two individuals or an individual a t two times to be having perceptual states which represents shape with the same content. Apart from this, we should note that Campbell is assuming that any internalist about shape perception would have to appeal 'to a phenomenal space account of shape perception, and that any externalist account would eschew such an account. However, an internalist may very well reject a phenomenal space account of shape perception. To be an internalist, all that is needed is that one deny that the property of having an experience representing a shaped object is a relational property. Being an internalist does not carry any commitment about the relation between representations of shape and phenomenal content. Similarly, being an externalist does not require any commitment to rejecting the phenomenal space account. Thus, even if the phenomenal space account were shown to be untenable, this would not show that externalism was to be favored over internalism. I suspect that Campbell has assumed that internalism carried a commitment to a phenomenal space account of perception of shape I3It may be t h a t Campbell has in mind passages (esp. sections 258ff) in Wittgenstein's Philosophical Investigations t h a t are sometimes called t h e private language argument, which is briefly mentioned early in the paper (p. 303). These arguments, if they are arguments, amount t o little more t h a n t h e assumption t h a t we cannot recognize subjective features of our experience and identify experiences a t different times as the same on t h e basis of memory. T h e thought t h a t we can recognize something about ourselves in a way no one else can or with a n authority no one else's can equal is neither a threat t o t h e normativity of judgments about experience, nor a guarantee of infallibility.

because he assumed that any internalist would want a reductive account of shape perception, and assumed that the only possible account would be a phenomenal space account. It is by no means clear that this would be the only reductionist option available to an internalist, so it is unclear even that an internalist with the commitment to give a reductive account of perceptual representation of shape would face difficulties if it were shown that the phenomenal space account could not be maintained. Finally, in connection with these last two points, it has occurred to me that Campbell may be supposing, though he nowhere says this in the paper, that there is a special connection between the view that shape properties are categorical and the debate between internalism and externalism because he supposes that it is only if shape properties are categorical that the internalist must appeal to the phenomenal space account of shape perception, with its supposed attendant difficulties. We have seen that the assumption on which this connection is based is false because one can be an internalist, and perhaps even a reductive internalist, without adopting the phenomenal space account of shape perception. Apart from this, however, it is unclear why one would suppose that similar issues did not arise for perception of clusters of causal powers through more than one sensory modality. To sum up, we have found no connection between questions about the nature of shape properties and the determination of the contents of representations of shaped objects. It appears that the main line of Campbell's argument is that only externalism could explain how perception of shape properties is possible if shape properties are categorical. However, on examination, the categoricity of shape properties appears to play no role in the argument. The rejection of internalism, given the categoricity of shape properties, is instead based on an objection to a phenomenal space account of shape perception. That kind of account, however, is not specially tied to internalism or externalism, nor to the categoricity or non-categoricity of the properties we perceive objects as having. Thus, the disconnection between the issues is complete.

29. SHAPE PROPERTIES AND PERCEPTION349

5 What is the relation between externalism and the rationality of the cross-modal transfer of information, and between externalism and the question whet her there are phenomenal differences between perception of shape in sight and touch? The argument for a connection between radical externalism and the rationality of cross-modal transfer of information is that (i) if externalism about shape perception is correct, then there is no difference in the phenomenal character of shape experience in sight and touch; and (ii) if there is no difference in the phenomenal character of shape experience in sight and touch, then "the sameness of property perceived in sight and touch is transparent to the subject"; and (iii) if the sameness of property perceived in sight and touch is transparent to the subject, then cross-modal transfer of information is rational (p. 304). (i) here, of course, is just [T2]. So a discussion of the argument for [T3] involves inter alia a discussion of [T2]. There are reasons to think that the cross-modal transfer of information is rational regardless of whether internalism or externalism is correct (see section 3 and footnote 2). But it is not clear that this argument should convince us of it independently even if we grant that externalism about shape perception is correct. We have already remarked that externalism is independent of the phenomenal space account of shape perception, and so independent of the question whether there is a difference in the phenomenal character of shape experience in sight and touch. Thus, we have no reason to believe (i), and, hence, no reason to believe [T2]. Further, it is not clear we have good reason to accept (ii) on the basis of a connection between the antecedent and consequent. Suppose that shape experience as such has no phenomenal character, and, hence, that there is no difference in the phenomenal character of shape experiences in sight and touch. Is it obvious that the content of experiences of shape should therefore be transparent to the subject? I can see that one may want to hold as a general matter than the contents of experiences are transparent, or at least accessible in consciousness, to their subject -I believe this is correct- but this does not seem to follow from the remark that there is no phenomenal difference between perception of shape in sight and touch. In addition, if one does hold the view that the contents of experience are accessible to the subject, then it is not clear why phenomenal differences would

prevent a subject from recognizing t h a t he perceives the same shape by sight and touch, since, by hypothesis, t h e representational content of the perceptions, even if involving different phenomenal characters, are the same. Thus, it does not seem t h a t phenomenal sameness of shape representation in sight and touch is what is crucial t o question whether cross-modal transfer of information is rational.

6

Conclusion

I conclude t h a t shape properties are categorical, that the crossmodal transfer of information is rational, but t h a t Campbell has given us no reason t o think t h a t any of [TI]-[T3]is correct.

Campbell, John. 1995. "Molyneux's Question". This volume, pp. 301-318. Martin, Michael. 1992. "Sight and Touch". In The Contents of Experience, ed. Tim Crane, 196-215. Cambridge: Cambridge University Press. Davidson, Donald. 1980. "The Individuation of Events". In Essays on Actions and Events. Oxford: Oxford University Press. Gregory, Richard L. 1990. Eye and Brain: the psychology of seeing. Princeton: Princeton University Press. Locke, John. 1975. An Essay Concerning Human Understanding. Oxford: Oxford University Press. O'Shaughnessy, Brian. 1989. The Sense of Touch. Australasian Journal of Philosophy 67: 37-58. Quine, W.V. 1985. "Events and Reification". In Actions and Events: Perspectives on the Philosophy of Donald Davidson, ed. Ernest LePore and Brian McLaughlin, 162-171. Oxford: Basil Blackwell. Shoemaker, Sydney. 1984. "Causality and Properties". In Identity, Cause, and Mind. Cambridge: Cambridge University Press. Wittgenstein, Ludwig. 1953. Philosophical Investigations. New York: MacMillan.


Shape Properties, Experience of Shape and Shape Concepts John Campbell

An internalist about perception of shape thinks that perception includes sensations of shape, which are what make the experience consciousness of shape. These sensations of shape are different in sight and in touch. An externalist about perception of shape thinks that what makes the experience consciousness of shape is the fact that it is responding to shape properties of the objects in the environment. So the externalist about shape perception can hold that shape perception is amodal. In 'Molyneux's Question' I tried to spell out these connections between the question whether shape perception is amodal and the issue of internalism vs. externalism. In particular, I proposed that we achieve a clearer focus on them by using a notion of primitive consciousness of shape, which subtracts from ordinary shape perception any grasp of the causal significance of shape properties. In their rich comments, Brian Loar and Kirk Ludwig raise many interesting problems, and I will not try to be comprehensive in replying to them. I will, rather, try only to clarify my main line of argument, and to add some brief remarks on the relation between experience of shape and concepts of shape. I begin by trying to

make vivid the reason for using the notion of primitive consciousness of shape.

1 Motivating the Idea of Primitive Consciousness of Shape Shape properties have causal powers. The shape of an object affects the ways in which it interacts with other things, and the ways in which one can make use of it. Ordinarily, people who perceive shapes have some knowledge of the causal significance of the various shape properties. Is knowledge of shape properties exhausted by this knowledge of causal significance? Or does shape perception provide a further, ineliminable dimension to our knowledge of shape? Suppose we try to set up the case of someone who has knowledge only of the causal significance of shape. This person is taught a theory about the causal significance of shape properties. This person -suppose we call her Bridget- learns, for example, that to smash something with a rock, it is better to use a corner than a flat surface. Since, as Kirk Ludwig points out, there are infinitely many different shapes, the theory she learns will take the form of a physical geometry, in which causal significance is not assigned directly at the level of the shape of the thing, but rather is given in terms of more basic notions such as 'corner', 'point', 'edge' and so on. The theory will still be formidably complicated, and full of special cases, but let us suppose that Bridget has the cognitive resources to master it. She will be able to derive from it conclusions about the conditional powers possessed by any object which has this or that particular shape. So long as the theory is stated using the shape concepts themselves, it might be acknowledged that mastery of the theory is enough for knowledge of shape properties, but that this does not show that the theory makes explicit everything that is involved in knowledge of shapes. Rather, you might say, there is more to knowledge of shapes than knowledge of their causal significance, but knowledge of the theory is sufficient simply because one could not grasp it without having this further dimension, provided by shape perception. So how will we set up the case so as to ensure that Bridget has knowledge only of the causal significance of shape properties? The obvious approach is to suppose that she is taught the Ramsey sentence for the explicit physical geometry, so that she learns only that there are properties P I , . . . , P,, such that.. . . We can suppose her to learn

this theory without having to draw on any prior grasp of shape properties, so that so far, shape perception does not come into play in her knowledge of the physical geometry. I am not supposing that Bridget lacks shape perception altogether. That would be one way to set up the example, but it is not altogether easy to suppose someone who perceives but has no shape perception whatever, given the spatial nature of the senses. We might bypass this by supposing that Bridget was born in a sensory deprivation tank, so does not perceive shapes or anything else, but nonetheless has the cognitive ability to learn the Ramsey sentence. Then the question would be whether that is enough for her to know what shape properties are. But we might also suppose that Bridget can perceive shapes, is taught the Ramsey sentence but simply does not relate it to anything she perceives: it does not occur to her that the properties the Ramsey sentence talks about are properties she can perceive. The question then is whether Bridget's knowledge of the causal role of these properties, as stated in the Ramsey sentence, is enough to constitute knowledge of shape properties. Can the bare knowledge of causal role constitute knowledge of which properties are in question, or does it need to be supplemented by perception of shapes? We can put the question by asking what Bridget learns when she finally does make the connection between the properties about which the Ramsey sentence talks and the shapes she perceives. On one view, she already knew what shape properties are, and she has simply learnt about another of their causal powers, namely what they look like. The alternative view is that before she made the connection, she did not know what shape properties are, she knew only that there are properties in virtue of whose possession objects will behave thus and so. When she makes the connection with shape perception, she learns which properties these are. In 'Molyneux's Question' I argued, in effect, for this alternative view. When she makes the connection with shape perception, Bridget learns what shape properties are. She learns what the categorical properties of objects are in virtue of which they behave in the ways described by the Ramsey sentence. The argument was that shape properties are explanatory of the behaviour of objects in a way that does not make sense if we think of them merely as collections of tendencies to affect the behaviour of objects. If our knowledge of shape properties really is exhausted by our knowledge of their causal significance, then the explanations we give which appeal to shape properties turn out to be virtus dormitiva explanations. And the usual defences which are given of virtus dormitiva explanations, in terms of their holding open the place for deeper microstructural accounts,

do not apply here. So if we are to make sense of the explanatory role of shape properties, we have to suppose that we understand them to be categorical properties of objects, and think of shape perception as giving us knowledge of shape properties as categorical. It is possible to take a more robust line here, and ask anyhow what is so special about the causal role of a property, why we should single that out as being what constitutes the property. Reichenbach (1958) held that spatial properties and relations can be reduced to temporal and causal properties and relations, and that temporal properties can be reduced to causal ones. This view does give a special place to causality. But the reductionist programme does not seem to work. It can even be argued that we have to appeal to spatial notions in explaining what causality is (Salmon 1984). Spatial properties and relations are just as fundamental as causal properties and relations. Shape is no less fundamental than cause. Once we abandon the reductionist programme, there is simply no reason to suppose that our knowledge of shape is exhausted by our knowledge of its causal significance. There is no reason to resist the common-sense idea that perception of shape gives us knowledge of it as categorical. This conclusion has a striking implication for the way in which we think of the content of shape perception. To bring this out, we can remark one way in which Bridget's case is unusual. Bridget learnt the causal significance of shape by learning an explicit theory. But the more common form for knowledge of the causal significance of shape is this: shape perception is systematically linked to action and to imagistic reasoning, and these linkages constitute one's grasp of the causal significance of shape. In 'Molyneux's Question' I tried to indicate the form which an articulation of these linkages would take. You might say that the content of shape perception is exhausted by these linkages to action and to imagistic reasoning. But that view takes the content of shape perception, as shape perception, to be exhausted by its grasp of the causal significance of shape. All there is to perception of the object as having a particular shape is that it is expected to behave thus and so in various imagined circumstances, and to provide various opportunities for action. So this view gives us no way of understanding how it can be that shape perception is perception of shape as a categorical property of an object, a property which underlies the tendencies of the object to behave thus and so. To understand how shape perception could be perception of shape as categorical, we have to consider what more there could be to the content of the perception than its being appropriately linked to imagistic reasoning and action. We have to give some account of this further dimension of the content of shape perception.

If there is such a further dimension to the content of shape perception, it means that we can, at any rate as a dramatic device, separate it off and call it 'primitive consciousness of shape'. There may be a kind of subtraction fallacy here, in that you might acknowledge that there is a further dimension to shape perception than its connections with imagery and action, but deny that this further dimension is capable of independent survival. But I think that it is helpful to suppose, even if only provisionally, that independent survival is possible, and to look at what is left in shape perception when we separate off the connections with imagistic reasoning and action. This primitive consciousness of shape will be what provides you with knowledge of shape as categorical. As I am here using the terms, internalism and externalism are different ways of characterising this further dimension of shape perception. In terms of the dramatic device, they are different ways of saying what primitive consciousness of shape is. According to the internalist, the further dimension of shape perception is provided by the presence in shape perception of sensations of shape. These sensations of shape are not to be understood as representational aspects of the experience. The problem the internalist is addressing has to do with the determinants of representational content; what more there can be to the representational content of a shape perception than its connections with action and imagistic reasoning. So the appeal to sensations of shape is not to be understood as simply an appeal to the idea that perceptions represent shapes. The appeal to sensations of shape is an appeal to an aspect of perception which, by being systematically connected to shape representation, might explain how shape perception could represent shapes as categorical. So the internalist has to insist on a sharp distinction between sensations of shape and representations of shape. Externalism, in contrast, is the view that this further dimension to shape perception, primitive consciousness of shape, is provided by the categorical properties themselves of which shape perception provides knowledge. It is in consequence of the geometrical aspects of the objects themselves that one's perceptions of them are perceptions of shape as cat.egorica1. This externalism can stay neutral on whether there are sensations of shape as well as perceptual representations of shape. It could be said that there are sensations of shape, and that it is in virtue of the relations between sensations of shape and representations of shape that the shape representations are representations of shapes as categorical; but that the shape sensations have to be individuated externalistically. Alternatively, it could be said that consciousness of shape is exhausted by its representational

aspects, that there are no sensations of shape, but that shape representation has the content it does in part because of the nature of the environmental features to which it is a response, and that it is in virtue of the nature of those environmental features that shape perception represents shape as categorical. The aim of 'Molyneux's Question' was to set out a link between this question of internalism or externalism about primitive consciousness of shape, and the question whether shape perception is modalityspecific or amodal. Loar and Ludwig express considerable puzzlement over the link I see here. So I will set out first the link between internalism and modality-specificity, and then the link between externalism and amodality.

2 Internalism and Modality-Specificity The internalist has to explain in detail how representations of shape are connected to sensations of shape, if the appeal to sensations of shape is to do the work of explaining how it is that we can represent shapes as categorical. I am not here going to consider how the internalist might go about giving this explanation; I want only to make some remarks about the implications of any such view for the modality-specificity of shape perception. The argument for saying that there are different shape sensations in sight than in touch is that we can always tell through which sense we are perceiving a shape. And for the internalist, the subjective discriminability of two sensations constitutes them being sensations of different types. Now if sensations of shape are different in sight and touch, what are the implications of this for the rationality of crossmodal transfer? If someone is trained to respond in a particular way to a tactual shape sensation, and is then presented with a visual shape sensation, and these are quite different sensations, is it rational to then give the response? We need to distinguish this case from the case in which the subject gives the learned response to another presentation of the original tactual sensation, and from the case in which the subject gives the response to the visual sensation because she has some background reason for supposing it would be right to give the same response. In these cases, it does seem to be rational to give the response. If however the subject is giving the response to a quite new sensation, with no background inductive reasoning to motivate doing so, then there is a sense in which giving that response is not a phenomenon of rationality. Of course, it may be that there is some hard-wired innate mechanism that the subject has, which dictates giving the same response to the new visual sensation as the

subject gave to the original tactual sensation. And as Loar remarks, it is not irrational to rely on innate mechanisms. But we still must not obliterate the distinctions here. The internalist position here is really quite fragile; I give it weight because it has some intuitive appeal, and often is what motivates a negative answer to Molyneux's Question, rather than because I think it will stand up to scrutiny. But notice that the reason given for saying that shape sensations are different in sight and touch is not compelling. The reason given is that we can always tell through which sense we are perceiving a shape. But this does not of itself show that the shape sensations are different, it shows at most that there is a difference between the total experience of seeing a shape and the total experience of feeling a shape. It would be consistent with that to hold that some analysis of the experiences is possible, and that the 'shape' aspects of the experiences are exactly the same though there is some further dimension along which the experiences are different, and that it is this further, separate aspect of experience that lets us know which sense-modality is in play. Introspection alone will not let us settle this. The internalist uses subjective matching of sensations as what constitutes them being of the same type. What we have just seen is that this criterion does not let us make all the distinctions we need: we want to ask about the sameness or difference of various aspects of experience, and introspection will not enable us to make the finegrained distinctions that matter here. There are other, obvious problems with subjective matching as a test for sameness of sensation, such as the fact that sameness of sensation is transitive while subjective matching is not. What motivates the appeal to subjective matching is the view that we have some privileged access to our own sensations, that I cannot go wrong in classifying which sensations I am having. If I think I am having an F sensation, I must be right about that; so if I think I am having two sensations which are alike in point of F-ness, I must be right about that too. Internalism elevates that point into a criterion of identity for sensations. But that seems to be a mistake. It leaves us with no way of making sense of the sameness or difference of the sensations had by different people, or by a single person at different times; problems of making sense of sameness or difference of sensation across modality are just another part of this pattern of gaps. And there seems to be no reason to suppose that only internalism can explain such privileged access as we have to our own sensations. It is, in fact, difficult to see how the internalist could even attach any clear meaning to the idea that the same categorical shape prop-

erties are perceived by sight as by touch. What could it mean to say that the visual and the tactual sensations are sensations of the very same property? Experience of correlations between the visual and the tactual is just that, experience of correlations, and does not of itself show that one and the same property is being perceived. But how is the internalist to go beyond the existence of correlations between the visual and the tactual? The causal view of properties suggests one possibility. Suppose we think of a property as a cluster of powers, including conditional powers, and powers to produce various sensations in us. We could find that as a matter of causal law, anything which has the relevant conditional powers plus the power to produce the tactual sensation in us also has the relevant conditional powers plus the power to produce the visual sensation in us. And conversely. So we have here just one causally unified cluster of powers; that is, on the causal theory, a single property. If the remarks I made earlier were correct, though, this approach is on the wrong track. We ought to think of shape perception as perception of shape as categorical, and we have to understand what it means to say that the same categorical property is being perceived through sight and touch. The dramatic device of primitive consciousness of shape helps in stating the problem here. We have to understand how it can be that we have primitive consciousness of the very same categorical properties in sight as in touch. Similarity of the causal significance of the properties is now set aside. And it is not enough for identity just that the properties are correlated. So what can we appeal to as making it the case that it is the very same properties that are being perceived through the two modalities? So long as the internalist persists in supposing that the visual and tactual sensations are quite different, there seems to be absolutely nothing that can make it the case that it is one and the same property that is seen and touched. Loar and Ludwig are quite sanguine about the possibility of combining internalism with the view that shape properties are categorical and that it is rational to suppose that one and the same property is seen and touched. But what we have found is that if you combine internalism with the view that shape properties are categorical, it is hard even to make sense of the idea that the same properties are seen and touched. It is anyhow obscure how internalism can solve the basic problem I set earlier: to explain how it is that we perceive shape as categorical. There is undeniably some immediate appeal to the idea that what makes it so that we perceive shape as categorical is the involvement

of shape sensations. But on reflection it seems that this is because our only understanding of what a shape sensation is already draws on our grasp of what it is to perceive shape as categorical. How the possibility of categorical shape perception could be explained by the existence of shape sensations is opaque: how could the involvement of these inner sensations affect the way in which the properties of external things are being perceived? How could they have any effect on content at all? It seems worth exploring the alternative to internalism.

3 Externalism and Cross-Modal Equivalences The externalist view says that what makes shape perceptions experiences of shape as categorical is that the external properties in question are categorical properties of objects. The environment plays a constitutive role in determining the character of the shape experience. So what makes one's consciousness consciousness of shape is the fact that one is using a neural system whose role is to pick up the shapes of the objects around one. In 'Molyneux's Question' I said that insofar as we are externalist about shape perception, we have to think of it as amodal, because it is the very same external shape properties that are being picked up on by sight as by touch. But to say that shape perception is amodal is not to deny that there can be informative identities involving shape properties; Loar is quite right to point out that there can be informative identities involving shape. But does this make sense? Is it really coherent to say that shape perception is amodal but to acknowledge that it can be informative to be told that it is the very same shape that one is seeing and touching? The quickest way to see that it is coherent is to remark that there can be informative identities involving shape properties within a single sense modality. In vision, perhaps the simplest and most familiar example is the square and the diamond (Palmer 1983). It can be informative to be told that the square is the diamond. Or again, consider a cube lying flat on a table, and a cube balanced on one of its corners. It can again be informative to be told that it is one and the same shape that one is seeing both times. In these cases, of course, it is true that the subject could typically spot the identity a priori, but it is also the case that the subject has to do some cognitive work to spot the a priori identity. I am here setting aside cases such as that in which one has an unfamiliar view of a familiar object, where the informativeness of the identity is owed to the fact that one's different

sightings of an object both give only incomplete information about its shape, and cases in which the doubt about identity of shape arose only because one did not trust one's senses. Anyone has to acknowledge that in a case in which vision gives only incomplete information about the shape of an object, and touch gives only incomplete information about the shape of the object, it may be informative to be told that one is perceiving the same shape twice. And anyone has to acknowledge that if one doubts the deliverances of either or both vision and touch, it can be informative to be told that one is indeed perceiving the same shape twice over. The present point is that within vision, one can have complete information about the shapes of two objects, and one may trust one's senses, but still find it informative to be told that one is perceiving the same shape twice over. This suggests that even if shape perception is amodal, one can have complete information about the shape of an object from vision, and complete information about the shape of an object from touch, and trust one's senses, but still find it informative to be told that one is perceiving the same shape both times. Are there cases, within touch, in which one perceives the same shape twice over and yet it is informative to be told that it is the same shape? Orientation does not seem to be as important in touch as in vision, if only because the reference information needed to establish directions is harder to obtain in blind conditions than when vision is being used. So we may not so easily find cases in which the informativeness of the identity depends on a change in the orientation of the shape. A striking case, though, is that in which we consider tactual comparisons of objects which are the same shape but quite different sizes. In vision it is easy to abstract from the difference in size and visually recognise sameness of shape across difference in size. This is not nearly so easy in touch as in vision: it can be informative to be told that two differently sized objects which one has felt are the same shape, even if one has complete tactual information about the shapes of each of them and trusts one's senses (Millar 1994). In the cases in which we have informative identities involving shape within a single sensory modality, there is a sense in which the identities are a priori, though it may take some cognitive work to come to know them. It is not that the cognitive work involves deductive reasoning. Rather, consider the case in which you see the same shape in different orientations. You can come to know that it is the same shape that you are seeing both times just by thinking about it; but the 'thinking about it' here does not involve deductive reasoning, but a more basic kind of image rotation, which takes place at the level of the perceptual system. It is also possible that someone could see the

same shape in different orientations but just not have a sufficiently powerful capacity for imagistic reasoning to recognise the sameness of shape by image rotation. This does not threaten the claim of the identity to be a priori, any more than the fact that someone can grasp an arithmetical truth without having the mathematical competence to prove it shows that the arithmetical truth is not a priori. In the tactual case, someone who feels differently sized objects of the same shape may be able to determine, just by thinking about it, that they are of the same shape, given complete tactual information about the shapes of the two objects. Of course, this may not be easy to do, but the identity may again be a priori, even if one does not have the cognitive capacity to recognise it oneself. Something similar may be true of the identity of shapes perceived by sight and by touch. It may be that the identity is a priori, and that it can be recognised just by thinking about it, perhaps by imagistic reasoning. If you do recognise the equivalence of the seen and touched shapes in this way, then there seems to be quite a strong sense in which cross-modal transfer is rational in your case. If you are trained to respond to a touched shape in one way, then are visually presented with that shape, you may give the learned response because you have by reasoning recognised the identity of the shape presented. This is a quite different case to that in which you are trained to respond to a tactual sensation in one way and then give that response to a quite different visual sensation, simply because you have been hard-wired to respond to that visual sensation in whatever way you respond to the tactual sensation. So the externalist holds that cross-modal transfer may be rational in a sense in which the internalist does not allow it to be rational. Even if you do not have the cognitive machinery required to enable you to recognise the identity of the seen and touched shapes -as anyone might have a deficiency in the capacity for imagistic reasoningso that you cannot yourself determine the identity of the seen and touched shapes just by thinking about it, still we have not returned to the radical incommensurability of visual and tactual presentations of shape postulated by the internalist; an incommensurability which, as we saw, threatens to make it unintelligible how vision and touch could be perceiving the very same categorical properties of objects.

4 Shape Perception and Shape Concepts Loar raises the question of the relation between perception of shape and grasp of shape concepts; I conclude with some brief remarks on

this. It does seem to me that we can make a distinction between two types of representation here, in terms of the kinds of operations that can be performed upon them. At the level of perception we have representations which are involved in imagistic reasoning: the rotations and zooms with which Shepard and others have made us familiar. And these representations are used in guiding action. Even an animal incapable of conceptual thought can engage in perceptually guided action. At the conceptual level we have representations which are involved in deductive reasoning: reasoning involving the application of the laws of logic. Imagistic reasoning and deductive reasoning seem to be quite different types of operation, and straight off it would seem that they must be being applied to quite different types of representation. To be capable of shape perception, one must have perceptions of a sort which can be operated on in imagistic reasoning, and which can be used in the control of action. If the above line of argument has been correct, there is a further dimension, which the internalist tries to capture by saying that one must have sensations of shape, and which the externalist tries to capture by saying that one must be suitably related to the shape properties of the objects around one. But what does it take to grasp shape concepts? To have a shape concept is to be able to form representations of shape which can figure in deductive reasoning. There must be some connection between grasp of shape concepts and the capacity for shape perception. Presumably, if you are to have conceptual knowledge of the shapes of things, you must be able to make shape judgements on the basis of perception of shape. But there is another connection between shape concepts and shape perception. Recall the story of Bridget with which I began. If what I said there is correct, then grasp of shape concepts depends on shape perception because without the shape perceptions you simply would not know what shapes are; you would not know which categorical properties the shape concepts identify. What more is needed than shape perception for grasp of shape concepts? So far I have said only that shape concepts must figure in deductive reasoning. But we can be a bit more explicit than that. To grasp a shape concept you must be capable of using it in subject-predicate thoughts -you must be able to apply the concept to objects. So the laws that you can use will include the laws of identity, the quantifier rules, and so on. You must also be able to vary the tenses of your ascriptions of shape. You must be able to say not just that this (currently perceived) object is (now) a cylinder, but that it was or will be a cylin-

der. There is a difference between the kind of time-reference that is possible at the purely perceptual level and the kind of time-reference that is available in conceptual thought . Why should there be such a further level of representation of shapes? It is a datum that we can manipulate representations of shape deductively, and that we do have ways of varying tense, but is this just a kind of epiphenomenon? What kind of reasoning would really exploit the fact that one can use shape concepts in these ways? I think the answer is that explicit causal reasoning involving shape concepts puts to work precisely those types of deductive reasoning (Campbell 1986). To put to work a grasp of causal laws governing the way in which the shape of a thing affects its behaviour, you have to be able to ascribe the shape property to each of a range of objects. Your grasp of what shape property is has to be such as to let you make sense of its being possessed by any of a range of physical things. And to articulate the causal relations between the objects around you over time, you have to be able to ascribe shape properties -and other physical properties- to objects as had at various times. So you have to put to work the ability to vary tenses in ascribing shapes to things. So I am in agreement with Loar that we manifest our grasp of shape concepts by putting them to work in explicit theoretical reasoning. But I do not think that this is what explains how we perceive shapes as categorical; for that explanation, we have to look elsewhere.

Campbell, John. 1986. 'Conceptual Structure'. In Charles Travis (ed.), Meaning and Interpretation. Oxford: Blackwell. Millar, Susanna. 1994. Understanding and Representing Space. Oxford: Oxford University Press. Palmer, Stephen E. 1983. 'The Psychology of Perceptual Organization: A Transformational Approach'. In J. Beck, B. Hope and A. Rosenfeld (eds.), Human and Machine Vision. New York: Academic Press. Reichenbach, Hans. 1958. The Philosophy of Space and Time. New York: Dover. Salmon, Wesley C. 1984. Scientific Explanation and the Causal Structure of the World. Princeton: Princeton University Press.

Contributors John Biro, University of Florida at Gainesville Ned Block, Massachusetts Institute of Technology Robert Brandom, Pittsburgh University Alex Byrne, Massachusetts Institute of Technology John Campbell, University of Oxford Fred Dretske, Stanford University Allan Gibbard, Michigan University Roger Gibson, Washington University Gilbert Harman, Princeton University Paul Honuich, Massachusetts Institute of Technology Mark Johnston, Princeton University Jaegwon Kim, Brown University Brian Lour, Rutgers University Kirk Ludwig, University of Florida at Gainesville William Lycan, University of North Carolina John McDowell, Pittsburgh University Sydney Shoemaker, Cornell University David Sosa, Dartmout h College Ernest Sosa, Brown University Robert Stalnaker, Massachusetts Institute of Technology James Tomberlin, California State University at Northridge Michael Tye, Temple University Enrique Villanueva, Universidad Nacional Aut6noma de Mkxico

Enrique Villanueva is now Research Fellow at the Centro de Neurobiologia at the Universidad Nacional Aut6noma de Mkxico in Juriquilla City, Querktaro, Mkxico, a university where he has been doing research and teaching since he completed his Graduate Study at the University of Oxford in 1972, and where he is carrying on research on philosophy of cognitive science. He has published a number of essays in Philosophy of Mind, Metaphysics of Persons and Philosophical History. Besides editorship of the present series he has edited the series Simposio Internacional de Filosofia, and the volumes Information, Semantics and Epistemology and El argument0 del lenguaje privado. He is author of Lenguaje y Privacidad, Ensayos de Historia Filosdfica and Las Personas.

Perception

Perception

Perception

Perception

Perception

Perception

Perception

Perception

Perception

Flavor Perception

Direct Perception

Tracking Perception

Depth Perception

Death Perception

Death Perception

Visual Perception

Auditory Perception

Death Perception

Direct Perception

The Problem of Perception

Science, Perception and Reality

Perception And Knowledge

Extra-Sensory Perception

Theories of Visual Perception

Perception Of Death

Externalist Theories of Perception

Concepts in Perception

Perception and Reason

The Nature of Perception

Encyclopedia of Perception

Perception

Perception

Perception

Perception

Perception

Perception

Perception

Perception

Perception

Flavor Perception

Direct Perception

Tracking Perception

Depth Perception

Death Perception

Death Perception

Visual Perception

Auditory Perception

Death Perception

Direct Perception

The Problem of Perception

Science, Perception and Reality

Perception And Knowledge

Extra-Sensory Perception

Theories of Visual Perception

Perception Of Death

Externalist Theories of Perception

Concepts in Perception

Perception and Reason

The Nature of Perception

Encyclopedia of Perception

Recommend Documents