Content, Consciousness, and Perception: Essays in Contemporary Philosophy of Mind

Content, Consciousness, and Perception: Essays in Contemporary Philosophy of Mind Edited by Ezio Di Nucci and Conor Mc...

Author: Ezio Di Nucci and Conor McHugh

39 downloads 928 Views 1MB 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

Content, Consciousness, and Perception: Essays in Contemporary Philosophy of Mind

Edited by

Ezio Di Nucci and Conor McHugh

Cambridge Scholars Publishing

Content, Consciousness, and Perception

Content, Consciousness, and Perception: Essays in Contemporary Philosophy of Mind, Edited by Ezio Di Nucci and Conor McHugh This book first published 2006. The present binding first published 2008. Cambridge Scholars Publishing 12 Back Chapman Street, Newcastle upon Tyne, NE6 2XX, UK British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Copyright © 2008 by Ezio Di Nucci and Conor McHugh and contributors All rights for this book reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior permission of the copyright owner. ISBN (10): 1-84718-775-7, ISBN (13): 9781847187758

TABLE OF CONTENTS

Editors’ Preface ..................................................................................................ix Chapter One: Introduction The State of Mind Conor McHugh and Ezio Di Nucci ....................................................................1 1 The mental and the physical ..........................................................................3 2 Content, perception and concepts ................................................................11 Bibliography...................................................................................................31 Part I: Concepts and Mental Content .............................................................35 Chapter Two Concepts: Representations or Abilities? Hans-Johann Glock...........................................................................................36 1 Are concepts representations? .....................................................................38 2 Are concepts signified by computational/neural signs?...............................45 3 Are concepts abilities?.................................................................................48 4 Concepts or concept-possession? ................................................................49 5 Is concept-possession a disposition or ability? ............................................51 6 Is the pragmatist explanation of concept-possession circular? ....................53 7 Are concepts more finely individuated than abilities?.................................57 Bibliography...................................................................................................59 Chapter Three Human Philosophy: Hume on Natural Instincts and Belief Formation Treasa Campbell ...............................................................................................62 1 The traditional use of the term “natural belief” ...........................................63 2 Wolff's propensity/disposition distinction ...................................................66 3 Natural instincts...........................................................................................67 4 Conclusion...................................................................................................72 Bibliography...................................................................................................72

vi

Table of Contents

Chapter Four Can Conscious Attention Restore Uncritical Realism? An Examination of a Neo-Russellian View of Reference Gloria Ayob........................................................................................................74 1 Identifying Wittgenstein’s worry about the Russellian view of reference ............................................................................................76 2 Campbell’s response: Conscious attention to an object does not require sortal classification...............................................................79 3 Refining the Wittgensteinian worry about the Russellian pre-reflective view..........................................................................................81 4 A problem for Campbell’s neo-Russellian semantics..................................85 Bibliography...................................................................................................87 Chapter Five On Peacocke's Theory of Concepts Asunción Álvarez...............................................................................................89 1 The theory....................................................................................................90 2 Reference.....................................................................................................92 3 Implicit conceptions ....................................................................................98 Bibliography.................................................................................................101 Part II: Consciousness and the Metaphysics of Mind .................................103 Chapter Six Deflationism about Qualia Dave Ward .......................................................................................................104 1 The explanatory gap ..................................................................................104 2 The obligations of a deflationist theory .....................................................107 3 Intrinsicality...............................................................................................111 4 Ineffability .................................................................................................114 5 Privacy.......................................................................................................118 6 Simplicity ..................................................................................................118 7 Direct availability ......................................................................................120 8 Conclusion.................................................................................................121 Bibliography.................................................................................................123 Chapter Seven Heidegger and AI: Technological Metaphor and Self-Understanding Richard Hamilton............................................................................................125 1 Introducing the subject ..............................................................................125 2 Reflective self-understanding ....................................................................128

Content, Consciousness and Perception

vii

3 Heidegger ..................................................................................................130 Bibliography.................................................................................................137 Chapter Eight Causal Exclusion and Overdetermination Markus E. Schlosser........................................................................................139 1 Non-reductive physicalism ........................................................................139 2 The causal exclusion argument..................................................................141 3 Events and property instantiations.............................................................143 4 Conclusion.................................................................................................155 Bibliography.................................................................................................155 Part III: Perception…………………………………………………………...157 Chapter Nine Metaphysical Disjunctivism and the Intentional Theory of Perception Francisco Pereira.............................................................................................158 1 Introduction ...............................................................................................158 2 Considering the argument from hallucination ...........................................163 3 Perceptual experience and distal objects ...................................................166 4 Phenomenal indiscriminability ..................................................................168 5 Fundamental ontological asymmetry.........................................................172 Bibliography.................................................................................................177 Chapter Ten Change Blindness and Counterfactual Dependence Nick Jones ........................................................................................................179 Bibliography.................................................................................................189 Chapter Eleven An Enactive Theory of Phenomenal Intentionality Julian Kiverstein .............................................................................................190 1 What is phenomenal intentionality? ..........................................................192 2 Against reductive representationalism.......................................................194 3 Perceptual content and demonstrative thought ..........................................199 4 Towards an enactive theory of phenomenal intentionality ........................201 Bibliography.................................................................................................210 Contributors ....................................................................................................212 Index.................................................................................................................214

EDITORS’ PREFACE

This book is about the mind. The introductory essay is intended to provide, for non-specialists, an overview of the concerns and themes of contemporary philosophy of mind, and to locate each chapter of the volume within the context of that manifold of overlapping concerns and themes. The essays themselves, on the other hand, constitute academic contributions to their respective specialties within the philosophy of mind. We hope, therefore, that the volume will be of interest both to specialists and to readers interested in getting to grips with what philosophers of mind do. All the essays that compose this collection were presented on June 30th and July st 1 2005, in Edinburgh, at the Mind2005 conference. The book could never have happened without the conference, so we want to take this chance to thank everyone involved with Mind2005. Our gratitude for making Mind2005 a success goes in the first place to our distinguished keynote speakers, Prof. Hans-Johann Glock, Prof. Christopher Peacocke, and Prof. Timothy Williamson. Their feedback at the conference improved all of the papers in this collection. Also, we want to thank Dr. Peter Milne, Head of Philosophy at Edinburgh University for all kinds of support. Thanks are also due to the conference’s other sponsors, the Scots Philosophical Club, the Scottish Postgraduate Philosophy Association, the Mind Association, and the Analysis Trust. Thanks also to Zoë Payne for helping us organise the conference and for setting up its website; to Astrid Johnston for a beautiful conference poster; to Lisa McKeown for helping us with submissions. There are numerous others, including speakers, commentators, chairpersons and helpful friends, who contributed to the conference, and to whom we are grateful. We want to thank our supervisors, Dr. Matthew Nudds (CM) and Dr. Bill Pollard (EDN). Conor McHugh is generously supported by AHRC doctoral award 2004/107894, and by a scholarship from the College of Humanities and Social Sciences at Edinburgh University. Ezio Di Nucci is supported by a scholarship from the University of Salerno. Finally, thanks to Amanda Millar and Andy Nercessian at Cambridge Scholars Press for making this book possible. Edinburgh, 24.4.2006

CHAPTER ONE: INTRODUCTION THE STATE OF MIND

CONOR MCHUGH AND EZIO DI NUCCI

“Mental processes just are queer,” Wittgenstein says in the Investigations (Wittgenstein 1953, §363).1 The bulk of the peculiarity of the mental—mental processes, events and states—is that it is both in the world and about the world. The mind has a unique perspective from within. The mental is in the world because it is part of the natural universe; and it is about the world because it represents it. Remarks like Wittgenstein’s reflect the widespread assumption that we should treat all the processes, events and states we call “mental” as constituting a unified set of phenomena or natural kind—or at least as exhibiting the same sort of “queerness”, and therefore as posing a unified philosophical problem. One might wonder if there is any justification for approaching matters in this way. There is, after all, an impressive diversity among the states and episodes that we label “mental”: why should we suppose that a momentary pain in the leg and a longstanding belief that the earth is flat ought to be understood as the same sort of element of reality (c.f. Rorty 1979)? What sort of property do they have in common? Candidates for unifying properties have included that of being composed of an irreducible metaphysical substance (most famously by Descartes; see, more recently, Robinson 2003); that of standing in certain causal-functional relations (see Putnam, e.g. 1967; Dennett 1991); a biological property (Searle 1992); and that of having a conscious subject (Kant 1929; and see more recently, e.g., Eilan 1995). One influential view has it that intentionality should be seen as the feature that unifies the mental. That is, the peculiar fact that (say) beliefs and pains can represent the world is what makes them mental. Intentionality will be the subject of 1

Thanks to Dave Ward and to Tom Roberts, whose comments on an earlier draft of this paper improved it considerably. The authors are responsible for any remaining errors. Thanks also to Nick Jones for suggesting the title.

2


the second part of this introductory essay. Whatever one’s view of what unifies the mental, there is no doubt that the notion of the mental is closely tied to the notions of consciousness and intentionality. Furthermore, minds are natural; their origin is different to that of other things that exhibit intentionality (words, for example, are human artefacts). Philosophy does not need to explain the origin of the mind (again, see Wittgenstein: we, as philosophers, are not doing natural history2), but that minds naturally emerged from the non-mental universe contributes to their being so queer.3 The queer nature of mind is what philosophy must explain; it must do so, says the fashion of the time, in a naturalistic framework; that is, approximately, philosophy mustn't use in its explanation anything that would make a scientist uncomfortable. It is widely believed that this explanation will depend heavily on facts about another element of the natural order: brains. But how do mental phenomena relate to brain phenomena? (This is another version of the metaphysical question that bugs the philosophy of mind: what sort of thing is the mind?) Ask people to point their index finger at their mind; some people will promptly point in the direction of their head; some others will stare back at you, because you have not fooled them. There is a relationship between mind and brain, or at least between mental phenomena and certain brain phenomena; some have thought it a relationship of identity (Smart 1959), others of supervenience (Kim 1966).4 But it seems peculiar to try to point to the mind; the grammar of pointing belongs to body-talk rather than mind-talk.5 In this introductory essay we aim to put the articles of this collection into context, by describing the state of philosophy of mind in the 21st century—the philosophical problems which the articles herein are addressing, and some of the interrelations between those problems. In the first part of the essay we will discuss the metaphysical question of whether the mental can be accommodated within a naturalistic worldview, without sacrificing what we typically think of as essential features of our conscious mental lives. The second part of the essay will discuss the notion of intentionality, drawing out its connections with the philosophy of perception and the theory of content and concepts. 2

See Wittgenstein (1953, §II, xii). So much so that some deny that the mental emerged from the non-mental, because there is no such thing as the non-mental; see Chalmers’s panpsychism (Chalmers 1996). 4 There are philosophers who think that the mind is in the brain; or at least who talk as if the mind is in the brain. Recently, at the 2005 APA Eastern Meeting in New York on Dec. 28th, John Searle repeatedly referred to the mind as being in the brain. In such circumstances, one is tempted to volunteer to cut someone’s head open and look for it. 5 For another version of the idea that the mind is not inside the brain, see recent proposals that the mind lies in the interaction between brain, body, and environment – see Clark (2001), also Hurley and Noë (2003). 3

Essays in Contemporary Philosophy of Mind

3

1 The mental and the physical In this section we address three different but interrelated questions that bear on the metaphysics of mind: (1) whether there can be psychophysical laws—deterministic laws that rule upon the workings of the mind by linking it to the physical; the existence of such laws would entail a reduction of the mental to the physical; (2) whether the subjective quality of our conscious experience—our qualia—can be reconciled with a physicalist world-view; (3) whether mind can be artificially re-created by man. All three questions have implications for what view we should take of the metaphysical status of the mind within nature. Perhaps they can all be seen as ways of asking the same question—“what sort of thing is the mind?”—but if one does not agree that all three questions address the same general problem, then one can just see (1)-(3) as giving a rhetorical arrangement for the discussion of the mental and the physical. We will mention, but not refute, one common objection to the claim that these three questions are formulations of the same problem. The objection is based on the distinction (Chalmers 1996) between easy and hard problems of consciousness. Easy problems are those that can be solved by the explanation of some mental phenomenon in terms of functional or computational facts about the brain (how the brain processes information). A hard problem, on the other hand, cannot be solved in functional or computational terms. Chalmers thinks that there is a part of the philosophy of mind which can be conceded to reductionists: problems such as, for example, explaining the nature of propositional attitudes and their role in causing behaviour, are easy problems, and thus materialists are right in thinking that they can be understood in physical terms. The materialist accounts will be massively complicated (and are in many cases still missing), but such an account can be given for these problems.6 But explaining the existence and nature of qualia—the subjective properties of conscious experience in virtue of which there is something it is like to enjoy such experience—constitutes a hard problem, according to Chalmers. He believes there is no physicalist solution to the hard problem. If one, with Chalmers, thinks that there is not one unified problem of the metaphysical status of the mind, but two different sorts of problems demanding different solutions, then one will not agree that formulation (1) addresses the same problem as formulation (2), because there will be laws for some features of the mind, namely those to do with how the brain receives and processes information, but not 6 One might be tempted to add, “and one day we shall have such a massively complicated physicalist story”; but it must be emphasized that for physicalism to be true, in its application to easy problems or to any problems, it is not required that man actually discovers the physicalist story.

4


for others, namely those to do with the subjective character of experience; and one will think that formulation (3) must be interpreted in such a way that only the former features can be artificially reproduced.

1.1 The anomaly of the mental Donald Davidson, in “Mental Events” (Davidson 1970), states mental anomalism (sometimes, as for example in Schlosser's contribution to this volume, called “psychological anomalism”) as the principle that “there are no strict deterministic laws on the basis of which mental events can be predicted and explained”7; thus mental events are contrasted with non-mental physical events, such as one’s pen falling to the ground, which fall under deterministic laws. If laws could be established which explained and predicted our mental events, on the basis of, say, brain-states, then there would be a reduction of the mental to the physical via psychophysical laws. The problem of the mental would dissolve; we would have reconciled the mental with the natural, and the philosophy of mind could have peace. But the anomaly of the mental is, according to Davidson, an “undeniable fact” (ibid., p. 207), and one that makes us human. It is not that philosophy must remove the anomaly, but rather that philosophy must show how mental anomalism is consistent with the dependence of the mental upon the physical. Mental anomalism can be seen as an expression of the intuitive thought that our mental lives constitute a realm of rational freedom, a realm incompatible with the determinism of natural law.8 Davidson aims to capture that thought, while maintaining that our mental lives are in causal interaction with the physical, and without adopting an implausible metaphysical dualism. According to Davidson, there is only one metaphysical sort of stuff, but it can have two different sorts of properties—physical and mental. His claim is that there can be no reduction of one sort of property to the other: mental properties, which are in the realm of reason, cannot be assimilated within the realm of law. Davidson points to normative constraints on the correct attribution of mental properties, that have no parallel at the physical level. That is, in order for a mental event, enjoyed by a particular subject, to have the property of, say, having a certain intentional content, that event must play a certain rational role in the subject's cognitive and behavioural economy.9 This role is constitutive of the event's having 7

Davidson 1970, p. 208 (page numbers for Davidson refer to the 2nd edition (2000) of his Essays on Actions and Events). 8 Davidson says he feels “in sympathy with Kant” (ibid., p. 207), where the latter says that the apparent contradiction between freedom and natural necessity must be “eradicated” (quoted in Davidson 1970, p. 207; from Grundrisse). 9 We say more about this below, §2.1.


5

that mental property. No such constraints are in force at the level of physical properties: at the physical level, there are no normative or rational relations, only causal and descriptive ones. Thus, there is a mismatch between the constraints on correct attribution of mental properties and the constraints on correct attribution of physical properties. Davidson sets out to show that there is no contradiction between the following commitments, each of which is individually plausible: (a) the principle that “mental events interact causally with physical events” (ibid., p. 208); (b) the principle of the nomological character of necessity: “where there is causality, there must be a law: events related as cause and effect fall under strict deterministic laws” (ibid., p. 208); (c) mental anomalism. The contradiction seems to arise from the fact that the first two principles imply that mental events are subject to psychophysical laws, which is what mental anomalism denies. There are at least four possibilities: one can side with Davidson and attempt to reconcile the three principles, or one can drop any of the three principles in such a way as to avoid the apparent contradiction. Epiphenomenalism, for example, would deny principle (a) (see Jackson 1982 and 1986), and claim that the mental does not interact with the physical. Unlike interactionist dualism, epiphenomenalism can maintain that mental properties are non-physical, without denying that the realm of the physical is causally closed. On the other hand, epiphenomenalism leaves the mental as something explanatorily redundant: it denies that, for example, you cry out because you feel pain. This is, surely, highly counterintuitive. On top of this, epiphenomenalism is false to the phenomenology of our introspection, since it is surely part of our phenomenology that our thoughts sometimes bring about our behaviour.10 Abandoning principle (b) does not seem a promising move: (b) merely expresses an understanding of causality that is common to almost all philosophical views of causality, both realist and anti-realist. What about (c)? Dropping mental anomalism amounts to claiming that there are psychophysical laws and that therefore the mental can be reduced to the physical. There is a fundamental disagreement between Davidson and those who propose to drop mental anomalism: Davidson thinks that mental anomalism is part of what the philosophy of mind must explain; his opponents think that it is a red herring. In an 10 This is not supposed to be a grand claim about mental causation, nor a claim about action being necessarily brought about by thoughts. We happen to think (see Di Nucci, unpublished MS) that some actions come about automatically, without any previous thinking on the part of the agent.

6


important sense, the defender of (c) and its opponent disagree upon the subjectmatter rather than its explanation. Davidson's own solution is a version of the identity theory, “a theory that identifies at least some mental events with physical events” (ibid., p. 209). Mental events that are identical to physical events are such that they have both a mental description (what Brentano 1973 called an intentional description, such that the event is about the world)11 and a physical description—that is, they have both mental and physical properties. Davidson calls his position “anomalous monism”, because it holds that there are no non-physical events, but it denies the existence of psychophysical laws linking mental and physical properties. This is because Davidson accepts identity of particular mental and physical events (event-tokens), but not identity of the mental and physical properties (event-types) possessed by those events; and only the latter would imply the existence of psychophysical laws. There are no laws because laws link properties, and, as we saw, there are normative constraints on correct attribution of mental properties (application of mental descriptions), which have no parallel at the physical level. Markus E. Schlosser, in his contribution to this volume (Chapter Eight), defends non-reductive physicalism from its best-known challenge, the causal exclusion argument. The exclusion argument is supposed to show that non-reductive physicalism is implausible because it holds on to both mental causation (the thesis that mental events cause physical events) and to the causal closure of the physical (the thesis that every physical event has a sufficient physical cause). According to the argument, the conjunction of these claims is inconsistent with the principle that events are not typically overdetermined by their causes—they do not typically have more than one sufficient cause. Schlosser sets out to show that at least three versions of the argument fail. The version according to which mental events are token-identical with physical events fails because it cannot rule out the causal relevance of mental properties. The version according to which mental events are metaphysically included and determined by physical events fails because, if physical events include mental events, then the two cannot be rival causes. Finally, the version according to which mental events are entirely distinct from physical events fails because, mental events and physical events not being independent causes, they cannot overdetermine their effect.

1.2 Consciousness Even if our experience of the world and ourselves could be predicted and explained through laws, just as the rest of the physical universe is understood through laws, it would still appear to us as if our experience were a unique realm. 11

See below, §2.1; and see Searle (1983).


7

That is, there remains something inherently peculiar (queer) about our experience, even if the strictest version of physicalism turns out to be true (and, as we stand, we don't know whether physicalism is true; let alone which version of it is true). Introspection tells us that our experience is private in a way that nothing else is; and it possesses intrinsic sensory qualities that elude characterisation in physical terms. The colour of your favourite jumper; the smell of a woman's hair; the salty taste of the sea; the feeling of being back home; the sight of the fire on in your grandfather's house.12 This aspect of experience has been called “what it is like” (Nagel 1974), phenomenal consciousness (Block 1995), qualia (Jackson 1982). The physicalist must understand qualia as physical properties, but every physical description seems to miss something about what an experience feels like to its subject. This puzzle about experience is illustrated by Jackson (1986) with a famous thought-experiment. Mary, a world-leading colour-vision scientist, not only knows everything that science knows about colour-vision, but knows all the physical facts about colour-vision. But Mary has never seen colours, because she has spent all her life in a black and white cell. Jackson asks whether, when Mary leaves the cell and sees colours for the first time, she learns a new fact—whether she acquires some information that she didn't have before, information about what it is like to see colours. It seems that, since Mary already knew all the physical facts, if she learns a new fact in leaving the cell, then that must be a non-physical fact. This is the position that Jackson himself endorses in his paper: Mary does learn a new fact in seeing colours for the first time; and since this new fact cannot be physical, Mary must learn a non-physical fact.13 Therefore, according to Jackson, there are nonphysical facts about phenomenal experience (qualia). Jackson’s paper seems to illustrate that our conception of consciousness allows for room between facts about consciousness and physical facts. The widely shared intuition that Mary learns something new when she sees colours supports the claim that facts about conscious experience are somehow different; and, perhaps, the claim that the phenomenal character of experience cannot be captured by anything that the physical sciences could tell us. Philosophers have deployed a plethora of other imaginary scenarios to support such a claim. For example, Chalmers (1996) asks us to imagine a zombie: a molecule-for-molecule replica of you, who lacks conscious experience. Although the zombie is physically just like you, and 12 Note that these examples have been deliberately chosen to show that the problem of phenomenal experience does not only apply to what are, after Locke, usually called secondary qualities, like colours or smells. It applies to any object of experience; and therefore also to what Locke called primary qualities. The literature on qualia is full of secondary qualities examples, like in Jackson (1986), only because they make the problem more apparent. 13 See Jackson (2003) for his recantation of this view.

8


therefore behaves just like you, it has no qualia. There is nothing it is like to be a zombie. The apparent ease with which we can entertain the supposition of zombies seems to suggest a gap between consciousness and the physical—just as Jackson urges with his case of Mary, the colour-vision scientist. What is interesting about Mary's thought-experiment is its extremism: it seems to force one either to concede that Mary does discover something new, namely a non-physical fact, or to suppose that, even though Mary has a new experience, she learns nothing new from it. It is not within the scope of this chapter to answer Jackson's challenge; we have presented his thought-experiment only as a very appealing way of introducing the puzzle about qualia. But we will mention David Lewis’s response. Lewis (1988) distinguishes between propositional knowledge— knowledge of facts, or knowledge that something is the case—and nonpropositional knowledge—which includes knowledge of how to do things, and knowledge of objects, also called “acquaintance”. Lewis claims that Mary, before leaving the cell, has all the propositional knowledge one can possibly have about colours and colour-vision. What she learns in coming out is some non-propositional knowledge, or know-how, about colours; she acquires the ability to recognize and distinguish colours. Note that Lewis here is very close to contemporary accounts of phenomenal experience according to which qualia just are, or are the base for (depending on the particular account) the subject's abilities to sift, sort, and track objects.14 Why should Mary learn something new? Because, according to Lewis, the know-how she obtains can be acquired only through direct experience of the object of knowledge (colour); this is unlike propositional knowledge, which can be acquired through, for example, testimony. That does not violate the example's stipulations, because Lewis isn't limiting Mary's knowledge of all the physical facts, which is purely propositional; and at the same time it does account for Mary's learning something new. Lewis thinks that his account vindicates materialism. Whether that's true isn't the point here. What is interesting is that even Lewis has to concede some irreducibility for the character of phenomenal experience: namely, that it can be discovered through experience alone. One cannot explain what it is like to see the sun setting on Edinburgh; or, rather, anyone's attempt to explain what it is like to see the sun setting on Edinburgh will fall short of finding out for oneself what it is like. There is only one way to find out: looking out of the window to see the sun setting behind Appleton Tower. There is an important lesson to be learned from this. Experiences are, necessarily, someone's experiences. It is their nature to belong to a subject and to be inseparable from that subject in so far as they are that subject's experiences. This is what is often expressed by calling them subjective. Experiences are subjective in 14

See, in this volume, Ward's discussion of Petit (2003).


9

the sense that they have to be some subject's experiences. The important point is that it is not required, for some form of physicalism to be true, that experiences can be somehow detached from their subjects. The challenge for physicalism is to provide an account of the character of phenomenal experience that is consistent with our naturalistic framework. A plausible physicalist account will naturally concede that experiences are subjective in the sense, outlined above, that they are essentially some subject's. The sense of subjectivity that physicalism will not and cannot concede is that the character of phenomenal experience is explained by mysterious non-physical entities. Dave Ward, in his contribution (Chapter Six), argues against the supposed mysterious character of qualia. He distinguishes five intuitively plausible features of qualia which appear problematic for the physicalist, and shows how each of them can be accounted for within a naturalistic framework. The features are intrinsicality, ineffability, privacy, simplicity, and direct availability to consciousness. Ward takes what he calls a “deflationist” approach: although nothing actually has all of these five features, an explanation is available for why it appears to us as if our experience does have these features. In brief, his position is that ineffability, privacy and direct availability are genuine features of experience, and that intrinsicality and simplicity are not. However, both the genuine features and the introspective evidence for intrinsicality and simplicity can be explained by appealing to the fact that the computational mechanisms giving rise to experience are massively complex, and that we have access to the deliverances of these mechanisms rather than the mechanisms themselves, since the mechanisms lie at the subpersonal level.

1.3 Mind as artefact Arguments like Jackson’s try to show that the mind cannot be captured fully in physical terms. Proponents of Artificial Intelligence disagree. If the mind can be artificially recreated by man, then it must not be so mysterious. Thus there is a fundamental relationship between AI and the philosophy of mind. If it were possible to recreate the mind artificially, then reductionists would have won the day. It is worth clarifying at the outset that the converse is not immediately true. It might be that reductionists are right even if the mind cannot be artificially recreated by man: there might be some nomological limitations upon man such that we cannot hope to re-create the mind, and this would still be compatible with the mind being part of the physical world. So AI has a particular place in the debate: its success could, theoretically, provide an argument for reductionism, though its failure cannot provide an argument against reductionism. There is a general objection to the connection between AI and the mind: if the mind's evolutionary history is essential to the mind being what it is—to the mind’s

10


essence—then AI is necessarily cut off from many philosophical questions about the mental. Man cannot replace nature, and that is independent of how similar the artificial mind can be to the natural mind. The artificial mind will always have a different history, and therefore it will never duplicate the natural mind. Searle (1992) seems to hold a similar position when he says that consciousness is a biological phenomenon, and that the mind is essentially organic. So, according to this objection, there is at least one thing that will always distinguish AI and the mind. There are at least two replies to this way of diminishing the importance of AI. First of all, one could take issue with the idea that the history of something is part of its essence. One could hold that the essence of the mind, and what philosophy of mind is interested in, is what the mind does: its functions and phenomenology. And even if that is causally related to the way in which the mind evolved, the history of the mind is not constitutive of what it is and does. Thus, to duplicate the mind it would not be necessary to duplicate its history; it would be sufficient to replicate what it does. Secondly, one could attack the motivation for the objection. The motivation seems to be the thought that the human mind is part of the natural, biological world, in which we understand the nature of things in terms of their evolutionary history, whereas an artificial mind could not be natural in this sense. But this thought involves an assumption about what counts as natural and what doesn't—about, so to speak, what is the will (or work) of God and what isn't. One could take a Humean line (see Hume 1777), and question this conception of “natural”: why should human artefacts, such as a mind created in a lab, not count as natural? One needs an argument for distinguishing between natural and non-natural things, when they both happen in the natural world. We can distinguish two different theses: Strong AI and Weak AI. Strong AI is the thesis that the mind is a computational system, and that therefore the mind could be artificially recreated. Weak AI is the thesis that even though a computational system could simulate the functions of the mind, a computational system could never be a mind.15 Therefore, according to Weak AI, the mind cannot be artificially recreated. The distinction between Strong AI and Weak AI gives rise to three different philosophical positions with respect to AI: one can accept Strong AI; one can reject Strong AI but accept Weak AI; or one can reject them both. The previous historical objection to AI can be mapped onto this distinction between Strong AI and Weak AI. One could think that if, as the objection goes, a mind’s history is essential to something being a mind, then the objection applies to 15

The term “Weak AI” is sometimes used to refer to a weaker view, that is agnostic about Strong AI, rather than a view that rejects Strong AI, as we have characterised it. The stronger characterisation suits for our purposes.


11

Strong AI, because nothing artificially reproduced by man can share the history of a natural mind, and therefore AI cannot create minds simply because it cannot (re)create natural minds. On the other hand, though, the historical objection could be powerless with regard to Weak AI, as long as the functional profile of the mind did not depend on its history. Then, the possibility of artificially reproducing something that was functionally identical with the mind would be independent of the mind’s history. But is it coherent to distinguish between Strong AI and Weak AI? Such a distinction seems to hide an assumption about explanation, namely that a model can explain something, by simulating its functions, without being or becoming that thing. If one endorsed the above-mentioned view that the essence of the mind is its functions, then one would think that the only way in which a computational system could properly and completely explain the mind would be by deserving to be called a mind, and so the distinction between Strong AI and Weak AI would not hold. It would not hold because a simulation that did not deserve to be called a mind could, at best, only provide an approximation or metaphor for the mind. But, as a thesis about explanation, Weak AI would not be a coherent alternative to Strong AI. Richard Hamilton's article (Chapter Seven) proposes, similarly, that Weak AI is no alternative to Strong AI, but he rejects them both. Hamilton presents us with a reading of Heidegger according to which it is a mistake to think that the mind can be artificially reproduced. The reason that it is a mistake, according to Heidegger, is to do with what he calls “reflective self-understanding”. Man cannot help understanding itself in light of the world to which it belongs. And, ours being an essentially technological world, man cannot help understanding itself in technological terms; and, in the case of AI, it cannot help understanding the mind in computational terms. Because this reflective self-understanding imposes itself upon the human being, one cannot have any proper theory of the mind—a theory that could be independently addressed and chosen, as required for the assessment of any theory. A technological view of the mind imposes itself at a pre-intellectual level, and prevents us from doing philosophy of mind: thus, no AI thesis can be assessed for truth.

2 Content, perception and concepts 2.1 Mental states and intentionality We have been discussing the prospects of naturalising the mental. As noted above, the mental bears a peculiarly intimate relation to intentionality. Intentionality is, roughly, the property of being about or directed on something. The suggestion that intentionality is the “mark of the mental” was explicitly formulated by Brentano (1973; it is defended by Fodor 1987, 2003); although the

12


notion of intentionality, and the view that it has a fundamental role to play with respect to the mental, go back at least as far as scholastic metaphysics (e.g. Augustine’s On the Trinity; Augustine 1873). An intentional state (or event, or item) is a state (event, or item) with intentional content; it is in virtue of having this content that it is about something. Thus, the belief that the earth is flat is an intentional state whose content is the proposition that the earth is flat; thus it is about the earth (inter alia). It is not about the words “the earth”, or some representation of the earth; it concerns that very concrete thing, the planet earth itself. The wish that the earth be flat, and the supposition that it is flat, are states of different types with the same intentional content, and directed on the same object or state of affairs. A pain in the leg, on some views, also has an intentional content: that my leg is injured, perhaps. It is directed on one’s leg. That is why it is a pain in the leg, not the arm (though the view that pains are intentional is highly controversial; see below, §2.2). For a state to have intentional content is for it to determine a set of conditions under which it would correctly represent the world. Having intentional content does not require that these conditions are satisfied. It was an error of the scholastics and, on some readings, of Brentano, to suppose that the object of an intentional state must be internal (“immanent”) to the state, and therefore existent. This is suggested by the notorious phrase “intentional inexistence” (see Brentano 1874, p. 88); the idea is that, since intentionality is directedness, an intentional state embraces the object of this directedness, and so the mere occurrence of the state ensures that there is something on which it is directed.16 On the contrary, intentional content ought to be understood as a semantic notion, related to the notion of meaning. A state with intentional content represents something, in a sense roughly like that in which an expression in language represents something (which is not to say that an intentional state comes to represent something in the same way as linguistic items do). The possibility of misrepresentation ensures the following peculiar fact: semantic representation is such that what is represented need not exist. I might believe that the fountain of youth is around the corner, despite the fact that there is no fountain of youth. This modifies the sense in which intentionality is relational, for a state can be intentional even when there is no object or state of affairs to which it is related intentionally—as is the case with my belief about the fountain of youth. Brentano’s claim is that the realm of the mental is the realm of the intentional; the task of accommodating the mind within nature then becomes the task of naturalising intentionality. While this is an important step, it does not obviously 16

According to the direct realist view of perception, the object of a veridical perceptual experience is internal to the experiential state—it is constitutive of the state. Thus, this view denies that perception is intentional. See below, §2.3.


13

move us any closer to success, for the task of naturalising intentionality is itself a notorious philosophical quagmire. Before going on to discuss the plausibility of Brentano’s claim, we will set out very briefly the problems surrounding intentionality. These problems don’t depend on one’s acceptance of Brentano’s claim: even if intentionality is not the mark of the mental, it is still a crucial feature of many mental phenomena—one that does important explanatory work, and thus demands assimilation into our worldview. The first of these problems, which we can call “the problem of content”, is this: how can there be mental states with intentional content at all? The problem of content is given force by what we take to be involved in having a naturalistic worldview—and by the excellent reasons we have to adopt such a worldview. Mental states are natural states. The problem of content is to explain how a natural state—one composed of the brute materials of an apparently meaningless natural realm—could have the property of being directed on, of referring to, of concerning, some other element of the world. The difficulty of the problem of content lies in the fact that intentionality is a normative notion. A state’s having an intentional content places it squarely in what Sellars (1956) calls “the space of reasons”. Part of what individuates an intentional content is the rational relations in which it stands; being in a state (a belief, say) with a certain content makes it appropriate, to some extent, for one to enter certain states, and avoid certain other ones, individuated by their intentional contents. My belief that the earth is flat makes it pro tanto rational for me to believe, for example, that there is at least one flat planet, and puts pressure on me to avoid believing that the earth is round. A state’s having an intentional content is (in part) its standing in normative, rational relations to other such states, episodes and actions. But it is hard to see how the notion that something ought to occur, and the notion of something’s being a reason for something else, can get any grip in the context of naturalism. The problem of content is given concise expression by Haugeland (1990, pp. 384-5): “given the vapid materialism now generally conceded (roughly: without any matter, there wouldn’t be anything else contingent either), how can it be that any part or feature of the universe represents or is a reason for another? How can there be norms among the atoms in the void? How, in short, is intentionality compatible with materialism?”

What the problem of content amounts to depends on the answer to another question: what is it for a state, episode or item to have intentionality? This need not be a conceptual or definitional question; it need not be understood as requiring an analysis of the concept of intentionality. Rather, it can be construed as a question

14


about a putative natural kind: what are the intrinsic and relational properties in virtue of which something falls under the natural kind intentional? The question of what it is for a state (say) to be intentional breaks into two distinct but interdependent questions, corresponding to two ways of picking out the intentionality of the state. The first is the state’s property of representing anything—of having an intentional content. This feature is contrasted against the property of being blank, meaningless, or empty of content. The second feature is the state’s having the particular intentional content that it has. This is contrasted against the property of having some other intentional content. Thus, there arise the following related problems. What is it for state M to have an intentional content at all? And what is it for state M to have the content C, rather than the content C’? Providing an account of what it is to have an intentional content—in the general or particular sense—may make the problem of content easier. For example, if one holds that a state M has the content C by virtue of an evolutionary function linking states of that type to situations of the type specified in C (Dretske 1995; Millikan 1989), then the problem of content is answered by reference to evolutionary functions. On the other hand, one might hold that intentionality is an intrinsic feature of certain biological systems (Searle 1992); this seems to leave unaddressed the problem of content. A third route is to deny that there is any substantive philosophical problem (e.g. Rorty 1979): intentionality is not a natural property, nothing beyond our attributions constitutes M’s having the content C, and so the problem of content does not arise. We cannot here argue for a particular view of content, but only show how these problems bear upon the question of how intentionality and mind fit together.

2.2 The role of intentionality in the account of the nature of mind The prospects of solutions to these general problems determine how much progress we make in our overall task—understanding the mind as a natural phenomenon—by conceiving of the mental in terms of intentionality. Nevertheless, we can consider independently of these problems whether we ought to conceive of the mental in this way. That is, we can ask what the role of intentionality in the account of the mental should be. Someone uncorrupted by philosophy might be tempted to think that it is only once we have done our metaphysics, and explained how mental phenomena arise in a physical world, that we face the task of accounting for the fact that all, or many, phenomena of this type exhibit intentionality. On this view, the mental is a peculiar natural kind, perhaps associated with intentionality, but whose nature is explicable independently of intentionality. Brentano’s suggestion may be read as the denial of this view. Now, one can read Brentano’s suggestion in stronger or weaker ways (we leave aside here the question of how Brentano himself intended it). It could be read as a


15

mere claim about demarcation: all and only mental states exhibit intentionality of a certain sort. Read this way, the claim has no implications for explanatory priority. More strongly, Brentano could be read as making a constitutive claim about mentality: having intentional content is part of what constitutes a state’s being mental. Even more strongly, it could be read as a reductive thesis: the concept of the mental is reducible to the concept of intentionality. Other readings are possible: for example, the thesis could be read as proposing a metaphysical, rather than conceptual, reduction. There are indeed considerations that, prima facie, count in favour of accepting a strong version of Brentano’s claim: at least the constitutive version. The intentionality of a particular mental state seems to be an essential feature of that state. Take my belief that the earth is flat. It is difficult to make sense of the possibility of the very same belief having some other content. The belief that the sun is flat, and the belief that the earth is round, would be different beliefs. We individuate beliefs by their contents. And it is even more difficult to make sense of the very same state failing to have any content at all—failing to be a representational state. These points hold not only of beliefs, but of other intentional states: of all the propositional attitudes, of occurrent thoughts, of imaginings, and so on. Now, the belief’s being mental and its being intentional both seem to be essential features of that belief; and that fact does not entail that one of the features is constitutive of the other. But the essential role of intentionality implied by the above remarks entails that we cannot strip intentionality away from belief and be left with something recognisable as mental. There is thus some attraction in the idea that part of what it is for the belief to be a mental state is for it to have the normative, relational property of intentionality. It is not clear that the nature of belief as a mental state could be explained without reference to intentionality. Someone unsympathetic to the stronger versions of Brentano’s thesis might try to point to mental states or events that are not wholly characterised by their intentional contents; if there are such states or events, and if the mental is a natural kind, then the property of being mental cannot be constituted by the property of being intentional. For example, the qualia of certain experiences17 have been argued to be essentially independent of those experiences’ intentional contents. On some views, pains are an example of mental phenomena that involve more than intentionality (e.g. Block 1996; but see Tye 1997, for a rejection of this view). What matters to pains, one might think, is not primarily a semantic content, but their phenomenology: it is essential to pains that they feel a certain way, and the way they feel (they hurt) is not fully characterised by specifying an intentional content such as “my leg is injured” (compare feeling a pain in one’s leg to merely 17

See above, §1.2; and Ward (this volume).

16


coming to believe of one’s anaesthetised leg that it is injured). Another oft-cited example is orgasm: although orgasm is perhaps partly constituted by a representation of ejaculation occurring, the experience seems to have sensational properties that go beyond any such representation. If this is right, then we cannot use intentionality to establish the unity of the mental as a kind. Pains and orgasms seem to have phenomenology that outruns their intentionality; on some views, they lack intentionality entirely. On the other hand, propositional attitudes have intentionality but, on most views, lack phenomenology: believing that the earth is flat does not seem to have a subjective phenomenal feel that distinguishes it from believing that the earth is round (though Searle 1992 can be read as claiming that propositional attitudes have phenomenology; and see Campbell’s chapter, this volume, for remarks on Hume’s “belief-feeling”; see, e.g., Dennett 1991, for the rejection of this). What, if anything, might tie these two sorts of phenomena together?

2.3 Perception: intentional content and phenomenology The task for the defender of the strongest version of Brentano’s view is to reduce phenomenology to, or identify it with, intentionality: they must show that the phenomenal character of a mental state or episode is nothing more than its having this or that intentional content. Perhaps this can be done for pains or orgasms (see, e.g., Tye 1997); or perhaps pains and orgasms comprise an isolated category of phenomena that cannot be understood in the same way as other mental states and episodes. The same cannot be said, however, for perceptual experiences—the exemplary case of the interplay of intentionality and phenomenology, since such experiences have phenomenal character and intentional content. Perception is the locus of any number of problems in metaphysics, epistemology and the theory of intentional content. Whatever we say about intentionality and phenomenology in this case is likely to have profound philosophical import. Unlike pains, it is clear that perceptual experiences are world-directed: they represent, or present, the world as being a certain way to their subjects. They have content, in this undemanding sense at least (we do not mean to imply that it is obvious that perceptual experiences have intentional content in the semantic sense). On the other hand, perceptual experiences, unlike beliefs, are uncontroversially and fundamentally phenomenological. There is something it is like to enjoy a particular conscious perceptual experience, and the nature of the experience cannot be understood independently of this phenomenological aspect. There might be something it is like to entertain a certain belief, but it was suggested above that what is essential for understanding a belief is its content, rather than any phenomenology it might have.


17

The subjective character of perception strongly suggests some sort of dependence, or interdependence, between these two features. To put it roughly, in experience the world is presented to one in the phenomenology. Consider a conscious experience of seeing red. Such an experience is a representation, or presentation, of redness—it is directed on a property in the world—only in so far as it has the peculiar phenomenal character that we associate with visual experiences of red things.18 The representational aspect does not seem to go beyond the phenomenal aspect. When one makes judgments (such as the judgment “that is red”) on the basis of such experiences, it seems that it is having an experience with that phenomenology that justifies one in making the judgment. And yet, on the other hand, the phenomenology is essentially world-directed. The phenomenal character of seeing red does not seem to be anything more than the redness in the world revealing itself to one. Phenomenology and intentionality seem, in such cases, to be inseparable; they are not independent features of experience such that we can abstract from one in order to see the other. What, then, is the role of intentionality in perceptual experience? According to the intentional theory of perception (or intentionalism), there is no more to the character of a perceptual experience than its intentionality. That is, the phenomenology of experience—its feeling as it does to its subject—is a representational feature; phenomenal properties are identical to, or reducible to, representational properties (c.f. Byrne 2001, for a recent statement and defence of intentionalism; also Harman 1990; Tye 1995) The intentional theory, confronted with the considerations adduced in the foregoing paragraphs, opts for a reductionist version of Brentano’s thesis, at least as it applies to perceptual states. It claims that the nature of perceptual experience should be explained entirely in terms of the notion of intentionality. The intentional theory of perception understands “intentionality” in the somewhat demanding semantic sense according to which the intentional content of an experience is understood in terms of the conditions under which the experience would be veridical. The view that perceptual experiences are intentional in this sense commits one to more than the uncontroversial claim that experiences are world-directed. Thus, a substantive intentional theory of perception must be distinguished from the mere claim that the subjective character of a perceptual experience can be fully described in terms of its being directed on certain objects or state of affairs. This latter claim, though entailed by most versions of the substantive intentional theory (Peacocke 1983 is an exception), is not equivalent to it. It is the claim that perceptual experiences lack non-representational qualia. 18

Perhaps there are unconscious or subconscious perceptual representations that have no phenomenal character for the subject. These remarks are intended only to capture what our experience seems like to us, introspectively.

18


In its adherence to this demanding sense of intentionality, the intentional theory assimilates perception to beliefs and desires. This suggests that we should understand perceptual experiences as occurrent states or episodes that represent the environment as being a certain way, and are caused (when veridical) by its being that way. This conception of perception is challenged by the so-called enactive view, which conceives of perceptual experience as being a process of interaction between perceiver and environment, in which the perceiver explores the environment, rather than something which is simply caused by the environment. According to the enactive view, your experience’s being of a tree is not simply a matter of the content “tree” being tokened in a mental state which you passively enjoy; rather, it is a matter of how the tree affects your sensory receptors, and how you expect it would do so, over an extended period, in which you might explore the tree in various ways, for example by walking around it (c.f. Noë 2004, for an influential version of the view). Now, it is a matter of some controversy just what the enactive view of perception is committed to. For it is not clear that intentionalism must deny, for example, that perception is a temporally extended process involving interaction with objects. Julian Kiverstein’s contribution to this volume (Chapter Eleven) presents an understanding of the enactive view, on which it has specific and substantive disagreements with intentionalism. According to Kiverstein, the intentional content of an experience is fixed by the subject’s sensorimotor knowledge: her grasp of how a scene’s phenomenal appearance to her is contingent on her movements. This entails that the phenomenal properties of conscious experiences not only are independent of their representational properties, but explain those experiences having the intentional contents they have. Pace intentionalism, intentionality derives from, rather than being the source of, phenomenology. The emergence of new views of perception, such as the enactive view, has stimulated reflection on the difficulties facing intentionalism—once installed as a near-orthodoxy. Consideration of these difficulties will help to put into context some other competing accounts. Broadly speaking, concern has focused on whether the intentional theory can do justice to the manifest role of perception in our cognitive, rational lives. We can distinguish two sorts of considerations that have led to such concern. The first consideration is the need to respect the phenomenology of perception. As we noted already, perception is distinct from the propositional attitudes in being associated with a salient phenomenology. The intentional theory purports to bring perception within our understanding by assimilating it in important respects to propositional attitudes: it holds that enjoying a certain sort of perceptual experience is just entertaining a certain intentional content—albeit in a distinctively perceptual way. This would suggest that phenomenology is to be explained by this distinctive


19

way in which contents are entertained in perception. The resultant worry, then, is twofold. On the one hand, intentionalism owes us a plausible account of this distinctive way that will explain why perceptual experiences have phenomenology—and it is not clear that such an account is in prospect. On the other hand, some philosophers claim that our introspective evidence tells us that the phenomenology of perception is not merely a matter of entertaining a certain content in a certain way or mode: the phenomenology tells us that in experience we grasp objects, or their sensible properties, in a distinctively sensory way that is inconsistent with intentionality in the semantic sense. This second worry is expressed by Mike Martin with the thought that experiences are partly constituted by their objects: “the phenomenological character of all perceptual experience requires us to view the transparency and immediacy of perceptual experience as involving actual relations between the subject and the objects of perception and their features. In just the case of veridical perception, the experience is a matter of certain objects being presented as just so, and in virtue of that, the subject ought to conform their beliefs to how things appear.” (Martin 2002, p. 402.)

Why is this inconsistent with the intentional theory? That theory is committed to a distinction between the content of a perceptual experience, and the object on which the experience is directed. One influential way to understand this distinction is in terms of sense and reference: on this understanding, intentional content is a phenomenon at the level of sense, while the objects of experience are referents. The intentional theory avoids what it perceives as the error of the scholastics19 by holding that the content of an experience, rather than the object, is a constituent of the experience. The object is external to the experience; and so the very same experience could have occurred in the absence of the object. The worry is that phenomenology tells us otherwise: objects of experience are immediate in the sense of being constituents of experience. The objects of experience are not merely represented (as they are, for example, by concepts); they are directly presented in experience. The second sort of consideration against the intentional theory is epistemological. It is a manifest fact that veridical perception is a source of knowledge. Furthermore, it is a source of knowledge about particular objects: in virtue of experiencing a tree, I can come to know about that very tree. My thoughts latch directly onto that particular; I do not need to be able to pick it out by, say, giving a unique description of it. This requires the tree itself to enter into my experience in some robust sense. But, it is claimed, this is inconsistent with intentionalism, for intentionalism claims that the object itself is not a constituent of 19

See above, §2.1.

20


the experience, but extrinsic to it. The flavour of the objection is given by McDowell (1994a, p. 193): “since there is no rationally satisfactory route from experiences, conceived as, in general, less than encounters with objects, glimpses of objective reality, to the epistemic position we are manifestly in, experiences must be intrinsically encounters with objects. But how could they be that if their aetiology were phenomenologically extrinsic?”

Note that this whole objection could be recast in non-epistemic terms. Perception enables us to think about particular objects (never mind knowing about them). We must, therefore, have experiences that involve those objects in all their particularity. The worry, then, would be that intentional content cannot achieve this particularity. It is not that states with intentional contents cannot refer to particulars; of course one can have a belief about a particular tree. It is rather that this particularity must have its source in something other than intentional content; the particularity of the content of that belief must be based on an experience that involves that very tree.20 Considerations along these lines have led some philosophers to conclude that the semantic conception of perceptual content is on entirely the wrong track. Thus we have seen renewed interest in direct realist accounts of perception. Direct realism used to be contrasted with indirect realism: the view that in perceptual experience one’s awareness of the external, mind-independent object of experience is mediated by awareness of mind-dependent sense-data (c.f. Russell 1912; Price 1950). But intentionalism is certainly not an indirect realist theory. It holds that one is aware of the object of one’s experience in virtue of one’s experience having a certain content, but not that one achieves awareness of the object by being aware of the content; just as one can think of an object by employing a certain concept, without thereby thinking of the concept. So direct realism, if it is to be in competition with intentionalism, must be saying more than that perceptual awareness of objects is unmediated. The crucial further claim is that a perceptual experience is a state of phenomenological awareness to which mind-independent objects are internal: one’s experiencing a particular object entails that that object is before one. One could not have had that experience (an experience of that type) in the absence of that object,21 because its presence is partly constitutive of the

20

But see, e.g., Burge (1991), Peacocke (1993), Soteriou (2000), for the view that intentional content in perception can be particular-object-involving in a robust enough sense to meet this objection. 21 This is not to say that one could not have had an indistinguishable experience in the absence of the object. Rather, any experience which one could have had in the absence of the


21

experience. Experiences are fundamentally relations to objects. This contradicts intentionalism as it is usually understood, since entertaining an intentional content is not a relational state in a sense that requires the presence at that moment of the object on which the state is directed (see §2.1 above: it is essential to intentionality that an intentional state can be directed on nothing). This is true even on views according to which the intentional content of experience is partially objectdependent:22 object-dependent contents entail the existence of their objects, but not their presence every time the content is entertained. It is a central feature of intentional states that they are fallible. A state’s having a certain intentional content cannot guarantee that the state correctly represents the world.23 What’s more, the correctness or incorrectness of the state’s content makes no difference to how it is for the subject to be in that state. The fallibility of our intentional states—the fact that we can represent the world, but be mistaken about it—is one source of many of the philosophical difficulties we are discussing. It follows from this fallibility that there is no entailment from a subject’s enjoying a state with a particular intentional content, to that state’s representing the world correctly. If perception is intentional, then one’s enjoying a perceptual experience with a certain content is something that could occur equally in circumstances that do or do not make it veridical. Therefore, for any veridical perceptual experience, there is a non-veridical experience that, to the subject, would be indiscriminable from the veridical one (assuming that sameness of content is sufficient for indiscriminability, or, equivalently, that discriminability entails distinctness of content; below, we will make use of the converse entailment). That is, nothing available to the subject of an experience guarantees that the experience is not illusory. This implication of the intentional theory is surely a plausible one. The phenomenology of an experience cannot itself guarantee that the experience is not misleading. Even putting aside the intentional view, it seems to be part of the fallibility of perception that veridical and illusory experiences can be subjectively indiscriminable. That is why we can be fooled by the illusory ones. Direct realism now comes under pressure. The pressure is usually applied with some version of the argument from illusion. Typically, the argument from illusion is presented as an argument in favour of the sense-data theory of perception (see Smith 2002, for a thorough statement of the argument). However, we can rehearse a refined version of the argument, as follows (strictly, this version is a kind of argument from hallucination). object would have been a different type of experience. See below, this section; and Pereira (this volume). 22 See above, note 20. 23 See Brewer (1999) for an argument against this claim.

22


Take any veridical perceptual experience E, directed on some object. By fallibility, there is a possible perceptual experience, E’, that is subjectively indiscriminable from E, but is non-veridical. There is no object on which E’ is directed. If two perceptual states are subjectively indiscriminable, then they are tokens of the same type of state. If two mental states are of the same type, then it is not the case that one has a type of constituent that the other lacks. Since the putative external object on which E’ is directed does not exist, E’ does not have that object as a constituent. Therefore, E does not have an external object as a constituent either. Therefore direct realism is false. One reaction to this argument is sense-data theory. The defender of sense-data theory accepts the argument from illusion, but refuses to give up the claim that the objects of experience are constituents of the experience. The sense-data theorist argues that, since the constituents of the experience must be the same in the veridical and non-veridical cases, the objects of experience must be internal, minddependent objects: sense-data. Sense-data theory can be part of an indirect realist view of external objects (e.g. Russell 1912); or, if it is adopted on epistemic grounds, it may lead to phenomenalism (e.g. Ayer 1940), motivated by the idea that whatever our experience enables us to have thoughts about must be something intrinsic to experience (sense-data, on this view). More recently, the most popular reaction to the argument from illusion has been to deny that the objects of experience are constituents of the experience. This leads to intentionalism. Both sense-data theory and intentionalism endorse the premise that subjectively indiscriminable perceptual states are of the same metaphysical type. Intentionalism does so because of two assumptions. The first is the assumption that indiscriminability entails sameness of content—the converse of the entailment we appealed to above, and a plausible commitment for anyone who thinks that the character of experience is to be explained entirely in intentional terms, and who holds that differences between the contents of experiences can be detected by the experiences’ subjects. The second assumption is the widely-accepted principle that intentional-state types are individuated by their contents. The denial of the premise that subjective indiscriminability entails sameness of type defines the view known as disjunctivism. The disjunctivist holds on to the direct realist’s intuition that the objects of (veridical) experience are constituents of experience, on either phenomenological (Martin 2002) or epistemological (McDowell 1982, 1994a) grounds. But the disjunctivist resists the argument from illusion, and its pull towards sense-data theory, by claiming that there is no fundamental metaphysical type common to veridical experiences, illusions and hallucinations:24 subjective 24

Disjunctivists agree that veridical experiences and hallucinations are of different types; they differ on where illusions—experiences that latch on to actual objects in the


23

indiscriminability is not sufficient for sameness of type. Non-veridical experiences lack certain constituents that their matching veridical experiences have; their indiscriminability is to be explained some other (psychological) way. The attraction of disjunctivism, then, is that, unlike intentionalism, it respects what seems right about direct realism—the object-involvingness of perception—while, unlike the sense-data theory, it avoids commitment to a metaphysically and epistemologically puzzling veil of perception between subjects and the world. What is the conflict between intentionalism and disjunctivism? Intentionalism is committed to the claim that E and E’ are of the same type. Metaphysical disjunctivism is precisely the denial of this claim. So it might seem, but Francisco Pereira, in his chapter (Chapter Nine), argues for the surprising conclusion that metaphysical disjunctivism is consistent with intentionalism—and, indeed, can be seen as a version of the intentional theory on which intentional content is objectinvolving. Part of the interest of Pereira’s paper is in its careful consideration of a question over which we have ridden somewhat roughshod in this piece: just what are the commitments of the intentional theory? According to Pereira, intentionalism is minimally committed to there being a type of state in common between E and E’, but not to their being of the same fundamental type. This is, so far, consistent with Martin’s (2004) influential version of disjunctivism, which holds that E and E’ both fall under the type subjectively indiscriminable from E, but that E, unlike E’, is of the fundamental type veridical experience of an x, where x is whatever E is an experience of. Intentionalism also holds that E’, like E, can be described intentionally, in terms of how it represents the world as being. Pereira differs from Martin in holding that this commitment is also consistent with disjunctivism. If Pereira is successful in showing that intentionalism and disjunctivism are compatible, then he will have offered a new way of accommodating our intuitions about perception. We have so far seen how experience, in virtue of its phenomenology, deserves an independent treatment within the philosophy of mind, and how that is the case independently of one’s stand on intentionality. In the next two sections we will consider the other unique element of experience: its epistemic priority.

2.4 The metaphysics and epistemology of perception It is no coincidence that we have found this interlacing of metaphysics, epistemology and the theory of content, when we came to the philosophical consideration of perception. Perceptual experiences are mental episodes or states; the metaphysics of these episodes or states must be intelligible as part of the wider environment, but present them as having properties they do not have—should fall within that dichotomy of types. C.f. Martin (2002).

24


account of the nature of the mental. Yet perceptual experiences are special in their epistemic role. They are a fundamental epistemic source for a vast range of our beliefs about the world. Perception is even more basic when it comes to issues of content: although there are concepts whose application is not tied to experience, it is arguable that no thought at all would be possible—no contents could be entertained—without the enabling contribution of perception. Thus, Kant’s Critique of Pure Reason begins: “There can be no doubt that all our knowledge begins with experience. For how should our faculty of knowledge be awakened into action did not objects affecting our senses partly of themselves produce representations, partly arouse the activity of our understanding to compare these representations, and, by combining and separating them, work up the raw material of the sensible impressions into that knowledge of objects which is entitled experience? In the order of time, therefore, we have no knowledge antecedent to experience, and with experience all our knowledge begins. But though all our knowledge begins with experience, it does not follow that it all arises out of experience” (Kant 1929, in Norman Kemp Smith’s translation).

Although the different metaphysical accounts of perception we have been considering can be contrasted in their epistemological commitments, any such account comes up against a very general epistemological task: explaining how perceptual states can yield knowledge. My perceptual experience might make it appear to me as if there is a tree in front of me; but why, on that basis, am I warranted in believing that there is a tree in front of me? Philosophers have found this fact—that perceptual appearances yield warrants for beliefs—puzzling, because of a number of apparently plausible points about perception—points that do not rely for their plausibility on any particular philosophical view of perceptual experience. Firstly, there are phenomenological and epistemological reasons to think that, in acquiring knowledge on the basis of experience, one does not rely epistemically on any mental states other than the perceptual experience. One does not, for example, rely on a belief that one’s experience is veridical. The experience alone suffices, in the absence of defeaters, to warrant the belief that, say, there is a tree in front of one. That is not to say that other mental states do not constitute enabling conditions for perceptual knowledge; it is just that they do not constitute epistemic grounds for such knowledge. Phenomenologically, we seem to believe what we see automatically, rather than going through, or relying implicitly on, any sort of reasoning. Epistemologically, we do not seem to have any grounds independent of experience for believing that our experience is veridical; thus any such belief would fail to provide epistemic support (see Alston 1993, for the case that there is no noncircular argument for the reliability of perception).


25

Secondly, as we have noted, perception is fallible. I could have been presented with a tree-appearance indistinguishable from the veridical case, and yet been mistaken in judging there to be a tree in front of me. There is nothing in the experience itself to rule out this possibility. It should be noted that this is not an exotic evil-demon-type scenario: our perceptual experiences lead us astray relatively often, and yet we are usually entitled to rely on them. The third point is more delicate. On some views, the way we take the world to be on the basis of perception is underdetermined by what is strictly presented to us in perception. That is, what the external environment itself contributes to an experience is less than the richly textured, objectively structured scene the subject finds himself inhabiting. Now, most contemporary philosophers would be uncomfortable with the idea that there is an experiential “given” that somehow falls short of being a presentation of objective reality (see Sellars 1956; McDowell 1994b)—the idea that there is a level of content that is a mere “blooming, buzzing confusion” (as William James described the experience of neonates; James 1890, p. 462), from which a representation of the objective physical world must be built up. Nevertheless, it seems to be true (see below, §2.5) that the information reaching the sensory receptors at any moment is relatively sparse. At a subpersonal level, one’s perceptual systems implement a certain conception of how the environment is arranged, in order to cause a personal-level experience that has richness, order and objectivity, rather than one that is a mere confusion. In that sense, in experience one imposes something beyond what is given; this is quite consistent with the sort of thought that motivates the direct realist, that in experience one is open to the environment. But then, how can an experience whose content goes beyond what is given by the environment warrant one in making a judgment which is solely about that environment? Considerations like these have given rise to a tradition in philosophy of investigating the contribution of the perceiver to the construction from perception of a conception of an objective world. This investigation has taken different forms. Kant held that a conception of objects interacting causally in space and time is something derived a priori from the constitution of our perceptual and intellectual faculties, which structures the deliverances of the world. Few contemporary philosophers endorse the Kantian project wholesale, but there remains a strong tradition, particularly in British philosophy, of demanding an epistemic contribution to empirical knowledge from the ways in which we must think about the deliverances of perception if we are to acquire knowledge from perception at all (c.f. Strawson 1959; Evans 1982; Campbell 1994; McDowell 1994b). Kant’s project is not that of refuting scepticism, about perceptual or any other beliefs. However, we can see Hume’s grappling with various sceptical threats as contributing to the same philosophical tradition of investigating the role of perceivers’ faculties in perceptual epistemology. In the context of scepticism, the

26


first two considerations we brought forward become salient: the fallibility of perceptual experience, and the fact that one can apparently acquire knowledge on the basis of an experience without further ado—in particular, without any independent basis for supposing that the experience is veridical. Hume’s contribution is, characteristically, more prosaic than Kant’s; he is concerned not with the a priori and the constitution of our perceptual and intellectual faculties, but with the habits of thought in virtue of which we are able to respond to experience as the presentation of an independent world, inhabited by persons and structured by causal interactions between objects, even when we apparently lack sufficient grounds for taking it to be such. For Hume, perception can suffice to ground world-concerning belief, despite its fallibility and relative sparseness, in virtue of certain other things we take for granted. Among these things is the reliability of our perception (see Alston 1993). Now, taking for granted the reliability of perception is not the same as having a belief that perception is reliable, and using that belief as a reason to take one’s experience at face value. Rather, in basing beliefs on experience without further ado, one implicitly manifests that one is taking it for granted; rather as one can perform a modus ponens inference, without having a belief in the validity of modus ponens that one could marshal in support of one’s conclusion (as Lewis Carroll 1895, famously pointed out, it is a mistake to see the validity of modus ponens as a further reason, in addition to the premises, for accepting the conclusion of such an inference). On some views (see Peacocke 2004), these propositions we take for granted in basing beliefs on experience are presuppositions: the beliefs are warranted as long as there is no reason to think the presuppositions false. Our entitlement to these presuppositions is rationally explicable: there is a rational basis for acting as if they are true. For Hume, however, the status of these propositions is not grounded in reason; they are where justification runs out. The orthodox interpretation of Hume has it that they are propositions we assent to by habit of thought, and that our entitlement to do so rests on pragmatic considerations. In her contribution, Treasa Campbell (Chapter Three) challenges this interpretation. She argues that Hume saw these “natural instincts” as something much more fundamental: not as a set of beliefs among others, but as propensities that constitute our belief-forming structure. They are not foundational beliefs, but rather principles that govern our whole beliefsystem at once. They are grounded in the nature of human engagement with the world. Given that they have this status, Campbell argues, natural instincts can be seen as establishing naturalised epistemic norms. It is a controversial question whether Hume should be seen as denying the possibility of an epistemic grounding for propositions such as that of the reliability of perception. Campbell’s interpretation sees Hume as doing something more subtle, and something that anticipates Kant: locating the ground of these


27

propositions not in reasons we have to think them true, but in the constitution of our nature as human beings living in the world. If Campbell is right, philosophy’s most renowned sceptic may in fact have provided the materials for an anti-sceptical case.

2.5 How does experience ground thought? If perception gives knowledge of, and grounds thought about, the world, it is only because a large majority of our perceptual states are veridical. Perception does the fundamental jobs in virtue of which it is of such philosophical interest, not just by representing the world, but by actually putting us in touch with it. Thus the epistemological questions we have been discussing presuppose that there are veridical perceptual states: that there are states of, say, seeing a tree, that can serve as a basis for empirical thought and knowledge. To see a tree is not merely to have it appear as if there is a tree in front of one; it is to be perceptually linked to a particular tree. It is only states of this sort that enable one to know that there is a tree in front of one, to think about that tree, and so on. It is therefore crucial that we have a satisfactory account of veridical experience—of the circumstances under which having an experience as of a tree count as seeing a tree. The difficulty is that it will not do to say that an experience as of a tree counts as seeing a tree just when the truth-conditions that fix its content are satisfied. For one could be standing in front of a tree, and have an experience as of a tree matching that one in every visible way, yet not be seeing the tree (it could, for example, have a hologram of an exactly matching tree obscuring it). It is natural to suppose that the concept of (veridical) seeing (or hearing, etc.) finds application when there is a causal dependence of the right sort of the experience on the object experienced. Thus, I am seeing the tree, not when the scene in front of me merely happens to be how it appears to me to be, but when things appearing to me that way is caused by the scene being that way. It must be caused “in the right way”, where that clause is spelled out in such a way as to exclude cases where the causal chain between the scene and the experience is so indirect that the case doesn’t count as one of seeing (suppose, for example, that the presence of the tree in front of me causes the projection of the matching hologram, and so indirectly causes my experience). Nick Jones, in his article (Chapter Ten), challenges one influential account of what this causal dependence must amount to: David Lewis’s counterfactual account. Lewis (1980) takes the relevant dependence relation to be one that holds between the experience as a whole and the scene as a whole. He proposes that the dependence of the experience on the scene must be such that, had the scene been different, the experience would have been matchingly different. If this condition fails, the experience is non-veridical. Jones presents empirical evidence of change

28


blindness: the failure to notice significant changes in parts of a scene to which one is not presently attending (c.f. Rensink et al. 1997). According to Jones, since the possibility of change blindness is ever-present, all cases of seeing are such that counterfactual dependence fails to hold: were the scene to change in any one of many ways, the perceiver’s experience would not change. Therefore counterfactual dependence of the experience on the scene is not necessary for seeing. Jones’s chapter can be taken as a contribution to a growing movement within philosophy and psychology that rejects the view of seeing as a matter of building up a representation of a whole scene that matches and depends on the scene (c.f. Noë 2004; Clark 2002). It claims that we can see particular details in a scene, by being appropriately responsive to them, while not seeing other details. Alternatively, one could understand Jones as pointing to a phenomenon that has long been noted by scientists of perception: the pervasive influence on our experience of general assumptions made by our perceptual systems (see above, §2.4). The visual system, for example, works as if the appearances of stable objects do not change suddenly; it “assumes” they remain the same. Thus we experience scenes as if they conformed to these assumptions. These are not a priori constitutive facts about our faculties in Kant’s sense, or about our human engagement with the world in Hume’s sense. Nevertheless, they can be seen as subpersonal analogues of the natural instincts discussed by Campbell. Thus a new version arises of the question with which we began the previous section: how can we be entitled to rely on the deliverances of a perceptual system that we know to be systematically biased?

2.6 Content and concepts We have been discussing perception primarily from an epistemological point of view. It is arguable, however, that the question of how perception can yield knowledge of the world is a special case of a more general question: how does perception put us in touch with the world at all? (see McDowell 1994b). That is, what role does perception play in enabling us to think of mind-independent objects? The answer to this question must connect perception to the possession and application of concepts, for we think of mind-independent objects by employing concepts of those objects. Concepts are constituents of thoughts; the concepts involved in a thought, and their mode of combination, constitute the content of the thought. When I think that the sun is shining, I represent a certain state of affairs, by combining the concept the sun with the concept of a role that the sun can play in local weather conditions. Our question, then, is how the application of a concept can be enabled by, and grounded in, experience. Conceiving of matters at this level makes apt another famous remark of Kant’s (1929, B75/A51), which has been


29

guiding us implicitly until now: “Thoughts without content are empty, intuitions without concepts are blind.” Now, it is not clear that this more general question about content can really be understood apart from epistemic considerations. Presumably, part of what constitutes experience’s enabling thought to be non-empty is the contribution of specific experiences to specific warrants for applications of concepts. That is, my experience enables me to think about trees in part by contributing to knowledgeable applications of the concept tree. This link between concepts and knowledge— between the semantic and the normative—will be endorsed by anyone who thinks that the nature of a concept cannot be explicated independently of its role in knowledge (e.g. Peacocke 1992). Gloria Ayob, in her contribution to the volume (Chapter Four), takes up the question of how perceptual experience can warrant the use of concepts. Ayob’s paper makes abundantly clear how intimately related these issues are to the metaphysics of the physical world; the realist about ordinary objects will want an account that explains how perception enables correct applications of concepts to be fixed by the mind-independent natures of the objects presented in experience. Such an account is the ambition of John Campbell (2002), who holds that a sort of preconceptual attention to an object can warrant the use of a demonstrative concept to refer to the object, thus enabling subjects to grasp such concepts. That is how demonstrative concepts ground our capacity to think of mind-independent objects. Ayob argues, on Wittgensteinian grounds, that no pre-conceptual acquaintance with objects can serve this purpose: grasping a demonstrative requires a richer level of thought, in which intentionality is already implicated. If Ayob is right, then the problem of content is not to be solved by focusing on the constitution of our perceptual faculties, as John Campbell supposes, but rather by focusing on the nature of social practices. It is only within the practices of a community that intentional content can emerge. Lurking in the background, when these issues are discussed, is the embarrassing lack of consensus in philosophy on what it means to attribute possession of a concept to somebody. It is clear that possessing the concept tree enables one to think about trees—to reason about them, to refer to them, and, as we have been emphasising, to know about them—in ways not available to those who lack the concept. But this seems to tell us what possessing the concept enables one to do, rather than telling us what possession of the concept consists in. Earlier we characterised the problem of content as one of the fundamental problems underlying much of philosophy of mind; since concepts are the primary vehicle for the representation of the world by human minds, the question of what concepts are is surely as monumental and far-reaching as any in philosophy. One influential line of thought in philosophy and cognitive science takes it that concepts are mental representations: they are inner mental particulars with semantic

30


properties. This line of thought is most famously defended by Jerry Fodor (1987, 2003). According to Fodor, our mental states and episodes come to represent the world in virtue of the properties of these particulars, and their combination. Concepts are constituents of those states and episodes; their semantic properties and syntactic combination thus determine the content of the state or episode in question. Roughly, for one to think that the sun is shining is for certain of these particulars, namely those representing the sun, and the property that the sun has when it is shining, to be tokened in a certain combination. To possess the concept of the sun is to have as a syntactic unit in one’s language of thought a particular that represents the sun. The notion that concepts are representational mental objects is in some ways an attractive one. It gives an intuitively intelligible picture of what it is to think a thought, or have a belief, and of how such episodes and states can have intentional contents. However, the representationalist view has come under increasing criticism in recent years. Hans-Johann Glock, in his contribution “Concepts: Representations or Abilities?” (Chapter Two), subjects the Fodorian view of concepts to a far-reaching critique. According to Glock, it misconstrues the nature of concepts—in particular, it cannot be reconciled with the fact that concepts can be shared—to suppose that they are objects or goings-on inside the minds of individuals who possess them. Nor, however, can concepts happily be thought of as abilities: Glock argues that this is a category mistake. The lesson, Glock claims, is that we should not be looking for an answer to the question of what concepts are; we should focus instead on the question of what it is to possess a concept. He concludes by suggesting that concept-possession should be construed as an ability. Glock limits himself to the methodological claim that we are likely to make more progress by focusing on the question of concept-possession than by focusing on the question of what concepts are. However, there might be a more profound point here. It seems plausible that the notion of a concept derives its sense from the notions of possession and application of a concept. This is a perfectly familiar scenario, though one that philosophers should take more notice of. As Glock points out, the question “what is a mortgage?” is secondary to the question “when does a house have a mortgage on it?”, in the sense that mortgages are nothing over and above the conditions under which houses are mortgaged. Similarly, perhaps, with concepts. If we discard the view of concepts as representational mental items, it becomes appealing to take the question “under what conditions does a subject possess a concept?” to be prior to the question “what are concepts?”. Some other notions of philosophical interest may also benefit from such a treatment. So, for example, we might think the question “what is a reason?” is secondary to the question “what is it to have a reason?”. Perhaps many of philosophy’s “what is X” questions should be addressed this way.


31

The most influential recent theory of concepts, that of Christopher Peacocke (1992), begins with precisely that assumption: that the best approach to understanding concepts is through an understanding of what it is to possess a concept. Peacocke individuates concepts in terms of the conditions under which thinkers can be said to possess them. The possession-conditions for a concept specify certain belief-involving transitions, inferential and otherwise, involving the concept. The transitions that individuate a concept have the status of primitive axioms: to grasp the concept, a subject must find the transitions “primitively compelling”. It is crucial that the transitions involve belief, for belief aims at truth, and, for Peacocke, the nature of concepts is not explicable independently of truth, and, concomitantly, reference. Asunción Álvarez’s contribution to the volume (Chapter Five) challenges Peacocke’s theory of concepts. Álvarez’s criticisms fall into two sets. The first set of criticisms has to do with Peacocke’s insistence on a referential constraint on genuine concepts: the constraint is that the possession conditions of any genuine concept must, together with the world, fix a reference for the concept. Álvarez finds this constraint too demanding, and presents a series of cases of apparently genuine concepts whose possession conditions are internally inconsistent, in the sense that they could not determine a reference. What’s more, she argues that the constraint is unmotivated by Peacocke’s framework. Secondly, Alvarez finds Peacocke’s explanation of the rationality of accepting primitive axioms unsatisfactory. Peacocke has introduced the notion of “implicit conceptions” to do this job; these are non-inferential, subpersonal attitudes that ground the understanding that is constituted by the personal-level acceptance of primitive axioms. Alvarez argues that implicit conceptions are both too strong and too weak to do the job. They are too strong in that they seem to do away with the need for a philosophical theory of concepts. They are too weak in two respects: firstly, in that they leave an unbridgeable, or at least unbridged, gap between, on the one hand, the implicit and non-inferential, and, on the other hand, the explicit and inferential; and secondly, in that they offer no explanation of what we have been calling “the problem of content”—of how meaning emerges from the meaningless.

Bibliography Alston, W.P. (1993). The Reliability of Sense Perception. London: Cornell University Press. Augustine, St., Bishop of Hippo (1873). On the Trinity (trans. A.W. Haddan). Edinburgh: T&T Clark. Ayer, A.J. (1940). The Foundations of Empirical Knowledge. London: Macmillan. Block, N. (1995). "On a Confusion about a Function of Consciousness", in Behavioral and Brain Sciences 18: 227-287.

32


——— (1996). "Mental Paint and Mental Latex," in E. Villenueva (ed.), Philosophical Issues 7. Northridge: Ridgeview. Brentano, F. (1973). Psychology from an Empirical Standpoint (trans. A.C. Rancurello, D.B. Terrell, and L.L. McAlister). London: Routledge and Kegan Paul. Brewer, B. (1999). Perception and Reason. Oxford: Clarendon. Burge, T. (1991). "Vision and Intentional Content", in E. LePore and R. Van Gulick (eds.), John Searle and his Critics. Oxford: Blackwell. Byrne, A. (2001). "Intentionalism Defended", in Philosophical Review 110: 49-90. Campbell, J. (1994). Past, Space, and Self. Cambridge, Mass.: MIT Press. ——— (2002). Reference and Consciousness. Oxford: Oxford University Press. Carroll, L. (1895). "What the Tortoise said to Achilles", in Mind 4: 278-280. Chalmers, D. (1996). The Conscious Mind. Oxford: Oxford University Press. Clark, A. (2001). “Reasons, Robots and the Extended Mind”, in Mind and Language 16: 121-145. Clark, A. (2002). “Is Seeing All It Seems? Action, Reason and the Grand Illusion”, in Journal of Consciousness Studies 9: 181-202. Davidson, D. (1970/2000). "Mental Events", in his Essays on Actions and Events (2nd ed.). Oxford: Oxford University Press. Dennett, D.C. (1991). Consciousness Explained. London: Allen Lane. Di Nucci, E. Mind Out of Action. PhD Dissertation, University of Edinburgh. Dretske, F. (1995). Naturalising the Mind. Cambridge, Mass.: MIT Press. Eilan, N. (1995). "The First Person Perspective", in Proceedings of the Aristotelian Society 95: 51-66. Evans, G. (1982). The Varieties of Reference. Oxford: Clarendon. Fodor, J.A. (1987). Psychosemantics. Cambridge, Mass.: MIT Press. ——— (2003). Hume Variations. Oxford: Oxford University Press. Harman, G. (1990). "The Intrinsic Quality of Experience", in Philosophical Perspectives 4: 31-52. Haugeland, J. (1990). "The Intentionality All-Stars", in J.E. Tomberlin (ed.), Philosophical Perspectives 4: Action Theory and Philosophy of Mind. Northridge: Ridgeview. Hume, D. (1777). "Of Suicide", in his Essays and Treatises on Several Subjects: In Two Volumes. Edinburgh: Donaldson and Creech. Hurley, S.L., and Noë, A. (2003). "Neural Plasticity and Consciousness", in Biology and Philosophy 18: 131-168. Jackson, F. (1982). “Epiphenomenal Qualia”, in Philosophical Quarterly 32: 127136. ——— (1986). “What Mary Didn't Know”, in Journal of Philosophy 83: 291-295. ——— (2003). "Mind and Illusion", in A. O'Hear (ed.), Minds and Persons. Cambridge: Cambridge University Press.


33

James, W. (1890). The Principles of Psychology. London: Macmillan. Kant, I. (1929). Critique of Pure Reason (trans. N. Kemp Smith). London: Macmillan. Kim, J. (1966). "On the Psycho-Physical Identity Theory", in American Philosophical Quarterly 3: 277-85. Lewis, D. K. (1980). “Veridical Hallucination and Prosthetic Vision”, in Australasian Journal of Philosophy 58: 239-49. ——— (1988). "What Experience Teaches", in Proceedings of the Russellian Society 13: 29-57. Martin, M.G.F. (2002). "The Transparency of Experience", in Mind and Language 17: 376-425. ——— (2004). "The Limits of Self-Awareness", in Philosophical Studies 120: 3789. McDowell, J. (1982). “Criteria, Defeasibility and Knowledge”, in Proceedings of the British Academy 68: 455-79. ——— (1994a). "The Content of Perceptual Experience", in Philosophical Quarterly 44: 190-205. ——— (1994b). Mind and World. Cambridge, Mass.: Harvard University Press. Millikan, R.G. (1989). "Biosemantics", in The Journal of Philosophy 86:281-297. Nagel, T. (1974). “What is it Like to be a Bat?”, in Philosophical Review 83: 435456. Noë, A. (2004). Action in Perception. Cambridge, Mass.: MIT Press. Peacocke, C. (1983). Sense and Content. Oxford: Oxford University Press. ——— (1992). A Study of Concepts. Cambridge, Mass.: MIT Press. ——— (1993). "Externalist Explanation", in Proceedings of the Aristotelian Society 93: 203-230. ——— (2004). The Realm of Reason. Oxford: Clarendon. Pettit, P. (2003). "Looks as Powers", in Philosophical Issues 13: 221-252. Price, H.H. (1950). Perception. London: Methuen. Putnam, H. (1967). "Psychological Predicates", in W.H. Capitan and D.D. Merrill (eds.), Art, Mind, and Religion. Pittsburgh: University of Pittsburgh Press. Rensink, R. A., O’Regan, J. K., and Clark, J. J. (1997). “To see or not to see: the need for attention to perceive changes in scenes”, in Psychological Science 8: 368-373. Robinson, H. (2003). "Dualism", in S. Stich and T. Warfield (eds.), The Blackwell Guide to Philosophy of Mind. Oxford: Blackwell. Rorty, R. (1979). Philosophy and the Mirror of Nature. Princeton: Princeton University Press. Russell, B. (1912). The Problems of Philosophy. London: Thornton Butterworth. Searle, J.R. (1983). Intentionality: An Essay in the Philosophy of Mind. Cambridge: Cambridge University Press.

34


——— (1992). The Rediscovery of the Mind. Cambridge, Mass. MIT Press. Sellars, W. (1956). "Empiricism and the Philosophy of Mind", in H. Feigl and M. Scriven (eds.), Minnesota Studies in the Philosophy of Science, Vol. 1. Minneapolis: University of Minnesota Press. Smart, J.J.C. (1959). "Sensations and Brain Processes", in Philosophical Review, 68: 141-156. Smith, A.D. (2002). The Problem of Perception. Cambridge, Mass.: Harvard University Press. Soteriou, M. (2000). “The Particularity of Visual Perception”, in The European Journal of Philosophy 8: 173-189. Strawson, P.F. (1959). Individuals: An Essay in Descriptive Metaphysics. London: Methuen. Tye, M. (1995). Ten Problems of Consciousness: A Representational Theory of the Phenomenal Mind. Cambridge, Mass.: MIT Press. ——— (1997). "A Representational Theory of Pains and their Phenomenal Character", in N. Block, O. Flanagan and G. Güzeldere (eds.), The Nature of Consciousness: Philosophical Debates. Cambridge, Mass.: MIT Press Wittgenstein, L. (1953). Philosophical Investigations. Oxford: Blackwell.

PART I: CONCEPTS AND MENTAL CONTENT

CHAPTER TWO CONCEPTS: REPRESENTATIONS OR ABILITIES?

HANS-JOHANN GLOCK

Concepts have occupied a central role in philosophy ever since the Socratic quest for definitions. But even if we abstract from merely terminological variations, concepts have been assigned a multitude of different and potentially conflicting roles. What we would nowadays call concepts have featured as: • components of propositions • universals or properties • meaning(s) of linguistic expressions • psychic or neurophysiological processes • “modes of representation”. In spite of this variety, one can detect a pervasive contrast between two fundamentally opposed conceptions of concepts (see also Haller 1971; Kenny 1989, p. 136; Rey 1998). On the one hand, we find objectivist or logical conceptions, positions which treat concepts as parts of what we think, components of propositions, thoughts or “contents”, and hence as phenomena that exist independently of the minds of individuals, paradigmatically in a realm of abstract objects. On the other hand, we find subjectivist or psychological conceptions, accounts which treat concepts as parts of our thinking, as mental episodes (events, processes, mechanisms) which feature in the stream of inner goings on. At present, this conflict manifests itself in at least three different contexts. One is a diffuse yet wide-ranging divergence of interest between logicians, conceptual analysts and formal semanticists on the one hand, psychologists, philosophers of mind and members of the artificial intelligentsia on the other. The former tend to study concepts in abstraction from individual minds (whether it be as sets of real or possible objects, functions on such entities, or as abstractions from intersubjective linguistic practices). The latter incline to treating them as internal representations such as images, stereotypes, or words realized in the mind or brain. At a more


37

specific level, this divergence is linked to debates in the wake of externalist theories of mind. It has become customary to distinguish between the wide and the narrow content of propositional attitudes, the former being determined by the causal interaction between a subject and her physical or social environment, the latter by what goes on in her mind. The third manifestation is Fodor’s vociferous attack on objectivism. Fodor has recently added a sustained subjectivist theory of concepts to his representational theory of mind or “RTM” (1998; 2003). Its central claim is that concepts are mental particulars. Fodor portrays his enemies as part of a wide conspiracy, which includes Ryle, Wittgenstein, neo-Fregeans, conceptual or inferential role semanticists, and holists of various sorts. He labels this movement “pragmatism”, because it links concepts to inferential rules and abilities rather than inner goings-on. And he has pronounced pragmatism to be “the defining catastrophe of analytic philosophy of language and philosophy of mind in the second half of the twentieth century” (quoted in Crane 2004, p. 4).1 Before having a closer look at Fodor’s Manichean struggle, we need to distinguish at least four philosophical questions that can be raised about concepts: Definition question: What are concepts? Individuation question: How are concepts individuated? Possession question: What is it to have a concept? There has been a tendency to run together problems of definition and problems of individuation in the extensive discussion of criteria of identity. Yet it is one thing to ask what (an) F is, another thing to ask how Fs are to be individuated. For instance, it is one thing to define what a person is—e.g. a creature that is self-conscious, rational, autonomous, and has a sense of her own past and future; it is quite another to specify how persons are individuated—e.g. by reference to bodily continuity, psychological continuity, possession of a brain, etc.. Put differently, criteria of identity must not be confused with criteria of application (see Williamson 1990). The latter concern the conditions under which something is an F to begin with, the former the conditions under which an F encountered in one context or under one description is the same as an F encountered in a different context or under a 2 different description. If Quine’s slogan “No entity without identity!” is on the right 1

See also Fodor (2003, p. 9; and 2004), where he uses the less misleading label “concept pragmatism” to cover all “epistemic” accounts that treat concept possession as a kind of knowledge, including those which invoke knowing that rather than knowing how. But even in that article he does nothing to dispel the impression that RTM and pragmatism exclude one another. This is misleading, since one can be a representationalist while subscribing to conceptual role semantics (e.g. Block, Boghossian). 2 Even with respect to a sortal noun “S”, which is by definition linked to criteria of identity, these criteria do not constitute a definition of “S”. That the use of “S” is subject to certain criteria of identity forms part of its meaning, but doesn’t exhaust it. For although some

38


track (see Strawson 1997, ch. 1; Glock 2003, pp. 47-52), a general term is well defined only if there are criteria of identity for things to which it applies. But although definition and individuation are linked in various ways, they are not the same. Once we distinguish these three questions, one further question arises: Priority question: Which is more fundamental, the definition question or the possession question? In what follows, I shall discuss the conflict between representationalism and pragmatism with a view to these four questions. Each numbered section argues for one of the following claims. Definition question 1. Concepts are not computational-cum-neural representations or signs, but rather what is expressed or represented by signs. 2. Patterns of neural firings cannot even be signs of concepts. 3. Concepts are not abilities. Priority question 4. The possession question is a better starting-point for an investigation of concepts than the definition question. Possession question 5. Concept-possession is a dispositional rather than occurrent property. 6. There is no circularity in explaining concept-possession by reference to abilities. Individuation question 7. Concepts are not more finely individuated than abilities.

1 Are concepts representations? For RTM, “the mind is pre-eminently the locus of mental representation and mental causation” (Fodor 2003, p. 8; see pp. 7-10). It is a “representational system”, and thinking is “representing things in the world” (Crane 2004). So-called propositional attitudes like belief and desire are mental states, and they “are constituted by relations to mental representations”. Furthermore, “mental processes sortals differ in their respective criteria of identity, not all of them do. “Cat” and “dog”, for example, have different definitions; yet they share the criteria of identity for the more general sortal “animal”, and there may even be a single criterion of identity for all material bodies (Lowe 1997, p. 623).


39

consist of causal interactions among these interrelated states and entities”; they are “causal chains in which each link is sufficient for its successor”. “The constituents of mental processes are causal interactions among … ideas” (2003, pp. 8n, 141, 10). As regards concepts, the central thesis of RTM is that they are “mental objects”. They constitute a kind of “mental representation” and hence a “kind of mental particular” (Fodor 1998, pp. 3, 7, 22). As mental particulars, they are “objects in the mind” of individuals, and they have causes and effects in the physical world (Crane 2004; Fodor 2003, p. 13n). As mental representations, they have “representational content”. They occur as part of propositional attitudes, and what they represent contributes to the content of these attitudes, to what we believe, desire, etc. They do so by determining the conditions under which the attitudes in which they occur are true (in the case of beliefs) or satisfied (in the case of desires). Unlike the “Theory of Ideas” one finds in Descartes and the British empiricists, RTM conceives of representations not as denizens of a separate mental realm, but as physical phenomena and, more specifically, as (patterns of) neurophysiological processes or events. Instead of ideas, it speaks, for example, of “cerebral neuronal representations” (Libet 1987). Similarly for Fodor, representations in general and concepts in particular are, in the final analysis, patterns of neural firings. This avoids the metaphysical embarrassment of having to postulate a realm over and above the physical world of matter, energy and tangible objects. But it also has a more immediate advantage. Mentalist representationalism holds that entertaining thoughts involves the occurrence of introspectable items, words or mental images that cross our minds and of which the subject is or can easily be aware. But as Frege, Wittgenstein, Ryle and Price argued convincingly, such occurrences are neither necessary nor sufficient for conceptual thought, and they do not determine its content. RTM is immune to these objections. This fact is concealed by the curious readiness of its proponents to sign up to the traditional view that entertaining a concept is “a matter of contemplating or viewing an ‘object’—a mental atom” (e.g. Fodor 2003, pp. 11-12). Yet this is not a position RTM can condone. Its representations, the particulars with which it identifies concepts, are neither “objects before the mind”, nor even “objects in the mind”, if the latter is understood as anything other than a neurophysiologically realized computer-programme. For they are physical phenomena in the head. For Fodor, concepts are ultimately patterns of neural firings or discharges. To be sure, the representations of his computational representationalism are elements of a computational architecture; they are classified in computational rather than neurophysiological terms, as symbols of a software programme. But these symbols are implemented neurophysiologically, and the particulars with which Fodor identifies concepts are neurophysiological phenomena. In any event, these computational representations

40


occur at a subpersonal level which is no more accessible to introspection or 3 authoritative self-awareness than neurophysiological phenomena. That inspectable items in the stream of consciousness fail to determine the content of conceptual thought is not a worry for either neural or computational representationalism. But another problem remains. It arises from an observation which was already central to Frege’s case against subjectivist conceptions of concepts (e.g. 1892, p. 60). Concepts can be shared between different individuals, whereas mental representations, considered as particulars in either the mind or the brain, cannot. 4 Shareability is a feature of the ordinary concept of a concept in both everyday life and in disciplines like psychology and the history of ideas. One central use we make of “concept” and terms that are equivalent in the relevant contexts (like “conception”, “idea” or “notion”) is in claims about different individuals or even groups of individuals either sharing a concept, or failing to do so. “Well, I share their concept of lying, if nothing else!” was recently my exasperated last line in a conversation about Tory accusations against Tony Blair. Claims of this kind are equally central to intellectual history, e.g. when it comes to comparing the Greek concept of eudaimonia with the contemporary concept of happiness. Such appeals to established use outside cognitive science would normally be given short shrift by supporters of RTM. Yet even Fodor accepts as one of the “non-negotiable conditions on a theory of concepts” that “concepts are public; they’re the sorts of things people can, and do, share” (1998, pp. 22, 28, 34). At the same time, some of them breezily dismiss the suggestion that shareability shows concepts to be abstracta rather than mental representations. Laurence and Margolis opine that the Fregean argument “isn’t the least bit convincing”: “Mental representations are subjective in that their tokens are uniquely possessed; they belong to one and only one subject. Their being subjective in this sense, however, doesn’t preclude their being shareable in the relevant sense, since, …, two people can have the same representation by each having tokens of the same type. When someone says that two people have the same concept, there is no need to 3

Contemporary RTM subscribes to Price’s admonition against the “inspective theory of thinking”: “The concept is not before the mind as an object of inspection. It is at work in the mind, but not as one inspectable content among others, … It shows itself not as a detectable item of mental furniture, but rather as a guiding force, determining the direction which the series of presented particulars [mental images or words] takes, …” (1953, p. 342; see also pp. 313, 332-4). 4 This term is preferable to Fodor’s “publicity”: my nose is public in the sense of being accessible to more than one observer; but I am relieved to state that I do not shared it with anyone.


41

suppose that she is saying that they both possess the same token concept. It would make as much sense to say that two people cannot utter the same sentence because they cannot both produce the same token sentence” (1999, p. 7).

This response applies to concepts the distinction between type and token which Peirce had devised for signs. Alas, it then proceeds to run the two together. The objectivist argument is precisely that concepts, qua shareable, could not be subjective in the sense in which tokens are. Types of mental representations are shareable, alright; but of course they are abstracta, just like the objectivist has it. Fodor also tries to account for the shareability of concepts by distinguishing between type and token. Unlike Laurence and Margolis, however, he acknowledges that types are abstract entities rather than particulars. the mental particular that’s in your head on occasions when you think dog is a token of the concept type DOG, just as the word that’s on your lips when you say “dog” is a token of the word type “dog”. In both cases, the tokens are concrete particulars and the types are abstracta. Likewise, the mental particular that’s in your head when you think that (judge that) dogs bark is a token of the mental representation type DOGS BARK (2003, p. 13n; see 1998, p. 3n).

What can be shared between different individuals are abstract concept-types, of which the unshareable mental particulars in each individual are tokens. Fodor’s explanation of shareability ties in nicely with his specific version of RTM, namely his “language of thought hypothesis”. Fodor (1975) argued that both the meaning of public languages and the intentionality of thought can be explained by a “language of thought”. External sentences are meaningful because they are correlated with internal symbols, sentence-like representations in the brain the tokening of which constitutes our thinking. The representations which have genuine propositions as their contents are sentences in “Mentalese”, physical tokens of computational types. When I engage in conceptual thought, Mother Nature inscribes words of a computer programme into my brain. Nevertheless, this solution of the shareability problem is inadequate for two reasons. For one thing, even if Fodor’s account does not trade on a straightforward equivocation, it sails under false pretences. On the one hand, the non-negotiable constraint on a theory of concepts, namely of shareability: ¾ is satisfied only by the abstract concept types, ¾ which are not particulars (mental or otherwise). On the other hand, the central claim of Fodor’s theory of concepts, namely that concepts are particulars: ¾ applies only to the concept tokens, ¾ which are not shareable.

42


This difficulty is not merely terminological. If Fodor persistently kept apart concept-types and concept-tokens, he would be forced to abandon the central claim of representationalism, the one which is supposed to set it apart from the wicked pragmatists, namely that concepts are mental particulars that can enter into causal relations. Indeed, my objection can be sharpened by reformulating it as follows. Fodor’s constraint leads to an argument the conclusion of which is the very opposite of what he claims: P1 P2 C1

Concepts must conform to the non-negotiable constraints on a theory of concepts. Mental particulars cannot conform to one of these constraints. Concepts are not mental particulars.

RTM no longer constitutes a genuinely subjectivist position. Its case against objectivism has evaporated, and Fodor’s case against pragmatism would have to be rejigged as follows: We entertain conceptual thoughts not by exercising an ability, but through the occurrence of neural tokenings of shareable concept-types.

Both Anne and Sarah believe that Socrates is human. According to Fodor, in Anne’s brain there occurs one neural token-sentence, and in Sarah’s brain there occurs a different neural token-sentence. Yet Anne and Sarah both believe the same, since both tokens instantiate the same Mentalese type-sentence SOCRATES IS HUMAN. Similarly, they share the concept HUMAN, because both “have tokens of literally the same concept type” (1998, p. 28). Even if Fodor owned up to the consequence that shareable concepts are not particulars, his position would still suffer from terminological incoherence. “Very roughly, concepts are constituents of mental states. Thus, for example, believing that cats are animals is a paradigmatic mental state, and the concept ANIMAL is a constituent of the belief that cats are animals… [M]ental states and processes are typically species of relations to mental representations, of which latter concepts are typically the parts” (1998, p. 6).

On the one hand, concepts are supposed to be constituents of mental states and processes, which in turn are declared to be relations to mental representations. On the other hand concepts are supposed to be parts of the mental representations themselves.


43

The latter is Fodor’s considered position: “concepts are the constituents of 5 thoughts” (1998, p. 25), not of states or processes of thinking thoughts. But that position revolves around a more systematic and insidious oscillation between subjective and objective categories, in particular the realm of signs and the realm of what signs signify. Fodor continues: “I’ll use “thoughts” as my cover term for the mental representations which, according to RTMs, express the propositions that are the objects of propositional attitudes. Thus, a belief that it will rain and a hope that it will rain share a thought, as well as a proposition which that thought expresses. For present purposes, it will do to think of thoughts as mental representations analogous to closed sentences, and concepts as mental representations analogous to the corresponding open ones” (1998, p. 25).

Anne and Sarah do not stand directly in a relation of believing to the proposition that Socrates is human. Rather, they stand in a relation to tokens of the thought or type-sentence SOCRATES IS HUMAN; and that thought expresses the proposition, it “means that” Socrates is human (see Fodor 1987, p. 17). By the same token, the concept HUMAN is a type-word which has the property being human as its meaning. At the same time, Fodor also maintains that “word meanings just are concepts” (1998, p. 2). This cannot allude to the relation between token-concepts and typeconcepts. For sign-types are not the contents or meanings of sign-tokens. For instance, “human” does not mean or express the sign-type “human”; it instantiates that type. What could mean the sign-type are signs like “‘human’” or “the word ‘human’”. The idea that concepts are word meanings is blatantly incompatible with treating them as words that have meanings. It suggests instead that they are components of the contents of propositional attitudes, parts of what we think 6 (believe, desire, etc.). Nor can this passage be dismissed as a slip. It is central to 5

Unlike Fodor, we should distinguish carefully between A’s belief in the sense of what A believes, e.g. that the sun is out, and in the sense of what A has, namely a thought-episode which might be characterized as a believing that the sun is out. 6 Fodor, like many contemporaries, equates meaning and content, e.g. when he describes believing that p as standing in a computational relation to a mental representation that “means that” p. In my view, this equation is untenable. A proposition, propositional content or thought is something that is or could be said or believed—a sayable or thinkable. However, propositions in the sense of what is or could be said are what we express or convey by the use of sentences, not the meanings of these sentences. Unlike the meaning of a sentence, what is said (believed, etc.) can be true or false, implausible or exaggerated (see Glock 2003, pp. 153-4). Furthermore, far from being identical with sentence-meanings, what is said on a particular occasion depends on sentence-meaning and context of utterance. But in the present context I shall disregard this need to distinguish meaning and content.

44


Fodor’s case for the shareability of concepts that they should be components of the contents of propositional attitudes. His theory of concepts is supposed to explain the possibility of psychological explanations which are both intentional and “nomic”, i.e. they invoke general law-like connections between propositional attitudes. Such explanations “appeal to covering generalizations about people who believe that such-and-such, or people who desire that so-and-so, or people who intend that this and that, and so on. In consequence, the extent to which an RTM can achieve generality in the explanations it proposes depends on the extent to which mental contents are supposed to be shared. If everybody else’s concept WATER is different from mine, then it is literally true that only I have ever wanted a drink of water, and that the intentional generalization “Thirsty people seek water” applies only to me” (1998, p. 29; see p. 7).

Fodor rightly rejects that suggestion, and proceeds to insist that belief contents and their component concepts are literally shared, rather than merely similar. It is not just that Fodor’s specific argument for shareability requires that concepts be parts of what we think rather than of representations, of mental sentences which express what we think. On the alternative assumption there is simply no reason to accept that concepts are shareable. As Peacocke and Rey have pointed out, even mental representation types need no more be shared between individuals that entertain the same thought than the type words need to be shared in sentences of public languages that express the same proposition. “One person might express the concept [city] by the word ‘city’, another by the world ‘ville’; still another by a mental image of bustling boulevards; but, for all that, they might have the same concept [city]; one could believe and another doubt that cities are healthy places to live” (Rey 1994, p. 186; see Peacocke 1992, p. 3).

Laurence and Margolis try to rebut this argument on two grounds. First, they deny that different types of mental representations can express the same concept, since they “will have substantially different inferential roles”. Secondly, even if different types of internal representations could express the same concept this would raise “no difficulties for the view that concepts are mental representations”: “If two or more different representations of different types express the same concept, then, of course, that concept cannot be identified with one or the other of these two types. But there is no reason why the concept shouldn’t be identified with a broader more encompassing type—one that has the mental representations of these other two types among its tokens. Just as particular Persian cats can be cats alongside Siamese cats and tabbies, so tokens of different types of representation can all be instances of a broader representation type” (1999, p. 76).


45 7

This response betokens a misunderstanding of the type/token distinction. Persian and Siamese cats both instantiate the kind cat or the property of being a cat. But there is no symbol or representation-type of which both “city” and “ville” are tokens. Of course, they have something in common. Yet it has nothing to do with the syntactic computational level at which type-words in Mentalese are supposed to be individuated. It lies rather at the semantic level: both mean the same, they represent the same kind of entity. This also guarantees that they do not differ in their inferential roles, pace Laurence and Margolis. More generally, there is no reason why two representation-types that play different roles in the thinking of an individual should not express the same concept. If you associate the term “grandmother” with an image of your grandmother and I associate it with the first 8 line of Brahm’s lullaby, this is no bar to our expressing the same concept by it.

2 Are concepts signified by computational/neural signs? On any coherent account that preserves the connection between concepts and shared thoughts, a concept is neither a token- nor a type-representation; rather, it is what is expressed by such representations. Concepts are not representations, but what is represented, just as the objectivist tradition has it. In a Teutonic vein, you might call this “Frege’s Revenge”. In an American vein, you might say: “Any way you cut up the pie, concepts ain’t in the head!”. It is central to RTM that “concepts are symbols” (Fodor 1998, p. 28). Yet even if patterns of neural firing could be symbolic representations (see below), concepts could not themselves be such representations, since they are among the kinds of things which representations represent or express. It is not easy to see how representationalism can escape this difficulty, short of dismissing the shareability constraint. Even then, a problem would remain. Representations require a medium. That is to say, they have representational properties by virtue of having non-representational properties. For instance, Rembrandt’s self-portrait in the National Gallery of Scotland represents a particular

7

This misunderstanding may have been facilitated by the more pervasive failure to keep apart that distinction from the one between properties and tropes. The type/token distinction is apposite in connection with Fodor’s idea of a language of thought, since the latter postulates signs of a certain kind. But the more general talk of types and tokens of mental states is misleading: what is at issue is the property of being in a mental state (e.g. believing that dogs bark), and the realization of that property in a particular individual at a particular time. 8 For a more thorough elaboration of these criticisms of RTM, see Glock (forthcoming).

46


individual on account of more basic properties, roughly the way in which it arranges colours and shapes (see Hyman 2006). The idea that thoughts and concepts are (mental, computational or neural) representations is incompatible with this defining feature of representations. Thoughts and concepts are all message and no medium. This just reinforces the lesson we already reached above by a different route. Thoughts and concepts are what is represented, the content of thinking, not what represents, i.e. what expresses these contents. But can’t we still preserve the idea that thinking conceptual thoughts is a matter of having representations? After all, the type-representations of computational representationalism occur in a medium, namely a medium of neural firings. They represent propositional or conceptual contents on account of their nonrepresentational physical properties. According to Fodor, for instance, these representations have certain syntactic properties—properties determining the way they are processed—on account of their physiological qualities, and they have certain semantic properties, properties determining what they represent, on account of their causal relations to the environment. The suggestion that neural firings are representations needs to be spelled out, however. And at this juncture the feature of RTM which allowed it to escape the aforementioned problems of mentalist representationalism come to haunt it. Because of their subpersonal status, the purported representations transcend introspection and self-awareness. But this makes it impossible to accommodate the epistemic dimension of representation. The non-representational properties of representations must be accessible to the subject of representation. After all, a representation is not just a sign of something, but a sign for someone—a subject of representation. Yet neural tokens of computational types are entirely inaccessible to the subject. This point is displayed clearly in Peirce’s famous semiotic triangle, and can be rephrased in terms of his theory of signs. One minimal condition on a thing R being a representation of another thing X is that R should convey information about X. But, following Peirce’s theory of signs (1933; see also Schwartz 1994), one can distinguish between different types of representation: Icons resemble what they represent Symbols are related to what they represent by convention Indices are connected to what they “represent” by causal dependencies or by other natural relations such as spatial or temporal proximity. Thus realist paintings are icons of what they represent. Linguistic expressions, with the possible exception of onomatopoetic ones, are symbols of what they represent. The word “dog”, for example, is connected to the animals not through any kind of resemblance, but through an arbitrary convention. Smoke, finally, is an index of fire, because it is a causal result of fire.


47

Given this distinction, one might deliver the following brief verdict on causal representationalism: a. Patterns of neural firings are certainly indices of external phenomena, but only for observers with neurophysiological measuring equipment, not for ordinary subjects of thought; b. they might be icons (but for the most part are not); c. they cannot be symbols. That neural firings are causal results of external events and causal preconditions of perception is agreed on all sides. The extent to which there is, for example, a spatial resemblance between the objects of perception and the neural activities that underlie perception, is a matter for empirical investigation. For the most part, no such iconic relation has been observed. Although experiments like those of Hubel and Wiesel show that particular neurons are involved in seeing lines of a particular orientation, there is no iconic similarity between the lines and the pattern of firing neurons. Finally, neural firings cannot be symbols because there is no one who uses them to represent anything in a conventional way. In the sense in which a symbol represents something, the firing of a neuron represents nothing, for we do not use neural cells or neural firings as symbols, and it makes no sense to suppose that the brain uses them as a symbol of anything. For something to be a symbol, it must have a rule-governed use, there must be a correct and an incorrect way of employing it (see Hacker 1987). Neural firings are not symbols, they do not have definitions or rules of syntax, and they are not employed by symbol-using creatures. They cannot be correct or incorrect, but only be or fail to be in line with observed regularities. To say that a neural firing represents an oriented line in the visual field is simply to say that there is a causal correlation between the latter and the former. But the relation between, for example, the noun “cat” and the animals it might be said to represent or stand for is conventional, not causal. Accordingly, the claim that conceptual thought involves neural representations is - false, as far as icons are concerned; - nonsensical, as far as symbols are concerned; - vacuous, as far as indices are concerned. For that last claim boils down to the indisputable but unexciting observation that neural firings in the brain are causally related in law-like ways to certain events outside the brain. Although indices are signs of some kind, they are not representations in the sense of icons and symbols. They are not the result of an intention to convey information about X, they simply do convey such information to someone cognizant of their causal correlation with X. As Peirce himself stressed, a purely natural relation like that between fire and smoke has no significance at all, unless someone takes the smoke to indicate a fire.

48


Proponents of RTM might respond by insisting that their notion of a symbol differs from the Peircean one. For them what counts is not a conventional connection between the symbol and what it signifies, but the fact that it has certain syntactic properties, properties which determine the role which the symbol plays in computational processes. But it is unclear why talk of syntax should have any more content in the absence of sign-using creatures than talk of semantics. By definition, syntax concerns the combinatorial properties of signs. But without creatures to whom signs represent something there can be no syntax, and neither can there be genuinely cognitive operation with signs. Of course, thinking and speaking involve causal mechanisms which operate at a subpersonal level. But these mechanisms provide the causal basis for the possession and operation of concepts (the understanding and use of words); they must be distinguished both from the concepts themselves and from their possession.

3 Are concepts abilities? I hope to have provided reasons for holding that RTM—whether in its mentalistic, physicalistic or computational manifestation—does not constitute the correct answer to the definition question. At this point we are entitled to turn our back on representationalism. But do pragmatist accounts of concepts fare any better? The most straightforward version of such an account identifies concepts with abilities. In response to his question “are concepts entities or are they dispositions?”, Price states in no uncertain terms: “a concept is not an entity, …, but a disposition or capacity” (1953, pp. 320, 348; see also p. 355). In the same vein, Geach pronounces that concepts “are capacities exercised in acts of judgment” (1957, p. 7; see p. 13). Geach points out that this identification of concepts and capacities does not fall foul of the shareability constraint—unlike the idea that they are mental particulars (or so I have argued). It does not entail that “it is improper to speak of two people as ‘having the same concept’”, since different individuals can possess the same mental capacities (1957, p. 14). Furthermore, concepts and abilities can both be acquired, applied and lost. Finally, we shall see in the next section that to possess a concept is to possess an ability. However, we cannot simply infer from this last point that concepts themselves are abilities. For our linguistic constructions need not comply with the naïve compositionalist recipe “(having x = having y) ⇒ (x = y)”, since having may amount to something different in each case. Furthermore, there are weighty objections to identifying concepts with dispositions, capacities or abilities. First, one crucial thing we do with concepts is to define or explain them. But to define a concept is not to define or explain a capacity. Secondly, concepts can be instantiated or satisfied by things. But abilities cannot. Thirdly, and relatedly,


49

concepts have an extension and an intension, which cannot be said of abilities. Insofar as the ability linked to possessing the concept F has an extension, it is not the range of things that are F, but either the range of subjects that possess F, or the range of situations in which these possessors can apply or withhold F. Fourthly, a concept can occur in a proposition or statement, but an ability cannot. Of course, abilities can occur in propositions in the sense of being mentioned in them. But concepts occur in propositions in yet another and more pervasive way. The concept of being human occurs in “Socrates is human”, even though there is no mention of a concept. The most straightforward pragmatist definition of concepts fails, therefore. Concepts are not identical with abilities, even if to possess a concept is to possess an ability. A more promising suggestion is that concepts are not themselves abilities, but the principles or rules which someone has mastered who possesses the relevant abilities. For instance, in so far as to have a concept is to be able to classify things (see sections 6-7 below), the concept itself is not the ability to classify, but the principle or rule according to which someone who possesses the ability can classify. Even this proposal creates categorical mismatches, however. It does not seem to be the case that to define a concept is to define a principle or rule. Rather, the principle features in the definition. Furthermore, at least prima facie principles or rules can occur in propositions only in the immediate sense of being mentioned, the sense that abilities are confined to. There may be a solution to this problem. If Strawson and Künne (2003, ch. 6.2) are to be trusted, universals or concepts can enter a proposition not just in the sense that the sentence expressing the proposition contains words or phrases referring to the universal/concept, but also in the more indirect sense that the sentence contains words or phrases expressing the concept. Why not extend this courtesy to principles or rules? One qualm would be that principles can be true or false, whereas concepts cannot. It is not obvious that stipulated principles of classification really are truthapt, syntactical appearances notwithstanding; but the onus here would be on a pragmatist pursuing this line. On the other hand, rules seem to escape this difficulty, since they are expressed by sentences in the imperative rather indicative mood. Note, however, that this uncontentious exemption holds only for regulative rules, not for constitutive ones. And it would seem that the kind of rules which constitute word-meaning, and by implication concepts, would have to be constitutive rather than regulative (see Glock 2005). In that case, invoking rules will pose the same kind of challenge for pragmatism as invoking principles.

4 Concepts or concept-possession? Whatever concepts are, they cannot be identified with either abilities or with representations in individual minds or brains. Furthermore, it may even be difficult

50


to identify them with principles or rules. At this point, I hope that you have become slightly weary of the definition question. It’s time to consider the suggestion that it is more auspicious to scrutinize concept-possession, rather than concepts themselves. We are dealing with the conflict between a Direct (Socratic) approach: What are concepts? What is F? and an Indirect (contextualist) approach: What is it to have a concept? What is it to know F? Fodor favours a direct or Socratic approach to the nature of concepts. One of his central gripes against pragmatism is that it reverses the “classical direction of analysis”. Instead of starting out with an explanation of “what it is to be the concept X, … the concept’s identity conditions”, they start out with an account of “the conditions for having concept X”. Fodor concedes that it is a grammatical truism that accounts of concepts and of concept-possession must be interconvertible: “if you know what an X is you also know what it is to have an X. And ditto the other way around”. Still, he insists that “an explanation of concept possession should be parasitic on an explanation of concept individuation”, just as an account of “knowing the meaning of a word” should be parasitic on an account of “being the meaning of a word”. For there is a general priority of “metaphysical” over “epistemological” issues: “… understanding what a thing is, is invariably prior to understanding how we know what it is” (1998, pp. 2-3, 5; 2003, p. 11). Appearances notwithstanding, however, there are good reasons for reversing these priorities. Socratic “What is…?” questions have a notorious tendency to shed more darkness than light, and to result in puzzlement and mental cramps. What is more, this tendency seems to be unaffected by the switch from the material mode, i.e. questions of the form “What is the essence or nature of F”, to the formal mode, i.e. questions of the form “What does “F” mean?”. At the same time, the switch from a direct to an indirect approach does promise a way forward. Consider the following pairs of questions: What is a mortgage? When does a house have a mortgage on it? What is a legal obligation? When is someone legally obliged to do something? What is a disease? When does someone suffer from a disease? In all these cases, the contextualist questions are more tractable than the Socratic ones. Presumably the reason is that the nouns concerned derive their content from non-nominal constructions. But I do not intend to advance a claim of either


51

conceptual or metaphysical priority here. Instead, I shall confine myself to a methodological case. It is easier to make progress on the topic of concepts if one starts out from concept-possession. Ironically, Fodor should be sympathetic to this line. After all, he opines hyperbolically, though not without some justice, that “the number of concepts whose analyses have thus far been determined continues to hover stubbornly around none” (2003, p. 6; see also 1998, chs. 3-4; and Davidson 1996, pp. 263-4). In view of this situation, it is surely both legitimate and natural to pursue an indirect approach, since it promises to avoid the deadlocks of a direct one. In fact, one such indirect approach, owed to Frege, is precisely to approach the question of what Fs are by elucidating the criteria for the identity of Fs. Thus the most helpful way of explaining what a direction is may be to elucidate under what conditions two lines have the same direction. And the most propitious way of explaining what a cardinal number is may be to scrutinize the conditions under which two sets have the same number of members. But now, the allegedly egregious strategy of inquiring into what F is by elucidating what it is to know (an) F is simply another form of such an indirect approach. What is more, there is no reason to suppose that this variant is any less legitimate. Barring sceptical doubts, which in sober philosophy should be confined to a special branch of epistemology, we have every right to assume that we know humdrum things like concepts or the meanings of words. In fact, one of the few things that must stand fast in reflections on linguistic meaning is that the meaning of a word is something that is known to competent speakers of the language. That is why the Wittgensteinian strategy of replacing the question “What is meaning?” by the question “What is it to understand the meaning of an expression?” is not just 9 defensible but entirely reasonable. Mutatis mutandis for “What is a concept?” and “What is it to know or possess a concept?”. Unlike gold and cats, but like legal obligations and mortgages, concepts are not things one can point to and investigate directly. Hence it may be more fruitful to ask what it is to have a concept, just as it is more fruitful to ask what it is for a house to have a mortgage on it or for a person to be subject to a legal obligation.

5 Is concept-possession a disposition or ability? This leads us straight to our third question, namely what concept-possession consists in. Pragmatism identifies having a concept with possession of an ability, capacity or disposition. In what follows, I refrain from deciding which of these is

9

For strikingly different manifestations of this strategy, see Baker/Hacker (1980, essay 17); Dummett (1993, chs. 1-2) and Peacocke (1992).

52

Content, Consciousness and Perception 10

the most appropriate general category. Barring that issue, identifying conceptpossession with an ability or disposition of some kind is inevitable, and it is accepted, willy-nilly, even by proponents of RTM. Recall their need to distinguish between concept-types and concept-tokens, what Price had earlier called “occurrent concepts”. Let us accept for the sake of argument that it is the “tokening” of a concept-type in the brain which causally explains whatever the ascription of an occurrent belief causally explains (see Fodor 2003, pp. 137-8). Still, such tokenings cannot account for so-called “dispositional” beliefs, since they are temporary events rather than states in even the most catholic sense. A fortiori, therefore, they cannot account for the possession of concepts, since the possession of concepts must be at least as stable as the possession of dispositional beliefs. Put differently, concept-possession must belong to the category of disposition or ability, since it combines two features. On the one hand, it is enduring rather than episodic. On the other hand, it is something which manifests itself in certain episodes, notably of overt or silent classification or inference. It should not come as a surprise, therefore, that Fodor himself claims that on the Cartesian notion which he accepts: “having a concept is: being able to mentally represent (hence to think about) whatever it’s the concept of. So the concept DOG is that mental particular the possession of which allows one to present—to bring before one’s mind—dogs as such. One has the concept of a dog if, and only if, one is able to think about dogs as such” (2003, p. 19; my emph.).

Keeping in mind the type-token ambiguity, the tokening of a mental particular accounts for having occurrent beliefs, but the ability to do so accounts for conceptpossession. In a similar vein, Fodor’s pragmatist opponents insist on the need to draw attention to the “connection between having a concept and having it play its characteristic role in one’s cognitive (and behavioural) life”. For all his initial complaints, Fodor resigns himself to the idea that concepts must play such a role. He even accepts that concepts “are individuated by their function”. It is just that he thinks their function is to represent, and that concepts are individuated or “distinguished by what they mentally represent”:

10

Glock (2000) argues that concept-possession is an ability rather than a disposition, since it is not automatically exercised given certain conditions. This makes sense of the fact that automatic discriminatory behaviour is less suitable for the imputation of concepts than intentional discriminatory behaviour. On the other hand, the exercise of concepts in judgments is not voluntary.


53

“the use we make of concepts is to represent, in thought, the things they are concepts of. […] concepts are a species of mental representations, and are distinguished by what they mentally represent. The concept C is, simply, whatever it is with which the mind represents in thought the property of being C; …” (2003, pp. 14-16).11

Other representationalists admit that the idea of concept possession as an ability or capacity is “initially attractive”. But they are wont to add “until one starts to figure out what kind of ability it might be” (Crane 2004). And it is to this problem that we must now attend.

6 Is the pragmatist explanation of concept-possession circular? Representationalists take exception to the idea that concept-possession is a practical ability. Fodor defines pragmatism as the view that “the essence of a concept is in the way we apply it to things in the world, together with the inferences we use it to draw”, and maintains that pragmatism treats concepts as “constructions out of dispositions to classify things and to draw and [sic!] inferences”. He characterizes the first component variously as the ability “to apply the concept to things”, or to “sort things” in such a way that “clear cases” of the concept “end up in one pile” and clear cases of non-instances “end up in the other”. He characterizes the second component as the ability to draw or acquiesce in inferences involving the concept. Fodor further opines that this means that for pragmatists, conceptpossession involves a species of “knowing how” (2003, pp. 14, 17). The paradigmatic kind of concepts are those expressed by a general term F and designated by noun-phrases of the form “the concept of being (an) F” (e.g. the concept of being holy, of being true, of being water, of being an apple) or “the concept of F-ness”, where F-ness is a dummy for the morphologically acceptable nominalization of F (e.g. the concept of holiness, of truth, or of water, or of an 12 apple). It is natural to identify possession of these concepts with the ability to 11

In at least one respect, this way of contrasting pragmatism and RTM must be misleading even by Fodor’s own lights. It is not clear what “being able to mentally represent (hence to think about)” the instances of a concept is, if not entertaining thoughts about these instances (the embarrassment is especially acute, moreover, if one accepts that concepts must have a cognitive function). But it would have to involve entertaining simple thoughts like that an individual a is C. But in one sense of applying C, one simply cannot entertain a thought involving C without applying it. To think “Ca” is to apply the concept to a thing, rightly or wrongly. Nor is there anything particularly epistemological about that. What Fodor must have in mind is that having C involves both correctly thinking that a is C and thinking, once more correctly, that b is not C. 12 Most contemporary authors treat these noun-phrases as equivalent, but I do not wish to commit myself to this.

54


classify things into those that do and those that do not fall under them, provided that such classification is kept apart from mere discrimination (see section 7 below). However, the idea cannot easily be extended to other things which are commonly called concepts. Even if we leave aside the contested case of so-called individual concepts (the concept of Aristotle), there remain logical concepts (the concept of negation) and the concepts of space and time. Possession of these concepts is much more naturally explained by reference to the ability to draw certain types of inferences than by an ability to classify. Even if we allow for a diversity of cases, identifying concept-possession with abilities is contentious. A first objection raised by representationalists is the charge of behaviourism. A completely paralysed person can possess concepts, without being able to sort things or to draw linguistic inferences. According to Crane, pragmatists will respond that such thinkers still have the disposition or ability to do so. But how can we characterize the relevant disposition in a way that doesn’t beg the question? I don’t have the disposition to play the violin simply because I would be able to play it if I were taught it. Similarly, a paralysed person doesn’t have the ability to sort Fs simply because she would be able sort Fs and non-Fs if she were not paralysed. In this form, the accusation of behaviourism is based on ignoring two important distinctions. The first is the distinction between first- and second-order abilities. The ability to learn how to play the violin is a second-order ability and entirely distinct from the ability to play the violin. Possession of a mind may be identified with certain second-order abilities (see Kenny 1989, ch. 2). Yet pragmatists do not align concept-possession with such a second-order ability, but with various firstorder abilities. But isn’t it obvious that such abilities part company with concept-possession in cases of general paralysis? Not necessarily. For we must also distinguish between the absence of an ability and the absence of internal impediments (Kenny 1989, pp. 67-8; Hyman 1994). Crane’s argument fails to keep apart possession of an ability from having the opportunity to exercise it (absence of external impediments) and from being in a state to exercise it (absence of internal impediments). A paralysed person may have the ability to sort things, if the impediment is removed. This is different from a person that needs to acquire an ability by learning and training. It is one thing to lack the ability to ski, another to be unable to ski because of a broken leg. There is an “all-in can” according to which a paralysed person cannot ski, or sort, or draw linguistic inferences. But the “can” invoked by pragmatists is not of this sort, and a person that is only temporarily paralysed can sort and draw inferences in the sense pertinent here. What about a human being that has never had the ability, e.g. because she has been paralysed from birth? Here the distinction between having the ability to sort and absence of internal impediments to sorting wears thin indeed. But so does the


55

insistence on the possession of concepts. Unless, of course, she has at least a capacity to distinguish things in perception or thought. Accordingly, the pragmatist has a second response: Concept-possession can equate to an epistemic or cognitive 13 rather than a practical ability. At this point, however, representationalists like Fodor see the opportunity to pounce. If to possess concept F is simply to be able to discriminate or to draw inferences about Fs in thought, pragmatism is circular, since it purports to explain thought. More generally, any epistemic capacity presupposes rather than explains representation. In Fodor’s view, pragmatists insist that concept-possession “is to be understood largely in terms of epistemological notions like warrant and justification”. But this means that they “have things back to front”: “epistemic capacities require ways of representing the intentional objects of epistemic attitudes. So epistemic capacities don’t constitute concepts, but merely presuppose them” (2003, pp. 19-20).

Fodor goes on to accuse the pragmatists of begging the question against RTM. But in fact it is his quote that commits a petitio, since it simply presupposes that cognitive achievements must be based on the having of mental representations. The general circularity charge begs the question in favour of RTM. It provides no independent reason whatever against the pragmatist claim that epistemic achievements amount to the exercise of an ability rather than to the occurrence of a representation (mental or neural particular). The argument is more attractive when it is spelled out with respect to the two types of abilities central to pragmatism. The more specific contention is that “thinking about something is prior to inference and discrimination” (Crane 2004). If A is capable of thinking about Fs: • A can also distinguish them from non-Fs, since it can think about how they differ from Gs. • A can also move from F-involving premises to G-involving conclusions. By contrast, how can a mind

13

Arguably, even such a non-practical capacity must be capable of being displayed in some way or other, if only through minimal reactions to the environment and/or to questions put to the subject. What if there is no capacity to discriminate even in this minimal sense? Couldn’t there be a developed capacity for abstract thought in a creature completely and forever incapable of reacting to her environment or displaying her thoughts in any way? In my view, the answer is negative (see Glock 1997). And if this be behaviourism, then the latter can’t be all wrong.

56

Content, Consciousness and Perception 14

(i) tell an F from a G, unless it is already able to think about Fs and Gs? (ii) infer from x being F to x being G, unless it is already able to think about Fs? The answer with respect to (i) is that a mind can discriminate Fs and Gs in perception, without automatically being credited with thought about Fs and Gs. This holds in particular if one holds, with RTM, that to have concept F is to be able to think about Fs or F-ness as such, i.e. in a way that distinguishes Fs from Gs even if all and only Fs are Gs, indeed, even if necessarily all and only Fs are Gs (see below). Similarly, it is possible that a creature can react to detecting Fs in a way appropriate to the presence of Gs without ipso facto being credited with inferences from Fs to Gs. Both representationalism and pragmatism seek to account for conceptual thought. Fodor is right, moreover, in holding that at least prima facie representationalism has a meatier tale to tell. Unfortunately, we have seen that the meatier tales served by mentalist, neurophysiological and computational representationalism do not add up. Austere representationalists such as Husserl (1900), by contrast, have disassociated the idea of representation from any specific connotations, whether they be mental images or words crossing ones mind, neural firings or computational symbols. But such positions once more face the task of explaining what having a representation amounts to. In Husserl's case, for example, we seem to be left with the idea that it is “just like” mental picturing, only without mental images. But that simply boils down to saying that having a representation of (an) F is to think about (an) F, which means that the explanation of thought has moved in a circle. A logical relation, thinking about (an) F, has first been construed as a pictorial one, and has then been robbed of the pictorial aspects which alone can give it any substance (Tugendhat 1976, pp. 62-3, 276-7). There is a very general notion of representation which applies to propositions and concepts, simply on account of the fact that both have semantic properties. Propositions are capable of being true or false, and they represent things as being a certain way. Similarly, concepts apply to the things that fall under them and have certain defining characteristics. But of course this general notion of representation is entirely unsuitable to explaining what thought is. For to represent Fs in this general sense is simply to think about Fs. If my arguments are sound, representationalism is reduced to uninformative claims like the following: to think that a is F is to represent a as being F, and to think of Fs qua Fs is to represent Fs. By contrast, pragmatism at least holds the promise of a genuine explanation: what it is to be able to think about Fs as Fs is spelled out in terms of other abilities. The idea is that there are preconceptual forms 14

Similarly, Crane inquires how a mind can tell Fs and Gs apart unless it is already capable of thinking about Fs and Gs.


57

of perception and action, and that these differ from the conceptual forms through the absence of certain cognitive abilities.

7 Are concepts more finely individuated than abilities? At this stage pragmatism faces a final objection, one which concerns the individuation question. For instance, creature with a kidney and creature with a heart apply to the same things, yet are different concepts; equilateral triangle and equiangular triangle apply necessarily to the same things, yet are different concepts. In current jargon, concepts are “hyperintensional”. Now, an ability is individuated by reference to its exercise. But, Fodor rightly points out, the same sorting and inferential performances can manifest the possession of different concepts. Confining ourselves to the ability to sort or discriminate, sorting equilateral triangles from all other figures is also sorting equiangular triangles from all other figures (Fodor 2003, pp. 25-6, 143-6). It seems to follow that concepts cannot be individuated by the exercise of an ability, and hence that they cannot be individuated by reference to abilities: P3 P4 C2

Abilities are individuated by their exercise (ability to Φ = ability to Ψ iff Φing = Ψing) In all possible situations, one and the same sorting activity can manifest different concepts Concepts cannot be individuated through the abilities which allegedly constitute their possession.

Fortunately, the pragmatist can repudiate the second premise: sorting triangles according to lengths is not the same activity as sorting triangles according to angles, even though the results are the same. The difference in the two activities can obviously be displayed by linguistic creatures, who can justify their sorting along different lines. It can even be manifested in non-linguistic behaviour. A creature that sorts on account of measuring lengths applies EQUILATERAL TRIANGLE, a creature that sorts on account of measuring angles applies EQUIANGULAR TRIANGLE. These are different activities, manifesting different abilities and concepts. And it is obvious that one can have one ability and one concept without having the other. Indeed, most children in fact learn how to measure lengths before learning how to measure angles. Consequently individuation is no obstacle to pragmatism. This leaves open the question of whether concepts should be individuated through their possession conditions, as Peacocke, for one, maintains. For one thing, most pragmatists agree that there is no precise way of individuating abilities. Thus Travis (2000) grants that linking concepts to abilities may not be much help in individuating concepts,

58


since it is not clear how abilities are to be counted. It is important to be precise about the scope of this concession. Like Travis, Geach (1957, p. 15) accepts that it is absurd to ask how many abilities are exercised in a judgment. Yet, unlike Travis, he also insists, rightly, that we can still distinguish different such abilities. More generally, one must distinguish between the possibility of enumeration and the possibility of individuation (see Strawson 1997, ch. 1; Glock 2003, pp. 47-52) of abilities and hence different concepts. Nevertheless there remains a problem. For it is at least plausible to hold that we can not just distinguish the concept of a dog from that of barking, but also specify that two concepts are involved in the judgment that dogs bark. So concepts and abilities may come apart on the issue of enumerability. Furthermore, the question remains how we individuate concepts. A concept cannot be individuated by the features that its instances do in fact possess. Nor can it be individuated by reference to the features its instances must possess, because in order to satisfy the concept of being triangular, x must possess the feature of being trilateral. An even stronger condition is that concepts are individuated through their definitions: the proper definition of the concept of triangularity is different from the proper definition of the concept of trilaterality. By this light, the concept expressed by a predicate “F” is identical with that expressed by a predicate “G” iff the two are synonymous. Of course the question of what synonymy amounts to is itself contested. Still, a plausible definition for two predicates being synonymous is that they should be interchangeable in all declarative sentences (except for special cases, e.g. when “F” and “G” are mentioned rather than used) in such a way that the results are what Frege (1979, p. 197) called “equipollent” or cognitively equivalent: nobody who understands them both can accept what the one says without also accepting what the other says (see also Rundle 1979, §54; Glock 2003, pp. 223-4). If synonymy thus understood holds the key to the identity of concepts, then the latter will make reference to a capacity, the capacity for linguistic understanding. This would reinstate a pragmatist account of concept-possession. It would also support a specific version of pragmatism, namely one which identifies concepts-possession with linguistic abilities, roughly, the ability to use, understand and explain linguistic symbols. In my view, by contrast, having concept F requires only the capacity for classification, and thereby for considered judgment that something is F, not the capacity to say so (Glock 2000). This position is not ruled out by the account of concept-individuation just canvassed. We may have grounds for ascribing concepts to creatures who lack the abilities by reference to which concepts are ultimately individuated, as long as there is some substance to the idea that their non-linguistic behaviour manifests concept F rather than concept G. Still, by way of conclusion I want to sketch briefly a different solution to the individuation problem. It is nonlinguistic, and identifies concept possession with abilities that are not confined to


59

language users. To possess the concept F is not just the ability to sort things into Fs and non-Fs, but the ability to classify them as Fs. And that is to distinguish Fs from non-Fs for the reason that the former possess certain features, namely those constitutive of being F. If A singles out triangles on account of the numbers of the angles, she exercises a different concept from B, who singles them out on account of the number of their sides. It might be held that the fact that x possesses the features constitutive of being F is simply identical with the fact that x satisfies the concept of an F. This may well be so, but doesn’t render the account circular. For even though “possesses the features constitutive of being F” and “satisfies the concept of an F” are necessarily co-extensive, they are not for that reason synonymous, i.e. they do not express the same concept. Circularity would only result if the explanation of what distinguishes concept F were to make use of the concept F itself, rather than of a term which is necessarily co-extensive with it. If this solution to concept individuation is feasible, it will link the two types of abilities with which concept possession is standardly associated. To classify things—unlike mere discrimination—involves reasoning and hence something like inference. Nevertheless, it extends to non-linguistic creatures, at least in principle. For instance, if chimpanzees learn to pick out certain shapes by paying attention to the number of sides, this is acquisition of one concept, if they do so by paying attention to the number of angles, this is acquisition of another concept. Even inferring is not confined to linguistic creatures on conceptual grounds. Consider the story of Chrysippus’ hunting-dog. In chasing a prey of which it has lost the scent, this dog reaches a cross-roads; it sniffs down the first path, then sniffs down the second path, then it immediately follows the third without sniffing. In the case of dogs, perhaps such behaviour could only be a rigid conditioned reflex. But I can see no reason for denying that this is an intelligible form of behaviour for a nonlinguistic creature capable of voluntary action. Furthermore, while the course of action of following the third path could be the conclusion of any number of lines of reasoning (the starting-point of Fodor’s argument concerning inferences), the nonlinguistic behaviour of this remarkable dog provides grounds for regarding its actions as based on a particular disjunctive inference (“p or q or r; neither p nor q; ergo r”).

Bibliography Baker, G.P. and Hacker, P.M.S. (1980). Wittgenstein: Understanding and Meaning. Oxford: Blackwell. Crane, T. (2004). “Something Else, Surely”, in TLS 7.05.04: 4. Davidson, D. (1996). “The Folly of Trying to Define Truth”, in Journal of Philosophy 93: 263-78.

60


Dummett, M. (1993). The Seas of Language. Oxford: Clarendon. Fodor, J. (1975). The Language of Thought. New York: Crowell. ——— (1987). Psychosemantics. Cambridge, Mass.: MIT Press. ——— (1998). Concepts: where Cognitive Science Went Wrong. Oxford: Clarendon. ——— (2003). Hume Variations. Oxford: Clarendon. ——— (2004). “Having Concepts: a Brief Refutation of the Twentieth Century”, in Mind and Language 19: 29-47. Frege, G. (1886/1953). Grundlagen der Arithmetik; English translation The Foundations of Arithmetic. Oxford: Blackwell. ——— (1892/1997). “Über Sinn und Bedeutung”, in M. Beaney (ed.), The Frege Reader. Oxford: Blackwell. [References are to the original pagination.] ——— (1893/1997). “Grundgesetze der Arithmetik Vol. I”, in M. Beaney (ed.), The Frege Reader. Oxford: Blackwell. [References are to the original pagination.] Geach, P.T. (1957). Mental Acts. London: RKP. Glock, H.J. (1997). “Philosophy, Thought and Language”, in J. Preston (ed.), Thought and Language. Cambridge: Cambridge University Press. ——— (2000). “Animals, Thoughts and Concepts”, in Synthese 123: 35-64. ——— (2003). Quine and Davidson on Language, Thought and Reality. Cambridge: Cambridge University Press. ——— (2005). “The Normativity of Meaning Made Easy”, in C. Nimtz and A. Beckermann (eds.), Philosophy and Science, Proceedings of GAP 5. Mentis: Paderborn. Hacker, P.M.S. (1987). “Languages, Minds and Brains”, in C. Blakemore and S. Greenfield (eds.), Mindwaves. Oxford: Blackwell. Haller, R. (1971). “Begriffe”, in J. Ritter (ed.) Historisches Wörterbuch der Philosophie (Vol. 1). Darmstadt: Wissenschaftliche Buchgesellschaft. Husserl, E. (1900/1968). Logische Untersuchungen (second edition). Tübingen: Niemayer. Hyman, J. (1994). “Vision and Power”, in Journal of Philosophy 91: 236-52. ——— (2006). The Imitation of Nature. Chicago: Chicago University Press. Kenny, A.J.P. (1989). The Metaphysics of Mind. Oxford: Clarendon. Künne, W. (2003). Conceptions of Truth. Oxford: Oxford University Press. Laurence, S., and Margolis, E. (1999). “Concepts and Cognitive Science”, in E. Margolis and S. Laurence (eds.), Concepts: Core Readings. Cambridge, Mass.: MIT Press. Libet, B. (1987). “Consciousness: Conscious, Subjective Experience”, in G. Adelman (ed.), Encyclopedia of Neuroscience. Boston: Birkhäuser. Loewer, B., and Rey, G. (1991). “Introduction”, in B. Lower and G. Rey (eds.), Fodor and his Critics. Oxford: Blackwell.


61

Lowe, J. (1997). “Objects and Criteria of Identity”, in B. Hale and C. Wright (eds.), Blackwell Companion to the Philosophy of Language. Oxford: Blackwell. Margolis, E. and Laurence, S. (eds.) (1999). Concepts: Core Readings. Cambridge, Mass.: MIT. Peirce, C.S. (1933). The Collected Papers of Charles Sanders Peirce Vol. II. Cambridge, Mass.: Harvard University Press. Peacocke, C. (1992). A Study of Concepts. Cambridge, Mass.: MIT Press. Price, H.H. (1953). Thinking and Experience. London: Hutchinson. Rey, G. (1994). “Concepts”, in S. Guttenlplan (ed.) A Companion to the Philosophy of Mind. Oxford: Blackwell. ——— (1998). “Concepts”, in E. Craig (ed.) Routledge Encyclopedia of Philosophy. London: Routledge. Rundle, B. (1979). Grammar in Philosophy. Oxford: Clarendon. Schwartz, R. (1994). “Representation”, in S. Guttenplan (ed.), A Companion to the Philosophy of Mind. Oxford: Blackwell. Strawson, P.F. (1997). Entity and Identity. Oxford: Oxford University Press. Travis, C. (2001). Unshadowed Thought: Representation in Thought and Language. Cambridge, Mass.: Harvard University Press. Tugendhat, E. (1976/1982). Traditional and Analytic Philosophy. Cambridge: Cambridge University Press. Williamson, T. (1990). Identity and Discrimination. (Oxford: Blackwell).

CHAPTER THREE HUMAN PHILOSOPHY: HUME ON NATURAL INSTINCTS AND BELIEF FORMATION

TREASA CAMPBELL

It is difficult to locate any area of Hume's philosophy which is not loaded with sceptical import, and his account of belief-formation is no exception.1 Yet, in spite of his sceptical arguments, there is a positive or constructive side to his writings. There is little doubt that Hume wished to distinguish between reasonable and irrational beliefs, "weaning our mind from all those prejudices, which we may have imbibed from education or rash opinion" (E182).2 For Hume a number of the most fundamental principles upon which we operate in the world have a particular set of characteristics, which marks them out. In order to draw specific attention to the significance of the existence and specific characteristics of this small group of operating principles, commentators have termed them “natural beliefs”.3 Though Hume never used the term “natural belief”, there is general agreement that such a 1

Early versions of this material were first presented at the Universities of Ireland Postgraduate Philosophy Conference 2005 and later at Mind2005. I would like to thank audiences on those occasions, especially Dr. Andrew Mason, my commentator at Mind 2005, for extremely helpful commentary and discussion. I would also like to thank Dr. Stephen Thornton and Dr. Stephen Bond for many helpful suggestions in the preparation of this paper. Work on this material was generously supported by a studentship award from Mary Immaculate College, Limerick. 2 I will use the following abbreviations in referencing works by David Hume: T: A Treatise of Human Nature. E: An Enquiry Concerning Human Understanding. A: “An Abstract of A Treatise of Human Nature”, in An Enquiry Concerning Human Understanding. 3 See Butler (1960), Gaskin (1974), McCormick (1993), Kemp Smith (1905).


63

class of beliefs exists for Hume. Discussion of what are termed “natural beliefs” has become central to those who consider Hume’s main concern to be revealing non-intellectual resources, located within our human nature, which enable us to interpret and respond to our experience.4 However, I take the fact that Hume never used this term as significant and I will argue that such a characterisation is a fundamental misrepresentation of the nature of the phenomena under consideration. I will make the case for reverting to Hume's term “natural instincts”, and seek to demonstrate how this term fully exposes the radical nature of Hume's theory. For in arguing that the phenomena in question are unavoidable instincts and not some form of belief, the phenomena become part of our cognitive abilities and are thus essential in establishing relevant and genuine epistemic norms for the human species. If philosophy is to have any relevance to us as human beings, it must take account of the cognitive capacities of our species. First, however, I would like to explain what it is that commentators mean when referring to “natural belief”. In doing so I can demonstrate how the term “natural belief” fails to take cognisance of the radical change for which Hume is calling.

1 The traditional use of the term “natural belief” The term “natural belief” was introduced by Kemp Smith to refer to a small group of beliefs that cannot be justified by reason, yet are unavoidable features of how we operate in the world. “Natural belief” is the terminology that is now widely used in the literature on Hume to refer to such features. McCormick tells us that there are a number of "important beliefs that, although he (Hume) never uses the term, commentators have come to call “natural beliefs.”” (McCormick 1993, p. 103.) These beliefs are unavoidable, indispensable and irresistible. Not only do many hold that such a class of beliefs exist, but for many it is a fundamental aspect of Hume's work. Butler, in his article “Natural Beliefs and the Enigma of Hume”, believed it was “to Hume's eternal merit to have suggested the existence of such a class.” (Butler 1960, p. 74.) Kemp Smith regarded Hume's doctrine of natural beliefs as his positive programme, his defence against scepticism. This new doctrine of beliefs is, Kemp Smith thinks, “one of the most essential and perhaps the most characteristic doctrine in Hume's philosophy.” (Kemp Smith 1905, p. 151.) Perhaps the simplest way to understand how it is we should understand the concept of “natural belief”, and what exactly is being indicated by the term, is to look at how its advocators mark it out from other more familiar categories of beliefs. 4

See commentators such as Garrett (1997), Kemp Smith (1941), Stroud (1987) and Strawson (1989).

64


Belief, on this view, can be crudely broken down into “rational beliefs” and “non-rational beliefs” and when further refined the category of “non-rational beliefs” can be broken down into “natural beliefs” and “irrational beliefs”. For Hume belief is a feeling; to have a belief in some idea is to feel a certain way about it. Beliefs consist “in their feeling to the mind”; “belief is something felt by the mind” (T146). Borrowing a term used by J.C.A. Gaskin in his 1974 article “God, Hume and Natural Belief”, I shall refer to “belief-feeling” (Gaskin 1974, p. 282) as the feeling that constitutes belief. There may be an impulse to appeal to the accompanying belief-feeling as a means of making the above categorisations of belief; however the very fact that the above categories are all considered to be some form of belief necessitates that they all have this attendant “belief-feeling”. There is no special kind of feeling which marks one form of belief out from another. What then are the grounds for the above classification of the general body of beliefs? The principle of differentiation is the ground upon which their justification lies. The first of these divisions is that between rational and non-rational beliefs. Rational beliefs, as the name suggests, point to reason as their source and in doing so can be justified rationally. Non-rational beliefs make no such claim on reason; without such a claim they relinquish the ability to appeal to rational justification in order to underpin their status. Rational beliefs can be modified in the light of new evidence. In the case of rational belief, thinking and assessing influence the belieffeeling, whereas thinking and assessing do not influence the belief-feeling of nonrational belief. Gaskin describes rational beliefs as beliefs which are the result of a "conscious rational assessment of evidence" (Gaskin 1974, p. 284). With rational beliefs, then, belief-feeling occurs, fails to occur or is modified as the involuntary effect of careful detailed assessment of the evidence. We turn then to “non-rational beliefs”, those in which thinking and assessing can have no effect on the belief-feeling. When we speak of “non-rational beliefs” we are in fact talking of two quite distinct forms of belief, “irrational beliefs” and what has been termed “natural beliefs”. This further breakdown is made once again along the differing statuses of their justification. Both are non-rational in that neither can appeal to reason as the basis for their justification. Where they part company is that for irrational beliefs the story ends there, the beliefs in question cannot lay claim to any coherent form of justification and they are thus classed as “irrational” or unjustified. They are baseless and can carry no authority in our knowledge system, as they are without any solid grounds. We thus arrive at the question of how it is we are to understand natural beliefs. With Hume arguing that many of our core beliefs are not grounded in reason (as traditionally conceived) and that evidence of the lack of rational support does not shake our commitment to them, the ability to distinguish between irrational and natural beliefs becomes crucial. The phenomena that Hume exposes, which have come to be characterised as natural beliefs, are non-rational, they have a certain


65

degree of force which for Hume is characteristic of belief,5 and they are unavoidable, this latter feature being of prime importance. The fact that they are unavoidable is crucial, because it takes more than not being rational (being nonrational) to be an irrational belief. Similarly, in possessing a certain degree of force they remain undifferentiated from other beliefs, as a degree of force is presented as a characteristic of all beliefs. “Natural beliefs” are non-rational, but necessary as a precondition of action, and they are universally held. They cannot be justified rationally but are impossible to give up; no amount of reasoning can eliminate them. For Hume, then, a number of our fundamental beliefs are not the result of a conscious rational assessment of evidence. Yet common experience reveals that such beliefs cannot be upset by such an assessment except in brief moments of “philosophical melancholy and delirium.” (T316) They may be non-rational, but one does not wish to call these beliefs either irrational or unreasonable, as they are indispensable for human action. Two questions now occur: which beliefs are to count as natural beliefs, and what characteristics do they have which make them natural beliefs rather then irrational or unreasonable beliefs? As said, in common with irrational beliefs, in the case of natural beliefs there is an inability to produce a grounding in reason. But, unlike irrational beliefs, there is no evidence which makes it more reasonable to adopt any alternative set of beliefs. The sceptical criticisms of natural beliefs can establish no other in their place. Unlike irrational beliefs, there is justification for acting as if natural beliefs are true. This justification lies in the unavoidability of natural beliefs. Natural beliefs then are to be understood as beliefs that are “universal, inevitable, unavoidable, or necessary pre-conditions for actions.” (McCormick 1993, p. 106.) It is clear that the vast majority of beliefs will fail to satisfy all of these stringent criteria. Indeed, most would not satisfy any of them, and as a result the set of beliefs which satisfies the criteria for being “natural” is extremely small. Gaskin (1974, p. 285) held that Hume locates the following set, which possess all of these elusive characteristics: (1) (2) (3)

belief in the continuous existence of an external world independent of our perception of that world (E160) Belief that the regularities which have occurred in our experience form a reliable guide to those which will occur (many locations). Belief in the reliability of our senses qualified to take account of acknowledged and isolatable areas of deception and confusion (many locations).

5 For Hume it is impossible to explain perfectly the feeling which constitutes belief, but he tells us that "this different feeling I endeavour to explain by calling it a superior force, or vivacity, or solidity, or firmness, or steadiness." [T146.] The distinction between fiction and belief is then given in terms of a greater level of force exerted by the latter.

66


It should be clear from this brief description that the nature, status and epistemological implication of what has been termed “natural beliefs” are of central concern to Hume's project. For they expose another vital step taken by the mind in the formation of human knowledge. However, it is my contention that the term “natural belief” introduced by Kemp Smith and continued throughout the literature is not a satisfactory description of the phenomena which Hume is exposing. I will argue this by reference to the distinction between propensities and disposition as set out by Robert Paul Wolff in his 1960 article “Hume’s Theory of the Mental.”

2 Wolff's propensity/disposition distinction Wolff, in his article “Hume’s Theory of the Mental”, argues that what we find emerging in Hume’s analysis of cognition is a distinction between what he terms “dispositions” and “propensities”. Of course, both of these terms have been used in a variety of different contexts throughout the history of philosophy, and so I want to be very clear on exactly how it is we are using them here and exactly what phenomena they denote. This clarity is all the more essential as the present usage of the terms “dispositions” and “propensities” differs from Hume’s own usage of the terms. Wolff acknowledges, as do I, that Hume “does not observe strict distinctions among “habit,” “custom,” “propensity,” and “dispositions”.” (Wolff 1960, p. 106.) Wolff, however, uses these terms to mark out two separate but closely related phenomena, the existence of which, though not the specific terminology, are found in Hume’s description of the cognitive process. Wolff’s distinction is as follows: "(The term ‘disposition’ is used) to describe the fact that an entity is prone to act or react in certain ways under certain conditions.” (Ibid.) "(The term ‘propensity’ is used) to describe the fact that an entity is prone to develop certain dispositions under certain conditions.” (Ibid.)

Wolff makes use of the example of Pavlov’s dog6 to illustrate this distinction. In this particular example Pavlov rings a bell repeatedly and immediately afterwards gives meat to his dog. After numerous repetitions of this process, simply ringing the bell provokes salivation in the dog. What is of significance to us here is Wolff’s claim that “it is not enough to say that Pavlov rang the bell every time the food was offered.” (Ibid., p. 105.) The repetition process is of course a necessary part of the explanation but crucially, according to Wolff, “we must add that the dog was capable of being conditioned. If it were not, no amount of bell-ringing could 6

Ivan Pavlov describes how animals (and humans) can be trained to respond in a certain ways to a particular stimulus. See Pavlov (1927).


67

produce the conditioned reflex of salivation.” (Ibid.) It is this capacity to form particular dispositions that we are here referring to as a “propensity”. In applying our “propensity”/“disposition” distinction, then, we can say that: “the dog’s salivation upon the ringing of the bell is a manifestation of a disposition (in this case a conditioned reflex) and the disposition in turn is a manifestation of the dog’s propensity to form dispositions of this sort.” (Ibid., p. 106.)

In doing so we have marked out two closely related yet distinct phenomena. A propensity is something that exists prior to any stimuli, and in accordance with Hume’s empiricism, “stimuli” is to be understood here as sensory in nature. Dispositions are dependent on both stimuli and propensity, while propensities precede both stimulus and dispositions. Wolff describes propensities as the “necessary conditions for the development of dispositions.” (Ibid., p. 107.) Dispositions are dependent upon the conditioning stimulus and follow after them. Stimuli are what Wolff calls the “individuating conditions of the disposition being formed.” (Ibid., p. 106.) He tells us “the nature of the stimulus determines the precise character, or content, of the disposition.” (Ibid.) Stimuli are intrinsic to the formation of a disposition; for example we can change the content of Pavlov’s dog’s disposition by altering the stimulus, for example substituting the ringing bell with a drum or horn. A second role Wolff attributes to stimuli is that they are “the triggers of the dispositions already formed.” (Ibid.) Each time we want to activate the disposition Pavlov’s dog has formed we must produce the appropriate sensory stimuli. Similarly, when the human mind is presented with perceptions conjoined in certain ways, its propensities are activated and it develops dispositions. For Wolff it is clear that the various principles invoked by Hume have the characteristics of dispositions and propensities and his article continues to draw this out. Regardless of how one feels about the particular choice of terminology, what is important at this point is simply that the reader acknowledges the logical distinction being made between a pre-stimulus, innate capacity (propensities) and what develops from them as stimuli are added (dispositions). With this distinction in mind I return to my case against the widespread use of the term “natural beliefs” in Hume studies.

3 Natural instincts My objection is not with the “natural” element but rather with the term “beliefs”. Hume himself does note that the word “natural” has many different meanings. Concerning the word “nature”, Hume says, “there is none more ambiguous and equivocal” (T525). In her article “Hume on Natural Belief and Original Principles”, Miriam McCormick identifies at least three ways Hume uses the word “natural”: to mean “ordinary”, to mean “part of nature”, and to mean “inseparable from the species” (McCormick 1993, p. 103). It is true that certain

68


characteristics may be called “natural” insofar as their causes are found in human nature. But they need not be necessary or unavoidable. I wish to retain the use of the term “natural” where it is to designate “unavoidability”. While retaining the use of “natural” in this sense, I nevertheless wish to put forward the case for substituting the term “belief” with the term “instinct” in this context. Having recognised that Hume himself never used the term “natural beliefs” to refer to the small number of unavoidable, indispensable and irresistible mental features which he discusses, Gaskin justifies his use of the term as follows: “With Butler I shall normally refer to them as “natural beliefs”: the phrase is not in fact used by Hume but is entirely consistent with his usage.” (Gaskin 1974, p. 284.) It is my contention that the term “natural belief” is not, as Gaskin claims, entirely consistent with Hume's usage. Hume at different times uses the following terms: “faculty”, “property”, “principle”, “quality”, “disposition”, “propensity”, “inclination”, “structure” and “capacity”, to describe these determinations of the mind which form our constant human nature. However, at key points in his texts when dealing with the phenomena which commentators have termed “natural beliefs” Hume chooses consistently to use the term “natural instincts”. Gaskin himself recognises that the phenomena in question “belong to the class of beliefs which Hume sometimes speaks of as ‘natural instincts’.” (Gaskin 1974, p. 284.) Indeed, when we try to conceive of that which is unavoidable, indispensable and irresistible to our very existence, the term “instincts” comes more readily to mind than “beliefs”. As we can see from the following passages, Hume, while never using the term “natural beliefs”, when speaking of such phenomena, regularly refers to “instincts”: “It seems evident, that men are carried, by a natural instinct or prepossession, to repose faith in their senses…” (E182.) (My italics.) “There is a great difference betwixt such opinions as we form after a calm and profound reflection, and such as we embrace by a kind of instinct or natural impulse, on account of their suitableness and conformity to the mind.” (T264.) (My italics.) “To bring us to so salutary a determination, nothing can be more serviceable, than to be once thoroughly convinced of the force of the Pyrrhonian doubt, and of the impossibility, that anything, but the strong power of natural instinct, could free us from it.” (E192.) (My italics.)

When speaking of causal inferences Hume says that “nothing leads us to this inference but custom or a certain instinct of our nature” (E190). In regularly referring to “natural instincts” Hume continually highlights the instinctual, innate, primacy of the phenomena to which he is referring. He is asking more then a mere reclassification of these phenomena as a unique form of beliefs.


69

He calls for us to see that these phenomena should no longer be viewed as any form of belief but as instinctual features of our being which are inseparable from our species. These instincts are inexplicable in themselves but explain much of human life. They are for Hume universal and necessary for all human beings. They are “inseparable from human nature, and inherent in our frame and constitution” (T633), and as such are indispensable despite their lack of rational grounds. As a result their causes are inexplicable and they are in a sense ultimate. On the existence of the external world, for instance, Hume says “tis in vain to ask, Whether there be body or not? That is a point, which we must take for granted in all our reasonings” (T238). We have no choice but to take “natural instincts” “for granted in all our reasoning.” We must then follow the blind dictates of human nature, because “Nature, by an absolute and uncontrollable necessity has determin’d us to judge as well as to breathe and feel” (T234). We are being asked to let go of the idea that these phenomena are beliefs and accept them in their true form as “a species of natural instinct, which no reasoning or process of the thought and understanding is able either to produce or prevent” (E90). The great insight of Hume, I wish to argue, is that these operating principles which traditionally have been viewed as beliefs are not in fact beliefs at all, but are instincts. They are not a class of beliefs, some necessary outgrowth of certain original, inexplicable and universal principles of human nature. They are original, inexplicable and universal principles of human nature. As such, they are a vital component of our belief-forming structure. They exist, not as a product of the belief-forming structures of the mind, rather they are an indispensable, instinctual part of those very belief-forming structures. Indeed these “natural instincts” so exceed every other in their power to inspire belief that, "upon their removal human nature must immediately perish and go to ruin" (T274). If we cling to the idea that such phenomena are beliefs then we cannot attribute them any place in our beliefforming process. For there is an implicit contradiction involved in identifying particular beliefs as part of a belief-forming structure, for how can a belief (natural or otherwise) be part of the mind’s belief-forming structures, which it must presuppose? That I operate in the world as if the continued and distinct existence of bodies is a reality or as if the reliability of inductive belief-formation is a reality is not the consequence of any belief which I affirm. Rather, it is a result of universal, unavoidable instincts to which my humanity binds me. These are fundamental components of how we operate in the world and of how we form beliefs. In short, they are instincts, and Hume’s challenge to us is to view them as such. These natural instincts are part of our very nature and are to be regarded as features of human cognition; we cannot proceed as if we were organisms of a very different kind. Natural instincts are fundamental principles of enquiry, or the substrata of thought; as such they can only be described, because there is nothing in terms of which they could be explained. If normative epistemology is to be relevant to

70


human beings, it must itself be built on a descriptive account of human cognitive abilities. Any account of norms depends on the cognitive abilities we have access to. The mind is, for Hume, determined in this way, emphasising again the unavoidable necessity to operate in accordance with certain natural instincts set out by our nature. As such the kind of justification, if any, demanded of a natural instinct is not the same as that required of beliefs. In characterizing the phenomena in question as instincts and not as beliefs, Hume has enacted a dramatic transformation in the epistemological landscape. Of possible concern7 is how the affirmation of such principles—an inner affirmation, not just a speech-act—could fail to be understood as a belief. But this inner affirmation is nothing more than the conscious manifestation of an inescapable instinct. That we operate in the world as if the continued and distinct existence of bodies is a reality is an instinct and not a matter of belief. The inner affirmation is wholly secondary; remove the instinct and the inner affirmation is no more, remove the inner affirmation and we will continue to operate in the same way. If we are to have any understanding of the universal, constant features of our nature which Hume discusses we must come to see them as natural instincts. Applying Wolff's distinction to Hume, then, we may accordingly say that instincts (innate propensities) combined with the stimuli of the senses to give rise to certain “natural beliefs” (dispositions). Nevertheless, commentators have obscured these instincts (propensities) and focused exclusively on what arises from them through the influence of stimuli (dispositions). In so doing it is the latter that have become the central focus. They simply merge together and no differentiation is made8. In identifying Hume's term “natural instincts” and Kemp Smith's term “natural beliefs” as interchangeable we fail to make the logical distinction between first level innate propensities or instincts and that which arise from such instincts, namely particular beliefs or dispositions. This lack of clarity has enabled a situation to arise where phenomena, which Hume describes and wishes to be treated as instincts, are instead treated as beliefs. Hume argued that humanity, like all the animal kingdom, has an inborn expectation that factors observed to be constantly conjoined in the past will continue to be conjoined in the future. This “natural 7

This concern was raised by Mind2005 conference commentator Dr. Andrew Mason during his review of this paper. 8 Dr. Mason’s review recognised that the distinction between propensities and dispositions would indeed be a fruitful one and held there to be much truth in the view that the universal, constant features of our nature which Hume discusses are not beliefs, but rather propensities to form particular beliefs. Dr. Mason held that there is potential for using this distinction to illuminate an area, in which the topic of “natural belief” is often discussed; that of Hume’s philosophy of religion as authors such as Butler, claim that belief in God is presented in the Dialogues as a “natural belief.”


71

instinct” gives rise to certain content-specific beliefs which may be wrong for certain pairs of factors, and weaker and stronger for others, but the instinct (propensity) to extrapolate such conjunctions is always there. When commentators refer to “natural beliefs” and Hume refers to “natural instincts” the same phenomenon is usually being discussed, viz., a universal, constant feature of human nature, yet they are fundamentally mischaracterised when described as beliefs. What should rightfully be termed “natural beliefs” are the individual, content-specific beliefs that inevitably emerge as our natural instincts are exposed to the stimuli provided by the senses. We have, for example, the individual, content-specific belief that the sun will rise tomorrow. But this is not what Hume is striving to reveal to us. What Hume's work masterfully exposes is that the expectation which I possess and that gives rise to this belief (that the regularities which have occurred in my experience are a reliable guide to those which will occur in the future) is not to be understood in terms of a belief; rather, it is an unavoidable instinct which I have as part of the very nature of the species to which I belong. Hume's terminology hits closer to what is central, namely, the exposition of these constant features of human nature as inseparable from the species. The term “natural instinct” indicates clearly that what is being referred to is not simply “a class of beliefs” but the original and ultimate instincts from which all action and all beliefs arise. The use of the term “natural instincts” over the term “natural belief” is more than a pedantic point of taxonomy, but rather calls for a radical transformation of our understanding of these phenomena, a move from regarding them as beliefs to viewing them as instincts. We must understand that for Hume the phenomena commentators term “natural beliefs” are seen, not as a set of beliefs which we construct, but as instincts that we possess by virtue of being human. We are constituted in such a way that we cannot help but implement them as part of the way we operate in the world as human beings. If we treat these as beliefs they can have no place in our belief-forming structures, but as instincts they can take their place in these structures as universal, constant features of human nature from which all beliefs arise. These instincts cannot simply be ignored by philosophers and discarded as nothing more the “some psychological quirks of our constitution” (Popkin 1966, p. 64). There exists an interconnected relationship between epistemic norms and our cognitive abilities. There is a real sense in which a description of our cognitive abilities is essential to establishing any genuine epistemic norms9. Hume's natural instincts are universal, unavoidable features of how we operate in the world; as such, any account of appropriate epistemic norms must incorporate them. As 9

See Brown (1996) for a discussion of the ties between epistemic norms and cognitive abilities.

72


specific features of human beings natural instincts have determined how we pursue the epistemologist's tasks. If we attempt to construct norms in a vacuum, disregarding such permanent and irresistible aspects of how we engage in the world, then the norms we establish will have no legitimate force for human beings.

4 Conclusion Quine believed that the “Humean predicament is the human predicament” (Quine 1969, p. 72) and Hume’s solution is very much a human solution grounded in the realities of the human species. Sometimes we have a tendency to overlook the most obvious of points, so I think it worth recalling the titles of Hume's two major philosophical works; “A Treatise of Human Nature”, and “An Enquiry Concerning Human Understanding”. The emphases are of course mine; I use them to highlight that in all his investigations Hume is dealing with human beings, the limitations of which can never be abstracted from, for they delineate the very scope and form of his investigation. Hume's avowed purpose is to provide an account of human knowledge and of how humans acquire such knowledge. Hume's is not an attempt to grapple with the nature of reality as a whole but with the operation and limitations of human nature. The term “natural belief” totally fails to capture the key features of phenomena described by Hume. What Hume has revealed are not beliefs but universal, unavoidable instincts in accordance with which all experience is processed. It is only in coming to see these phenomena as instincts that a muchneeded humanisation of our understanding of reason, knowledge, and justification can begin.

Bibliography Brown, H. I. (1996). “Psychology, Naturalized Epistemology, and Rationality”, in W.T. O’Donohueand R.F. Kitchener, (eds.), The Philosophy of Psychology. London: Sage. Butler, R. J. (1960). “Natural Belief And The Enigma Of Hume”, in Archiv Für Geschichte Der Philosophie 40: 71-100 . Garret, D. (1997). Cognition and Commitment In Hume's Philosophy. New York: Oxford University Press. Gaskin, J.C.A. (1974). “God, Hume And Natural Belief”, in Philosophy 49: 281294. Hume, D. (1985). A Treatise Of Human Nature. Harmondsworth: Penguin. ——— (1988). An Enquiry Concerning Human Understanding. Chicago: Open Court. Kemp Smith, N. (1905). “The Naturalism of Hume (I)”, in Mind 14: 149-173. ——— (1941). The Philosophy Of David Hume. New York: Garland.


73

McCormick, M. (1993). “Hume On Natural Belief And Original Principles”, in Hume Studies 19:103-116. Pavlov, I. P. (1927). Conditioned Reflexes. London: Routledge & Kegan Paul. Popkin, R. H. (1966). “David Hume: His Pyrrhonism and His Critique of Pyrrhonism”, in V.C. Chappell (ed.), Hume. London: Macmillan. Quine, W.V. (1969). Ontological Relativity and Other Essays. New York: Columbia University Press. Strawson, G. (1989). The Secret Connexion. Oxford: Clarendon. Stroud, B. (1978). Hume. London: Routledge & Kegan Paul. Wolff, R. P. (1966). “Hume's Theory of Mental Activity”, in V.C. Chappell (ed.), Hume. New York: Anchor.

CHAPTER FOUR CAN CONSCIOUS ATTENTION RESTORE UNCRITICAL REALISM? AN EXAMINATION OF A NEO-RUSSELLIAN VIEW OF REFERENCE

GLORIA AYOB

John Campbell has recently suggested that we ought to preserve the “commonsense picture” of reference, on which conscious experience of an object explains our capacity to grasp demonstrative concepts that refer to it.1 In order for experience of objects to play this explanatory role, Campbell argues that experience has to be a pre-reflective relation to objects. This suggestion echoes Russell’s Principle of Acquaintance: in the case of simple unanalysable referring terms, a pre-reflective acquaintance with objects explains our capacity to understand referring terms. On the Russellian view, experience cannot explain our grasp of referring terms if it isn’t a pre-reflective relation to objects. Wittgenstein took issue with this claim in his later thought. Very roughly, Wittgenstein came to realise that we understand referring terms because we can make sense of the activity of referring to the objects we encounter in experience. But the act of pointing to an object and meaning something by pointing to it draws on a whole repertoire of concepts. So if experience of objects is taken to be a prereflective relation then the worry is that it fails to explain our ability to grasp demonstrative concepts. 1

I would like to thank my respondent Tom Roberts for providing extremely helpful comments on this paper. I would also like to thank the audience at Mind2005 in Edinburgh for raising challenging and stimulating questions in discussion. In addition, I would also like to thank my colleagues at the University of Warwick for reading and commenting on earlier drafts. Thanks especially to Tim Thornton and Michael Luntley.


75

In light of this Wittgensteinian worry, it is interesting that Campbell seeks to revive the Russellian view of semantics. Campbell thinks that he has the resources to meet the Wittgensteinian objection to the Russellian view by appealing to the notion of conscious attention. Conscious attention to an object is the same relation as Russell’s pre-reflective acquaintance with the object, but there is one important element added: conscious attention is not only linked upwards to conceptual capacities, but it is also linked downwards to underlying information-processing mechanisms. This downward link to information-processing mechanisms, Campbell thinks, is what overcomes the problems that Russell had in the wake of Wittgenstein’s objection. This downward link to information-processing mechanisms supposedly restores uncritical realism, i.e. the view that the nature of the object fixes the correct application of the concepts that we apply to the object. (Pears 1987, p. 29). In what follows I want to evaluate this response. I begin by identifying the specific concern in Wittgenstein’s later thought about the Russellian view of reference that I am interested in. Wittgenstein is concerned about what is required for understanding the uses of demonstrative terms. Russell and Campbell think that simple demonstratives such as “This” and “That” are meaningful terms that constitute the most primitive explanatory ground in terms of which we make sense of our capacity for language and intentionality more generally. For Wittgenstein, this is problematic because it raises the following question: how does pre-reflective attention to objects fix the rule under which a demonstrative term is understood to relate to the objects that we consciously attend to? In the absence of a language practice that fixes the grammar of demonstrative terms, Wittgenstein argues that pre-reflective attention to objects cannot fix the grammar of demonstrative terms and that we end up with the rule-following paradox. I then look at a claim that Campbell makes which may be understood as a response to this concern. The claim is that we do not need to employ sortal classifications in order to consciously attend to an object and to thereby have an understanding of demonstrative concepts that refer to it. I argue that this response does not actually address Wittgenstein’s concern because Wittgenstein is not concerned with the metaphysical problem of the individuation of objects, but with the grammatical problem of how we are able to refer to objects in the absence of a language practice. Next, I go on to clarify a confusion about the rule-following paradox. Finally, I conclude that if Campbell’s appeal to the liaisons between conscious attention and information-processing mechanisms is meant to address the Wittgensteinian worry I have identified, then this threatens to undercut our normative practices, leaving us with a causal model of concept use. On the other hand, Campbell might argue that his appeal to information-processing mechanisms is not meant to address this

76


worry. If this is the case, however, then it is far from clear how he has managed to reinstate a neo-Russellian view of semantics that addresses Wittgenstein’s worry.

1 Identifying Wittgenstein’s worry about the Russellian view of reference Campbell thinks that the Russellian view preserves “the common-sense picture” on which our experience of objects controls the pattern of use of a demonstrative term, and that we should not lose sight of this common-sense picture when thinking about how experience explains our capacity for grasping demonstrative concepts (Campbell 2002, p. 4). But if this move to preserve the common-sense picture is to work, then Wittgenstein’s worry about how a pre-reflective relation to objects can explain our capacity to refer needs to be addressed. To bring the difference between the Russellian view and the Wittgensteinian view that I am interested in into sharper focus, I want to consider the question of how primitive the referential relation can be if it is to be a relation in terms of which we understand our capacity for meaning and intentionality. According to Russell, simple demonstratives such as a bare “This” are meaningful. We understand the meaning of a simple demonstrative by being acquainted with the object to which the simple demonstrative refers. Russell’s interest in the acquaintance relation lies in its explanatory role. On the most basic level, an expression can be intelligible if the subject is acquainted with the thing denoted by the word. The idea that the acquaintance relation has an explanatory role to play in a subject’s ability to grasp words and expressions rests on the following model of language: “Words all have meaning, in the simple sense that they are symbols which stand for something other than themselves.” (Russell 1992, p. 47.)

In order for the explanation of our capacity to understand words in terms of our acquaintance with objects to be informative, our acquaintance with these objects (which gives us an understanding of these terms) has to be non-intentional. Only if this acquaintance is non-intentional can it explain, in non-circular terms, our capacity for intentionality, i.e. our capacity to use concepts that are about a world that is external to our language. So, on this view of meaning, we first understand words by being acquainted with the bare objects to which these words refer. Understanding propositions follows from this acquaintance with the object. Acquaintance with an object is thus free of content: in being acquainted with an object, I am seeing the object independently of any propositions involving that object. Russell states:


77

“It would be an error to suppose that “this” means “the object to which I am now attending”. “This” is a proper name applied to the object to which I am now attending. If it be asked how I come to select this object, the answer is that, by hypothesis, I am selecting it, since it is the object of my attention. “This” is not waiting to be defined by the property of being given, but is given; first it is actually given, and then reflection shows that it is “that which is given”.” (Russell 1956, p. 168.)

By speaking of “an actually given” object, I take it that Russell construes attention to objects as the upshot of a (causal?) process, where the object is given to the subject passively, i.e. without the subject’s undertaking to do anything. Furthermore, because this attention to objects is not propositionally structured, the subject attends to bare objects. For Russell, it is this non-intentional, pre-reflective acquaintance with something external to language that enables us to understand the simplest terms in our language, i.e. simple demonstratives such as “This”. Thus, on Russell’s view, the objects referred to by simple demonstratives are given to us in experience non-intentionally. In so far as the objects referred to are given non-intentionally, our grasp of these referring terms is primitive and nonintentional. However, from the Wittgensteinian perspective, even if attention to objects is the upshot of a process (e.g. a causal process) that does not involve our capacity for intentionality, this by itself says nothing about our ability to understand demonstrative terms. This is because the referring role of demonstrative terms is not, and cannot be, given to the subject pre-reflectively. In order to appreciate the referring role that demonstrative terms play, a subject has to draw on her conceptual capacities. In the context of the present discussion, Wittgenstein’s argument for this claim may be formulated in the following way: referring to something is a move within a language-practice. In the absence of a language-practice, we cannot refer to objects. What makes an utterance of a demonstrative term a reference to an object is the fact that we use the term in a particular way, a way that is determined by its position in our language-practice. For instance, we refer to an object when we describe it, or when we give an order that involves the object referred to, or when we ask a question about the object. These are all moves within a language-practice and, in all these cases, what fixes the referring role of a demonstrative term is the position at which we station the term.2 In §49 of the Philosophical Investigations, Wittgenstein states: [Referring] and describing do not stand on the same level: [referring] is a preparation for description. [Referring] is so far not a move in the language-game—any more 2

Wittgenstein discusses the need for stationing in §29 and §257 of the Philosophical Investigations.

78

Content, Consciousness and Perception than putting a piece in its place on the board is a move in chess. We may say: nothing has so far been done, when a thing has been [referred to]. It has not even [been referred to] except in the language-game. This is what Frege meant too, when he said that word had meaning only as part of a sentence. (Wittgenstein 1953, p. 24)3

Wittgenstein is raising the problem of what is required for understanding the use of demonstrative terms. He observes that demonstrative terms need to be positioned or stationed as referring devices if they are to be understood, and that this stationing presupposes “a great deal of stage-setting in the language” because it is language that gives a place to referring devices. The act of referring to things does not make sense in a vacuum; in being able to refer to things, we presuppose the existence of the grammar of referring devices, e.g. demonstrative terms.4 And language provides the context that supports the grammar of referring devices. So, even if there is a sense in which our attention to objects is the upshot of a causal process as some theorists argue, we cannot conclude from this that our capacity to understand demonstrative terms consists in non-intentionally attending to objects. To say this would be to conflate the understanding of demonstrative terms with a (causal?) process that accompanies the use of demonstrative terms regularly. Our capacity to understand demonstrative terms draws on our mastery of a conceptual repertoire within which the role of referring devices is fixed. It is the mastery of a conceptual repertoire that fixes the referring role of demonstrative terms, rather than the pre-reflective acquaintance with an object that is the upshot of a causal process. Furthermore, it is only once the role of demonstrative terms as referring devices is fixed that we can grasp the meaning of demonstrative terms. In the usual case, we also need to perceive the object referred to by a demonstrative term in order to know which object it is that the demonstrative term refers to. This is a point that a Wittgensteinian can agree on. But the key difference is that, for the Wittgensteinian, this perceptual link with the object is conceptually informed and embedded within a language practice. Otherwise we would not be able to understand the demonstrative term that refers to the object we currently perceive. The problem with Russell’s pre-reflective attention to objects (construed as the upshot of a causal process) is that it cannot support the “stationing” required for reference to be so much as possible. This “stationing” requires us to exercise our conceptual capacities. Our capacity for intentionality is what helps to support the “stationing” of a demonstrative term as a referring term. Hence referring to objects cannot be so primitive as to exclude our capacity for intentionality. By claiming that acquaintance with an object is non-intentional, Russell removes the support required to “station” the demonstrative term as a referring term, and thereby removes the meaningfulness of the demonstrative term altogether. 3 4

I have replaced “naming” with “referring”, to adjust this passage to the present discussion. Argument taken from §257 of the Philosophical Investigations.


79

The upshot of Wittgenstein’s argument is that simple demonstratives cannot be so much as meaningful, if our understanding of these demonstratives consists in being acquainted pre-reflectively with bare objects. From this it follows that if the referential relation is as primitive as Russell takes it to be, then it cannot be a relation in terms of which we understand our capacity for intentionality.

2 Campbell’s response: Conscious attention to an object does not require sortal classification John Campbell seems to think that the worry I have just outlined arises because philosophers such as Wittgenstein fail to appreciate the richness our pre-reflective perceptual relation to objects. As Campbell would read it, we would only be inclined to think that referring to objects needs to be “stationed” by a language practice if we think that the primitive perceptual relation between a subject and the objects she perceives cannot do this job and thereby fails to deliver knowledge of reference, i.e. the knowledge that is given by being acquainted pre-reflectively with the object. If it can be shown that being perceptually related primitively does in fact deliver knowledge of reference then, Campbell claims, we can say that knowledge of reference is pre-reflective and does not depend on the subject’s exercise of conceptual capacities. In this vein, a rejection of the meaningfulness of simple demonstratives (whether it be Wittgenstein’s rejection, or Quine’s rejection) is taken to be a claim that we need sortal classification in order to situate the use of a demonstrative term, i.e. to attend to the object that is being referred to by the demonstrative, and thereby to have knowledge of reference. Based on this reading of Wittgenstein’s concern, Campbell’s Russellian defence against Wittgenstein’s concern takes the following shape: the “stationing” required for reference to be possible occurs in something more primitive than language use, namely in the information-processing mechanisms that underpin the perceptual relation between subjects and objects. Campbell’s key claim is this: in the case of everyday objects of perception such as trees and tables, we do not need to be able to classify them in order to be able to understand demonstrative concepts that refer to them. All we need is to consciously attend to the object in question. This prereflective conscious attention should give us an understanding of demonstrative concepts that refer to it. So, the use of a demonstrative term is situated by the prereflective perceptual relation to objects, and not by a language practice. If we nonetheless insist that sortal classification is required to consciously attend to the objects of perception, then, according to Campbell, this is because we assume ontological relativism. The only reason why we might think that “Gavagai” may be translated as “a temporal stage of a rabbit’s life” or as “rabbit part”, rather than as “rabbit”, is that we are sceptical about the thesis that the fundamental point of contact between language and the world is between the demonstrative and the

80


conscious attention to the object. (Campbell 2002, p. 228). However, once we understand the liaisons between conscious attention and the underlying information-processing mechanisms, we see that ontological relativism is not a plausible theoretical option. As it turns out, the perceptual system uses location together with Gestalt organisation principles to organise the world. Location is used by the perceptual system as a binding parameter. A binding parameter is the characteristic of the object that the visual system treats as distinctive of that object, and uses in binding together features as features of that thing. (Campbell 2002, p. 37). In short, what the visual system does is to bind the features at one location as features belonging to the same object. Since an object always occupies only one location at any given moment, this means that the perceptual system picks up objects long before we even grasp the concept of an object. Because our perceptual systems pick up objects long before we exercise our conceptual capacities, the referential relation between demonstrative concepts and objects is not simply a function of our (supposedly arbitrary) conceptual scheme. The referential relation is basic even prior to the use of concepts and, most importantly, it can explain our grasp of demonstrative concepts because it is basic in this way. If, however, this explanation is to do justice not merely to the use of demonstrative concepts as the upshot of a causal perceptual process but also as a normative undertaking, then such an explanation has to include an account of how the link between conscious attention and information-processing mechanisms can be a justificatory link. Campbell thinks that we can speak of a justificatory connection holding between conscious attention and information-processing mechanisms because, by consciously focusing on an object, we define the objective of the computational procedures that kick into play at the information-processing level. The idea, very roughly, runs as follows: While it is true that we may not be aware of which computational procedures are being used, Campbell argues that conscious attention can be said to justify the use of the computational procedure because we set the target for our information-processing mechanisms.5 Since we

5

The information-processing mechanisms are subpersonal mechanisms that cause our conscious experiences. It is somewhat unusual to speak of justification at this level, but Campbell thinks we can use the term “justification” here in the following way: “I myself may not be aware of which computational procedures are being used; but it is my conscious attention to the object that defines the objective of the computation. It is in that sense that conscious attention justifies the use of the computational procedure: the use of computational procedure can be justified or criticized by reference to the objective defined for the procedure by conscious attention.” (Campbell 2002, p. 26, my emphasis). Justified according the neurologist scanning a brain, perhaps, but I am not entirely convinced that we can shift from the neurologist’s viewpoint to the conscious subject’s viewpoint and


81

define the target in this way, “the use of the computational procedure can be justified or criticised by reference to the objective defined for the procedure by conscious attention.” (Campbell 2002, p. 26). So what we have here is an appeal to findings in cognitive science about how the perceptual system works to show that ontological relativism is not a plausible theoretical option. Even prior to the employment of concepts, objects are absolutely basic because the perceptual system uses location as a fundamental parameter in organising the world. Because we set the target for computational procedures by consciously attending to objects, we can appeal to the liaisons between conscious attention and information-processing mechanisms to give a full-blooded explanation of our grasp of demonstrative concepts. Conscious attention to objects both causes and justifies our use of demonstrative concepts and for this reason conscious attention explains our ability to grasp demonstrative concepts. The question to which I now turn is: does this constitute a satisfactory response to the Wittgensteinian worry about the Russellian view of reference?

3 Refining the Wittgensteinian worry about the Russellian prereflective view To argue that we do not need sortal classification in order to understand a demonstrative concept on the basis of experience is, I think, to miss the point of Wittgenstein’s fundamental worry about the “stationing” of referring devices on the Russellian view of reference. In order to try to identify the exact point of Wittgenstein’s worry, I want to make a distinction between two claims: (a)

The sortalist’s claim: You need to know what kind of object you are referring to, in order to be able to refer to it.6

(b)

Wittgenstein’s claim: You need to master the technique of referring in order to be able to refer to an object.

Both claims are similar in a superficial respect: They both say that knowledge of reference cannot be acquired independently of surrounding knowledge. In the sortalist’s case, this surrounding knowledge is given by the sortal concept F. When placed together with a simple demonstrative, the result is a complex demonstrative, “That F”. According to the sortalist, the most primitive unit of significance is the still speak meaningfully of “justification” in regard to information-processing procedures. I would like to thank the editors for highlighting this tricky question. 6 Campbell thinks that David Wiggins holds this sort of view. See Campbell (2002, pp. 7275) for Campbell’s discussion of Wiggins.

82


complex demonstrative. Simple demonstratives, on their own, are not meaningful. Like the sortalist, Wittgenstein also claims that knowledge of reference cannot be acquired independently of surrounding knowledge. This surrounding knowledge is constituted by the mastery of a language, within which the technique of referring has a use. So, according to Wittgenstein, a simple demonstrative cannot be meaningful on its own if a grasp of its meaning involves a pre-reflective acquaintance with the object referred to. The subject has to exercise her mastery of a language in order to grasp the meaning of a demonstrative term. But Wittgenstein and the sortalist have very different reasons for claiming that knowledge of reference cannot be acquired independently of surrounding knowledge. The sortalist’s claim is essentially a metaphysical claim: we need surrounding knowledge, i.e. knowledge of sortals, in order to individuate the object that is referred to by the demonstrative term—and it is, of course, only possible to understand the demonstrative term if we have individuated the object referred to.7 So, for example, when somebody points to a chess piece and says “That!”, we need to know whether he means, “That pawn!”, or “That bit of wood!”, in order to grasp the meaning of the demonstrative concept. In response to the sortalist’s reason for rejecting the Russellian view of reference, Campbell’s appeal to the liaisons between conscious attention and information-processing mechanisms might have some bite. On Campbell’s view, understanding demonstrative reference consists in consciously attending to an object—and this attention is the upshot of a subpersonal process that involves binding the features found in the same location as features of a particular object. So, our attention to objects delivers individuated objects even before we apply concepts. This means that, although we might get it wrong about what sort of object is being picked out by a demonstrative concept, we cannot be wrong about which object is picked out. Campbell states this point unequivocally: “Demonstrative reference, on the view I am recommending, does not intrinsically require that you get it right about the sort of the object.” (Campbell 2002, p. 78.)

What the liaisons between conscious attention and information-processing mechanisms reveal is that sortals do not play the constitutive role that sortalists take them to. Rather, sortals are merely an “external causal factor in the act of conscious attention.” (Campbell 2002, p. 77.) By this Campbell means that sortals at most help us to orient our attention, but sortals do not individuate the objects that make understanding demonstrative reference possible. That is done by the informationprocessing mechanisms in the visual system at the subpersonal level. 7

Campbell calls this The Delineation Thesis: “Conscious attention to an object has to be focused by the use of a sortal concept which delineates the boundaries of the object to which you are attending.” –Reference and Consciousness, p.69.


83

But Campbell’s response, although adequate for the sortalist’s anxieties perhaps, does not address Wittgenstein’s worry as I raised it in Section II above, because Wittgenstein’s worry is not a metaphysical worry about the individuation of objects. Even if the perceptual system does individuate objects for us prior to the employment of conceptual capacities, Wittgenstein’s worry is a grammatical worry that still stands: reference only makes sense when it is positioned within a language practice, so acquaintance with objects which is primitive and does not involve our capacity for using concepts cannot give us a grasp of demonstrative terms. This is because a primitive, pre-reflective acquaintance with objects does not fix the referring role that demonstrative terms play. A primitive acquaintance with objects does not bestow upon demonstrative terms any grammar at all, be it referential or otherwise. In order to grasp that demonstrative terms refer to objects, we need to be capable of stationing the referential role that demonstrative terms play within a context that can support this role. It seems that only a language practice can provide this support. Therefore, in order to grasp the meaning of a demonstrative term, we have to be exercising our conceptual capacities. So why does Wittgenstein think that a primitive acquaintance with objects cannot bestow upon demonstrative terms any grammar at all? Consider for a moment a case where somebody is trying to describe to her housemate a paint colour that she saw earlier in a catalogue, which she thinks would go nicely on their living room walls. The housemate asks her what this colour looked like, and she points to their kitchen table top and says, “Like that.” In this case, the act of pointing and the utterance of the word “that” is primarily being used to describe something (namely, the paint colour she saw in the catalogue), rather than to refer to something (i.e. the kitchen table top). If the housemate only grasped the referring role of demonstrative terms, she would completely miss the point of the use of the word “that” in this case—namely, that it describes a colour that might suit their living room. In this simple case, we see that what appears to be a demonstrative term can be used in more than one way: although in the standard case it is used to refer to objects, it can in some cases be used to describe something. What this shows is that there could, in principle, be various ways in which a demonstrative term may be used. Hence the grammar of a demonstrative term needs to be fixed if we are to grasp the meaning of the term. The problem with Campbell’s suggestion is this: merely by attending to an object pre-reflectively, we do not, and cannot, know how the demonstrative term is related to the object because we do not know under what rule the demonstrative term is being related to the object. It may be related referentially, but it doesn’t have to be (as the brief example above illustrates). There does not seem to be anything about attending to an object which forces a demonstrative term to be used referentially.

84


In fact, this is already stretching things a bit further than Wittgenstein himself would probably allow: it seems most plausible to say that what fixes the grammar of a demonstrative term is the context in which the term is being used. This context is made available by a language practice that subjects draw on when they grasp the meaning of demonstrative terms. Without a language practice, it seems far from clear how there can be any rules at all that fix how a demonstrative term relates to an object. In the absence of a language practice the act of pointing, for instance, does not make any sense because there are no rules in terms of which we can understand the act of pointing. One suggestion to overcome this problem might be to say that we could interpret the act of pointing as meaning that the subject is referring to an object. But this suggestion results in the rule-following paradox: because the act of pointing (or simply consciously attending to an object) is, on the Russellian view, pre-reflective and does not draw upon the resources of a language practice, there is nothing to compel this interpretation. Interpreting the act of pointing under another rule (for instance, that conscious attention to an object relates a demonstrative term to the object descriptively), we may just as well conclude that the act of pointing does not mean that the subject is referring to an object. This is the rule-following paradox: a content-free action, or state, is semantically inert and therefore stands in need of interpretation. But because an act of pointing (or consciously attending to an object) can be variously interpreted, this deprives the act of pointing of any meaning since the act of pointing can be made out to accord with rules that may be inconsistent with each other. The rule-following paradox arises here because, if conscious attention to an object is free of content, then there is nowhere to position the demonstrative term since the grammar of the demonstrative term (as a referring device, for instance) does not in fact exist in the absence of a language practice. The only way to avoid the paradox is to say that the act of pointing, or simply consciously attending to an object, does not have to be understood under an interpretation. Consciously attending to an object can be non-inferentially meaningful if it is embedded within a language practice, i.e. a practice that a subject draws upon when she consciously attends to objects. This is why Wittgenstein says: “To understand a sentence means to understand a language. To understand a language means to be master of a technique.” (Wittgenstein 1953, pp. 80-81.)

To understand a sentence involving a demonstrative term means to understand a language. This in turn presupposes a mastery of a technique. Having mastered the technique of referring, and various other techniques, the subject is capable of stationing uses of a demonstrative term and thereby is capable of grasping the meaning of the demonstrative term.


85

4 A problem for Campbell’s neo-Russellian semantics As I have identified Wittgenstein’s primary concern with the Russellian view of reference, the problem is to state how pre-reflective attention to objects fixes the rule under which a demonstrative term is understood to relate to the objects that we consciously attend to. In order to provide a plausible defence of the neo-Russellian view that he is expounding, I think Campbell must address this concern. First, I want to point out that Campbell’s reading of the rule-following paradox is shaped by an assumption that he foists upon Wittgenstein which I believe Wittgenstein himself does not make. If we take away this assumption, I am doubtful as to whether Campbell would still be able to meet Wittgenstein’s worry. I shall try to show below that, given the resources within Campbell’s neo-Russellian view of semantics, Campbell does not adequately address the Wittgensteinian worry as I have raised it. So what is it that Campbell thinks Wittgenstein’s objection to a Russellian semantics is? On Wittgenstein, Campbell says the following: “On the common-sense picture, your knowledge of reference controls the pattern of use that you make of the term. … In the later Wittgenstein and in Quine, the problem is that they think the common-sense picture cannot be sustained. There is only the pattern of use: there is no such thing as a knowledge of reference which controls the pattern of use, and to which the pattern of use is responsible. In the later Wittgenstein, the form the resulting problem takes is that the pattern of use now seems arbitrary, since it is no longer thought of as controlled by knowledge of reference. This is the issue he confronts in his discussion of rule-following.” (Campbell 2002, p. 4.)

What this passage shows is that Campbell takes the Wittgensteinian rulefollowing objection to rest on the assumption that there is no link between experience of an object (what he calls “knowledge of reference”) and the use of demonstrative terms referring to it. Hence Campbell takes his task of defending a neo-Russellian semantics to consist in showing that Wittgenstein is wrong: there is a link (which is both causal and justificatory) between experience of objects and concept-use, and this link is established by the fact that there are liaisons between experience and information-processing mechanisms (Campbell 2002, p. 5). As I briefly mentioned earlier, the introduction of these liaisons is meant to show that our uptake of objects in perception is absolutely basic and occurs prior to the employment of a conceptual scheme. Because of its priority, this uptake explains why we find the referential relation so fundamental in our engagement with the world, and why our use of demonstrative terms is constrained by the object itself. But in fact Wittgenstein makes no such assumption: The rule-following paradox does not depend on the presupposition that “there is no such thing as a knowledge

86


of reference which controls the pattern of use and to which the pattern of use is responsible.” Rather, as I read him, Wittgenstein’s point is that the rule under which a demonstrative term is understood to relate to our experience of objects (and thereby the rule under which knowledge of reference is acquired) is provided by a language practice. The embedding of demonstrative terms within a language practice entails that our grasp of demonstrative terms cannot consist in a prereflective relation to objects that delivers a non-circular explanation of our capacity for intentionality. So, accepting that Wittgenstein does not deny that there can be a link between knowledge of reference and the pattern of use, I think that Campbell cannot adequately address the Wittgensteinian worry as I have raised it. Suppose we do say that Campbell’s appeal to information-processing mechanisms is meant to address the Wittgensteinian problem I have identified. The suggestion, then, has to be this: the liaisons between conscious attention and information-processing mechanisms fix the rule under which a demonstrative term is taken to relate to the objects that we consciously attend to, namely that demonstrative terms refer to these objects. But this seems to amount to a strange form of reliabilism, a reliabilism about the rules that fix the grammar of demonstrative concepts. I shall call this “concept reliabilism”, because what is fixed by the link between conscious attention and information-processing is the rule that fixes the meaning of demonstrative concepts. Concept reliabilism runs contrary to our actual justificatory practices: in our ordinary semantic justificatory practices, we undertake a responsibility for using a demonstrative term referentially. It is not something that just happens to us. If the referential role of demonstrative terms were something that we were smacked by, so to speak, then how could we speak of justification at all? A causal model would be more appropriate in explaining what happens when we use demonstrative concepts on the basis of our experience of objects. In fact Campbell thinks that there is a causal link here as well, but it is difficult to see how he could secure the justificatory link without collapsing into a form of concept reliabilism. On the other hand, Campbell might argue that his appeal to informationprocessing mechanisms is not meant to address the problem of how pre-reflective attention to objects fixes the rule under which a demonstrative term is understood to relate to the objects that we consciously attend to. Perhaps there are further conditions that need to be met if such a rule is to be fixed. If this is the case, however, then it is far from clear how he has managed to reinstate a neo-Russellian view of semantics that addresses Wittgenstein’s worry about how a primitive, prereflective attention to an object could possibly explain (by causing and justifying) our grasp of demonstrative concepts as terms that refer to objects. Yes, it is true that there may be a reliable causal link between the object and the subject’s underlying information-processing mechanisms which results in a sentient


87

understanding of the object at the level of conscious attention. Because it is a reliable link, we may—if we adopt an externalist view of justification—speak of the particular information-processing streams that kick into action as being justified for doing so because the target object has certain properties. But if we reject concept reliabilism, then the normative link between conscious attention and concept-use has yet to be fixed. A person who has not yet acquired the technique of referring cannot be said to have an experience of an object that delivers an understanding of demonstrative terms that refer to it. To sum up the problem here: Campbell does not seem to have provided us with an adequate response to Wittgenstein’s worry. Wittgenstein raises the question of why pre-reflective attention to an object fixes the grammar of demonstrative terms. Now, if Campbell’s appeal to the liaisons between conscious attention and information-processing mechanisms is meant to address this question, then I have tried to show that we are left with a form of concept reliabilism. This makes it difficult to see how our language practice can be normative, instead of simply constituting a causal process. On the other hand, if Campbell is not appealing to the liaisons between conscious attention and information-processing mechanisms to address Wittgenstein’s question, then it is far from clear how pre-reflectively attending to an object can explain (by justifying) the use of demonstrative concepts that apply to the object. There may in fact be a reliable link between conscious attention to an object and underlying information-processing mechanisms, but this does nothing to fix the normative link between conscious attention and conceptuse. It looks as though there are in fact two distinct notions of justification at work here: first, there is a justificatory link understood in reliabilist terms that holds between information-processing mechanisms and conscious attention, then there is a normative link which involves the subject’s conscious engagement that holds between conscious attention and concept-use. If it turns out there are in fact two very different notions of justification at work here, then I think that Wittgenstein’s worry about the Russellian view of reference remains unaddressed, namely: how can experience of an object, if it is pre-reflective, fix the grammar of demonstrative concepts?

Bibliography Campbell, J. (2002). Reference and Consciousness. New York: Oxford University Press. ——— (2004). “Reference as Attention”, in Philosophical Studies 120: 265-276. Pears, D. (1987). The False Prison, Vol. 1. Oxford: Oxford University Press. Russell, B. (1992). Principles of Mathematics. London: Routledge. ——— (1956). “On the Nature of Acquaintance”, in his Logic and Knowledge: Essays, 1901-1950. London: Routledge.

88


Wittgenstein, L. (1953). Philosophical Investigations. Oxford: Basil Blackwell.

CHAPTER FIVE ON PEACOCKE'S THEORY OF CONCEPTS

ASUNCIÓN ÁLVAREZ

How are we to understand the notion of concept, the very concept of concept itself?1 One natural way, it seems to me, is to take Fregean sense as a model, and impose similar constraints on a theory of concepts. This approach has the advantage, among others, of allowing for a distinction to be made between publicly shared, objective concepts, on the one hand, and private, subjective mental representations on the other—a distinction which I believe is desirable for various reasons. One problem with Frege's account, however, is that he regarded senses as Platonic abstract objects belonging in the Third Realm, which he described as a non-physical and non-mental plane of existence. Besides the obvious metaphysical issues which this claim raises, there is also the question of how thinkers (who can usually be described as physical and/or mental entities) can come to grasp senses if senses are neither physical nor mental. One way to avoid these entanglements is Dummett's strategy—via Wittgenstein—of defining meaning in terms of use, thus steering clear of reified senses. A theory of meaning would thus be a theory of understanding: the meaning of a term being defined by what understanding that term involves. Translated into mental terms, the strategy would be to individuate concepts (and indirectly the thoughts they compose) in terms of the thinker's abilities.

1

I would like to thank Christopher Peacocke, who very kindly provided comments on this paper during and after the Mind2005 Conference at the University of Edinburgh. I am also grateful to David Galloway, David Papineau, Mark Sainsbury, and Mark Textor at King's College London for their feedback; as well as to audiences in Edinburgh and at the University of London June 2005 Graduate Conference in Philosophy.

90


Peacocke's theory of concepts is one of the most recent versions of this approach, and possibly the most influential one. What follows is an attempt to summarise my reading (or misreading) of Peacocke's theory of concepts, as well as to forward some of the reasons why I am unhappy with it. Generally speaking, I have found myself coming up with questions which fall under either one of two categories. On the one hand, there is a cluster of issues related to Peacocke's characterisation of judgment as truth-seeking and referential. The other set of worries concerns the notion of implicit conceptions, a more recent development in his theory, which I find troublesome in a different way. The structure of this paper follows the drift, so to speak, of these worries. Thus in the first section, I will sketch out, very broadly, the outline of Peacocke's theory of concepts as presented in his A Study of Concepts (Peacocke 1992; hereafter SoC). In the second section, I will touch on some points related to Peacocke's characterisation of concepts as eminently referential, which I find problematic. More specifically, I will question Peacocke's claim that the possession conditions which define a genuine concept must constitute a coherent set, or else the alleged concept will turn out to be spurious. And finally, in the third section, I will examine and discuss implicit conceptions.

1 The theory As mentioned above, Peacocke's theory, as put forth in SoC, aims to provide an account of concepts which can take advantage of the explanatory power of Frege's Sinn/Bedeutung distinction, while avoiding its metaphysical and epistemic pitfalls. Peacocke's way of doing this, like other Neo-Fregeans', is to define thought in terms of the thinker's abilities. Nonetheless, unlike other Neo-Fregean accounts of contents (such as Geach's2) which do not regard concepts as abstract objects but as mental (and thus subjective and psychological) capacities, Peacocke's theory takes concepts at their Fregean face value—i.e. as objective, mind-independent, nonpsychological entities. In this way, in Peacocke's theory, concepts are not identified straightaway with abilities, but rather it's the conditions for the possession of concepts—as distinct from concepts themselves—that are specified in such terms. Thus Peacocke's Principle of Dependence (PD) has it that: “There can be nothing more to the nature of a concept than is determined by a correct account of the capacity of a thinker who has mastered the concept to have propositional attitudes to contents containing that concept (a correct account of "grasping the concept").” (SoC, p. 5.)

2

C.f. for example Geach (1976).


91

Note that the PD plays here a double role, providing an account both of how concepts are individuated and of what it is to possess them. What form must this account of concept individuation/possession take, though? According to Peacocke, such an account should take the following General Form (GF): Concept F is that unique concept C to possess which a thinker must meet condition A(C). (SoC, p. 6)

where A(C) is the condition whereby the concept in question is individuated or possessed. And what form, in turn, must the individuation/possession condition A(C) take? Peacocke's paradigmatic example is that of the concept of conjunction, for the individuation/possession of which he provides the following "transitions": p, q ⊃ pCq⊃ pCq⊃

pCq p q

The individuation/possession conditions for the concept of (classical) logical conjunction would be given by these transitions: that is, in order to possess the concept of conjunction, a thinker would have to find these transitions primitively compelling, where "primitively compelling", Peacocke specifies, means that (a) the thinker must find the transitions compelling (obviously); (b) he must find them originally compelling, i.e. not by inferring them from previous premises or principles; (c) he needn't take the correctness of the transitions as answerable to anything else in order to possess the concept in question. Again, note that the compelling character of the defining transitions is spelt out in terms of belief, and thus of truth: a thinker will find a given transition compelling if, and because, he believes that it is correct, i.e. truth-preserving. A different kind of concept for which Peacocke specifies individuation/possession conditions is the perceptual concept . In this case, the individuation/possession conditions are not specified in terms of transitional forms which the thinker must find primitively compelling, but directly in terms of the thinker's dispositions to belief. These conditions, according to Peacocke, would be the following: a. The subject is willing sincerely to assert some sentence of the form "___ red ___" containing the word "red" (or some translation of it). b. He has some minimal knowledge of the kind of reference it has (e.g., that it is a colour word). c. He defers on his use of the word to members of his linguistic community. (SoC, 29)

92


Thus, the concept would be that unique concept in order to possess which a thinker should be disposed to believe certain completing contents involving —that is, to believe that certain completing contents involving are true. His disposition to belief is further specified by the requirements that he be aware that "red" is a colour word, and that he defer in its use to other speakers of his language So far, so good. But, beyond the specification of a concept's possession/individuation conditions, either in terms of primitively compelling transitions or of dispositions to belief, Peacocke places a further demand on a theory of concepts. Namely, such a theory must provide what he terms a Determination Theory (DT), which should explain how the semantic value of a concept (i.e. its reference) is determined by its possession conditions together with the world. For according to Peacocke, concept possession and individuation are necessarily linked to reference: “the level of reference is inextricably involved with concepts [...] Concepts are individuated by their possession conditions; the possession conditions mention judgments of certain contents containing the concepts; judgment necessarily has truth as one of its aims; and the truth of a content depends on the references of its conceptual constituents. It would be wrong, then, to regard the inferential relations in which concepts stand as grafted onto a structure of concepts that can be elucidated without any reference to reference. Referential relations are implicated in the very nature of judgment and belief.” (SoC, p. 17.)

Peacocke does not bring the DT into his account of concepts only out of “reverence for the classical doctrine that sense, together with the world, determines reference.” (SoC, p.17.) His thesis is stronger than that: namely, that any process involving concepts is by this very fact referential in nature. Not, of course, in the naïve sense that there is an actual referent for every concept; but rather in the sense that the possession conditions which define a concept necessarily involve reference inasmuch as they involve belief and truth. At this point, however, I would like to raise a few questions.

2 Reference One of the most interesting things about the DT is that it serves as a regulatory principle for conceptuality. The possession conditions for a concept, as we have seen, determine, together with the world, the reference of that concept. Should it not be possible to determine what has to be the case for something to be the reference of a concept, then, according to Peacocke, the supposed concept in question is no such thing: it is spurious, a pseudo-concept. Indeed, Peacocke goes so far as to claim that "meeting the possession condition for a concept can be


93

identified with knowing what it is for something to be the concept's semantic value (its reference)" (SoC, p. 22). In this view, to possess a concept is to know what it is for something to be its reference. If a concept cannot be assigned a reference—if it is not possible to describe a state of affairs such that in it the concept in question would have a semantic value—then it isn't a concept at all. Thus, the alleged concept <noon on the Sun> would be spurious because it is not possible to specify what state of affairs should obtain in order for the reference of the concept to be satisfied. For any genuine concept, there must be a set of conditions such that its satisfaction by an arbitrary object would ensure that the concept applied to it: if satisfaction of those conditions is impossible, then the concept defined by the conditions is spurious. A different kind of failure of the DT is that which results from an inconsistent set of possession conditions. This inconsistency can take either one of two forms: a set may contain conditions which are mutually incompatible; or it may be that its conditions, taken as a whole, fail to preserve truth. I will refer to the former form as internal inconsistency, and to the latter as inconsistency with respect to truth preservation (or simply non-truth-preservation). An obvious example of an internally inconsistent set of possession conditions would be one containing both a term p and its negation ¬p as conditions. On the other hand, Peacocke himself gives as an instance of inconsistency with respect to truth preservation the case of Prior's connective tonk, which is defined by a set of non-truth-preserving rules, namely p p tonk q

⊃ ⊃

p tonk q q

would be a spurious concept because its defining possession conditions constitute an inconsistent set with respect to truth preservation: it is not possible to assign a truth function to tonk such that truth is consistently preserved (to wit, notice what happens in the case which makes p true and q false.) Again, this would make a spurious concept and "p tonk q" a pseudo-thought. Can a theory which rejects all inconsistent concepts—that is, all concepts whose defining possession conditions constitute an inconsistent set, either internally or with respect to truth preservation—as spurious provide an adequate account of conceptuality? Consider the concept of (with a capital “G”) as possessed by any average Christian, Muslim, or Jew. The possession conditions for such a concept will presumably include the attribution of omnipotence to God—which immediately leads to such internal contradictions as made manifest by the Stone

94


Paradox.3 Shall we then say that the great monotheistic religions, and their respective theological systems, are based on a pseudo-concept? Maybe. But perhaps more importantly, consider Frege's logical system: despite the internal inconsistency caused by Axiom V (which renders the whole concept of set inconsistent) one can hardly describe it as a barren effort, as the Grundgesetze contains many valuable mathematical insights which are not rendered void by Russell's Paradox. Consider, on a related note, inconsistent set theory. Are we to say that these are mere fabrics of spurious concepts, which can yield no real knowledge? It should be mentioned, at this point, that Peacocke has claimed4 that the internal inconsistency of many concepts poses no problems for his theory: for spurious concepts, in his view, are so because of their unsatisfiability, their inability to have a semantic value. Thus, a concept such as can be a concept despite the inconsistency of its defining possession conditions; however, a pseudo-concept such as is spurious not because its possession conditions are inconsistent— they are not—but because they are not truth-preserving. Nonetheless, one might reply to this that even though not all non-truthpreserving sets of possession conditions are internally inconsistent, all internally inconsistent sets of possession conditions seem to be also non-truth preserving. If it is the preservation of truth that defines real concepts, internally inconsistent concepts will systematically fail to deliver, as an inconsistent set of possession conditions cannot possibly be truth-preserving. Which would, again, seem to indicate that we deal with vast numbers of spurious concepts on a regular basis. Given that pseudo-concepts appear to play a pivotal role in such relevant systems of thought, perhaps it might be worth considering more closely what differentiates genuine concepts from their fake counterparts. Spurious concepts, it appears, can enter into inferential relationships with other concepts or pseudoconcepts, and yield results which make sense to the thinkers who arrive at them. It is thus only the Determination Theory that distinguishes real from fake concepts, and which requires the consistency of the sets of possession conditions, in order to guarantee that truth is preserved and reference can be determined. But is this how we want a theory to characterise concepts—as necessarily linking belief, truth, and reference? Is this the nature of concepts? Imagine, for instance, a logician who, in addition to standard classical logic, masters several non-standard logics and is able to employ them with equal 3

An old theological conundrum that runs as follows. If God is omnipotent, then either He can create a stone which He cannot lift, or He cannot create a stone which He cannot lift. If He cannot create a stone which He cannot lift, then He is not omnipotent. But if He can create a stone which He cannot lift, then He is not omnipotent either. Therefore, if God is omnipotent, then He is not omnipotent. 4 In his reply to this paper in the Mind2005 Conference.


95

competence in the appropriate context. However, certain axioms contained in those logics—such as those which define negation—are mutually incompatible. According to Peacocke's theory, the logician can genuinely master the concept defined by only one of those sets of axioms—namely, the concept defined by the axioms which he finds primitively compelling and in the truth of which he thus believes. But what about that part of his work in which employs other logical systems, and other definitions of negation? I find it hard to agree that all he is doing is engaging in some sort of pretence or simulation with no relevance whatsoever to knowledge. And even if it could be written off as such, the fact would remain that a very large part of human thought processes consists of this sort of "pretence" in which belief and referential truth are, at least temporarily, suspended. A possible answer to this problem might be to argue that a thinker's reasoning capacities are "compartmentalised". That is, his reasoning capacities might be divided into a number of subsets—some mutually exclusive, some overlapping— which would apply or not according to the context. Thus, faced with a situation which involves deploying some non-standard logic, the logician would simply use his non-standard logic module, so to speak, while retaining his standard logic capacities intact, if unused. However, it seems to me that the idea of primitively compelling possession conditions disallows by definition this sort of conceptual skipping around. Take the concept of negation as defined in classical logic. According to Peacocke's theory, if we find its possession conditions primitively compelling, then we believe in them—we believe, for instance, that the formulation "if ¬ ¬p, then p" is true. How can we then blithely switch to a different set of possession conditions which denies that from ¬ ¬p it follows that p? Given that both sets of possession conditions are mutually incompatible, the logician is being asked to change beliefs every time he finishes teaching introductory logic for undergraduates and moves on to his graduate seminar in intuitionistic logic. Which is, I think, quite a stretch (Brouwerlike cases notwithstanding). It could also be argued that a classical logician who honestly believed that from ¬ ¬p there follows p could be able to work on intuitionistic logic by engaging in some sort of conscious make-believe in which he pretended that from ¬ ¬p it does not follow that p. However, by engaging in such pretence he would be able to arrive at inferential conclusions and results, just in the same way as he would via classical logic. So, given that belief and simulation do not here differ significantly in their inferential dynamics, it might be questioned whether the possession conditions which define concepts need be primitively compelling at all. Yet another possible answer might be that the concept of negation as defined in classical logic is a completely different one from the concept of negation as defined in intuitionistic logic. Thus the logician would be able to find their respective possession conditions primitively compelling, on a separate basis. But surely there

96


is something undesirable about simply writing classical and intuitionistic negation off as two distinct, unrelated concepts—for even if they are distinct concepts, there is a relation between them in that they both are varieties of negation, which the theory does not capture as it stands. Moreover, the possession conditions for these two distinct concepts would still be mutually incompatible—which raises the question of the consistency of conceptual systems. In Peacocke's theory, every concept must be internally consistent, that is, there must be no contradiction among its defining possession conditions. But concepts, by their very inferential nature, stand in relations to whole sets of other concepts. Must these sets of concepts, I wonder, be consistent too among themselves? Whatever the case may be, though, even if we define classical negation and intuitionistic negation as distinct concepts, the problem remains of explaining how the logician can stop finding the formulation "if ¬ ¬p, then p" true if he had previously found it primitively compelling. It might be argued, though, that it isn't such a stretch if we assume that belief functions—or at least is grounded—on a subpersonal level. Indeed, Peacocke has gone on to add the notion of subpersonal doxastic implicit conceptions to his theory of concepts as a way of patching up some of the problems caused by his definition of concepts in terms of possession conditions. I have my own misgivings with respect to this move—explaining conceptual thought in terms of subpersonal processes—but I expound them in the next part of this paper. Before moving on, however, I would like to point out that a theory of concepts should account, it seems to me, for the fact that it is possible to entertain mental contents without an immediate commitment as to their truth values, and that such a thought process is not irrelevant to knowledge or rationality. If possessing a concept necessarily entails finding its defining conditions primitively compelling— that is, believing that the formulations of its possession conditions are true—then it shouldn't be possible to ponder, to envisage, or to fantasise. At the very least, if we hold that to possess a concept is to find its defining conditions primitively compelling, then those mental processes—which involve only a sort of temporary "suspension of disbelief" in the concepts involved—should not be categorised as strictly rational, given that they don't involve legitimate concepts. How are we to categorise concepts which are temporarily "indeterminate" in this way? Are they genuine concepts which find themselves under probation, so to speak? Are they spurious concepts? If so, then the possibility of rational yet disengaged and temporarily non-committal intellectual consideration would be precluded, as it would involve non-concepts. Should a third category be established altogether for this kind of concept, then? A further problem is that it is hard to fit concepts defined in terms of primitively compelling conditions with inductive thought in general—which nonetheless seems to play an important role in our mental lives. Related to this question is the fact that


97

the primitively compelling character of possession conditions makes it difficult to explain how new knowledge can be reached from the premises which a thinker already possesses: for it is hard to imagine how a thinker might come to question or doubt or find inadequate the definition of a concept which he masters if its defining conditions already involve his belief in their truth. Peacocke's theory appears to consist of two layers, or stages (which do not necessarily follow a given order). In the first stage, concepts are defined in terms of their possession/individuation conditions: thus giving an account of a particular concept amounts to no more than giving an account of what it is for a thinker to master that concept, of what conditions a thinker must meet in order to posses the concept in question. In the second stage, the Determination Theory is added as an extra constraint to the theory of possession conditions, in order to ensure that concept possession and individuation are linked to reference. Even though Peacocke claims that a DT is an indispensable part of the theory, it seems to be possible to have a theory of concepts which accepts the first stage but not the second one (in fact, that is basically what such theories of conceptual content as Robert Brandom's amount to.) Peacocke's insistence on the DT seems to spring from his conviction that "judgment aims at truth". I take it that by this Peacocke means that the thinker who emits the judgment must believe that his judgment is true—a reading which is supported by Peacocke's previous emphasis on the primitively compelling character of possession conditions. But surely this cannot be enough. Guessing, for instance, can also be said to aim at truth inasmuch as the guesser is trying to "get it right", and surely it is an activity which is ruled by a different set of constraints and norms than that which governs judgment. Peacocke has corrected his dictum5, specifying that it is only rational judgment that aims at truth. But this, I think, is also unsatisfactory, for it might be argued that truth-directedness is neither a sufficient nor a necessary condition for rational judgment. On the one hand, irrational judgment can sometimes be said to aim at referential truth—as in the case of the proverbial madman who earnestly insists (and genuinely believes) that he truly is Napoleon Bonaparte, or Jesus Christ. In this way, truth-directedness would not be exclusive of rational judgment, and thus not a sufficient condition for it. On the other hand, I am not convinced that a rational judgment need be aimed at referential truth at all, in the sense that the thinker must believe that his judgment is correct. I find Italian gelato di lampone delicious, and superior to any other kind of ice-cream in the world. I am also perfectly aware that this is a judgment based on purely subjective grounds, and cannot be extended to other people's tastes. And yet

5

Ibid.

98


I doubt that my judgment, which is entirely based upon my own prejudices, can be described as irrational. It could be argued perhaps that my preference is a truth-directed judgment: after all, it is true that I do find that kind of ice-cream delicious. But it is not my own preferences that I am making a judgment about: I am making a judgment about a state of affairs in the world, which is entirely distinct from the meta-judgment which takes that judgment as its object. The former would have the content ; the latter would have the content . Both judgments are not inseparable: to wit, I might make the former judgment unconsciously, and thus be unable to form the latter meta-judgment which contains it, and which seems to require awareness of my judgment. I have given a rather whimsical example, but the moral can be applied to almost any non-generalisable subjective taste or preference. The fact that I am making a judgment about a state of affairs in the world does not mean that that judgment is thereby truth-seeking, as I might be aware that my reasons for making that judgment are highly biased and so take that judgment with a pinch of subjective salt. Thus truth-directedness would appear not to be a necessary condition for rational judgment either. By rational judgment, Peacocke seems to understand a double-barrelled inferential-referential process—where each of the halves, the inferential and the referential, would correspond to the two stages of the theory just cited. Yet perhaps it might be possible to give a more restricted definition of judgment which did not involve necessarily linking inference to reference. For in Peacocke's theory the definitions of inferential judgment, truth, and belief are mutually dependent—that is, each is defined in terms of the other two. Perhaps an alternative would be to define judgment exclusively in inferential terms, without defining its rationality and correctness by means of reference, and explaining reference (and so belief and truth) separately. This would be more akin to the original Fregean doctrine of sense and reference, whereby cognition would not attain referents (if they exist at all) directly, but only through the medium of sense. In this conception, there would be a one-way relation between the domain of sense and reference, whereby concepts would determine referents, but not the other way round. This is obviously a project too large to discuss here, but which I hope to expound elsewhere in the near future.

3 Implicit conceptions I would like now, however, to turn to the notion of implicit conceptions. My discussion will necessarily be briefer, given that this is a more recent development of the theory and I believe that, even though I have only a rather sketchy


99

understanding of implicit conceptions and their role within Peacocke's larger scheme, the theory itself still seems to be very much a work in progress. In his article “Implicit Conceptions, Understanding, and Rationality” (Peacocke 1998), Peacocke explains that his account of possession conditions in SoC was problematic inasmuch as it did not account for the rationality of accepting primitive axioms. Related to this is the question of learning: how can we rationally accept new principles which do not follow from those which we already accept? Implicit conceptions are meant to solve these problems. In order to illustrate the problem of learning, Peacocke uses the example of a young student being taught the axiom A → (A or B) for the first time. Explaining the acquisition of this axiom in terms of its defining rules as coded in truth tables is clearly inadequate. For on the one hand, the student can accept the correctness of the axiom without being acquainted with truth tables at all. And on the other hand, accepting the correctness of the relevant truth tables is also a different matter from accepting the correctness of A → (A or B), and can be quite independent from it. Peacocke solves the problem by claiming that the student's understanding the connective “or” either involves, or can be identified with, the content that any sentence of the form “A or B” is true iff A is true or B is true. This content is reached by a sort of simulation, the exercise of an imaginative “off-line capacity”— and it is this content that constitutes an implicit conception. Yet isn't this sort of intuition precisely what truth tables are purported to encode? Can't this be described as the textual, or "imaginative", equivalent of a truth table? Not really: for, according to Peacocke, implicit conceptions are subpersonal propositional attitudes of which the thinker need not have explicit knowledge, whereas the axioms and rules specified in truth tables—and which, in turn, define concepts—belong in the personal, fully conscious level. Moreover, implicit conceptions, as the contents of imaginative simulations, are non-inferential, as opposed to the inferential nature of full-blown concepts. I will now quickly run through the main problems which I find with this latest aspect of Peacocke's theory. As said, given that the theory of implicit conceptions is still not fully developed, I will restrict myself to broadly outlining some questions which the theory must address in order to provide an adequate account of concepts, rather than giving detailed objections. Firstly, I would like to question the necessity of positing implicit conceptions altogether, given that the introduction of such a notion appears to make concepts redundant. If implicit conceptions belong in a subpersonal level, it would seem that the task of providing an account of them should best be left to (computational) psychology. Thus, if concepts are to be accounted for in terms of implicit conceptions, a philosophical theory of concepts would turn out to be superfluous, as it could be replaced altogether by a psychological theory of implicit conceptions.

100


To this objection Peacocke has retorted6 with an analogy to vision: just as an understanding of the physiology of the eye does not eliminate the need for a cognitive theory of vision, so the existence of a psychological underpinning (under the form of implicit conceptions) does not remove the need for a philosophical account of concepts. But this analogy, I believe, is inaccurate. For according to Peacocke's description of conceptual dynamics, it would seem that full-blown conscious concepts are merely the personal-level, explicit outer shells for subpersonal, implicit conceptions, and concepts can (and ultimately, must) be explained in terms of implicit conceptions. Not so with vision: the physiology of the eye does not enter any account of the cognitive process of seeing. Whereas a theory of vision need not refer to the physiology of sight, a theory of concepts must, according to Peacocke, refer to implicit conceptions. And so, again, if implicit conceptions are invoked in order to account for the same explananda that concepts were meant to explain, why keep concepts in the picture at all? Secondly, I fail to see how a theory which resorts to subpersonal entities can explain the passage from the non-inferential to the inferential. We are told that inferential, explicit concepts on the conscious, personal level are grounded on noninferential, implicit conceptions which belong in a subpersonal level. Yet it is not explained how exactly the passage takes place, when it does. It is claimed that understanding a given concept, such as that of conjunction, relies on certain subpersonal thought processes of an imaginative, non-inferential nature, which appear to enact and simulate in some way the content specified by the possession conditions for the concept in question. But the explanation goes here only in one direction, it seems to me: we can establish what implicit conceptions underlie a given concept, but how are we to determine which concept a given set of implicit conceptions will produce, or indeed whether it will result in any concept at all? Moreover, and perhaps more importantly, is not clear at all how the leap from subpersonal, unconscious non-inference to personal, conscious inference takes place. This must be explained without falling in the Homunculi trap. Finally, a more general objection might be that an account of concepts in terms of more basic components—which seems to be what the introduction of implicit conceptions into the theory is meant to provide—should explain how the meaningful arises out of the non-meaningful, how sense arises out of non-sense. But Peacocke's implicit conceptions are still meaningful and indeed rational, if on a subpersonal level. Isn't positing implicit conceptions to explain concepts just taking the explanation one step back? Inferential, explicit concepts can be explained in terms of non-inferential, implicit conceptions—but in both cases we are dealing

6

Ibid.


101

with meaningful entities which, moreover, preserve rationality. The question remains of how rationality and meaning emerge to begin with.

Bibliography Geach, P. T. (1976). "Saying and Showing in Frege and Wittgenstein", in K.J.J. Hintikka (ed.), Essays on Wittgenstein in Honour of G.H. von Wright, Acta Philosophica Fennica 28: 54-70. Peacocke, C. (1992). A Study of Concepts. Cambridge, Mass.: MIT Press. ——— (1998). "Implicit Conceptions, Understanding and Rationality", in Philosophical Issues 99: 121-41. ——— (2004). The Realm of Reason. Oxford: Oxford University Press.

PART II: CONSCIOUSNESS AND THE METAPHYSICS OF MIND

CHAPTER SIX DEFLATIONISM ABOUT QUALIA

DAVE WARD

This paper aims to provide a schematic solution to one formulation of the “explanatory gap” problem for phenomenal consciousness: the alleged impossibility of explaining certain features of our conscious experience in structural, functional or physical terms.1 I begin by setting out the problem, and giving five features of experience that a reductive theory of phenomenal consciousness must account for. Drawing on the work of several “deflationist” theorists, I argue that, for each feature, a convincing reductive explanation can be given either for the feature itself, or for the introspective evidence which leads us to posit it. The goal is to draw together elements from different deflationist proposals and demonstrate that they can be made to cohere and support each other, resulting in a promising framework for the explanation of phenomenal consciousness.

1 The explanatory gap “The explanatory gap” refers to the alleged inexplicability of phenomenal or qualitative features of conscious experience in structural, functional or physical terms. The arguments in favour of its existence are familiar (inverted or absent qualia, the knowledge argument), but all aim at a common end2. All aim to demonstrate that facts about qualia do not logically/conceptually supervene on facts about structure and function—that our concept of phenomenal consciousness is not 1

Thanks to Andy Clark, Richard Clarkson and Pepa Toribio for helpful comments and criticisms of earlier versions of this paper. 2 For the introduction of the term “explanatory gap”, and a clear exposition of the problem, see Levine (1983). For the inverted qualia argument, see Shoemaker (1982). The knowledge argument was originally stated in Jackson (1982).


105

such that the facts about it can be fixed by physical facts. The clearest example of such an argument is the absent qualia, or “zombie”, argument, as championed by Dave Chalmers.3 Zombies are creatures who are physically and functionally identical to their human counterparts, but differ only in their absence of phenomenology. Because of their physical and functional identity to us, they will be behaviourally and psychologically identical to us. Not only will they say and do the same things, they will also have the same beliefs, desires, and other propositional attitudes as us (since it is thought that these are amenable to functional analysis in a way that the phenomenal features of our mental lives are not), including the same beliefs about their conscious experience as we have. However, Chalmers holds, it is logically possible that this psychological identity fails to ensure phenomenological identity—our physical duplicate might be a phenomenological zombie. Can we coherently imagine such a creature? Chalmers claims that “it just seems obvious”4 that we can, and his argument relies on this intuition. If we are really able to imagine such a creature, this shows that our concept of consciousness does not depict consciousness as logically supervening on physical facts. This would secure an explanatory gap; if we are to give a satisfying explanation of the phenomenal in terms of the physical, we need to be able to see how the physical gives rise to the phenomenal. Our concept of the phenomenal needs to be reducible to the physical in such a way that, given things are a certain way physically, we can see that they must be a certain way phenomenally. If this is not the case, and there is even conceptual room for phenomenal facts to vary whilst physical facts are fixed, we cannot be said to have reductively explained phenomenology in terms of physical facts. So long as Chalmers is correct in holding that zombies are a conceptual possibility, it seems that we cannot hope for a reductive explanation of phenomenology. Chalmers uses a variety of arguments such as this to support a taxonomy of the problems of consciousness into “Easy” and “Hard” varieties. “Easy” problems include the analyses of memory, reportability, and attention, all thought to be amenable to explanation in structural and functional terms. “The Hard Problem” consists in explaining phenomenal consciousness—the subjective character of experience which is thought to elude the structural and functional explanations appropriate for the “Easy” problems. Here are three possible responses to this dichotomy: 1) Accept it—be swayed by Chalmers’s zombie argument, and conclude that phenomenal consciousness is not amenable to the sorts of reductive explanation we use to answer the “Easy” questions. Types of theorist in this 3 4

Chalmers (1996), ch. 3. Ibid., p.96.

106


camp will include property dualists, such as Chalmers, who argue that reductive physicalist explanations are inappropriate here since consciousness is, in some sense, non-physical, and mysterians, such as Colin McGinn, who argue that we lack the cognitive resources necessary to provide such an explanation.5 2) Deny it—the zombie argument is misguided, since there is no such thing as phenomenal consciousness to explain in the first place, and therefore nothing left to explain once the “Easy” questions are answered. This is the “eliminativist” option—it is questionable whether any philosopher genuinely holds (or has held) this view, though Dan Dennett and the Churchlands are often accused of eliminativism6. 3) Deny it—though there is such a thing as phenomenal consciousness, it can be explained by answering sufficiently many “Easy” questions about cognitive processes and processing, and the relationships between these processes. The apparent difference in kind between “Easy” and “Hard” questions is illusory. We are misguided to think that zombies are conceivable, and once philosophers/scientists have shown us how to think about consciousness correctly, we will see this. This is the “deflationist” option, which I shall be defending for the remainder of this paper. It seems to me that many theorists who are accused of eliminativism really belong in this camp—no theorist really denies that there is such a thing as conscious experience to be explained. Rather, some theorists take issue with the way in which this experience is described—arguing against the connotations of the technical notion of qualia, for example. Dennett, properly read, is a paradigmatic example of such a theorist, and many of the ideas I will appeal to in what follows are present, in some form, in his work.7 I take the deflationist task to be to provide an explanation of how qualia (or phenomenology, or subjectivity—I think the terms are interchangeable in this context) can be explained in terms of structurally and functionally characterisable processes—the sort of processes that solutions to the “Easy” problems of consciousness are expected to make reference to. In what follows, I aim to show how existing work, taken together in a suitable way, suggests how this might be done. 5

E.g. Chalmers (1996), McGinn (1990). Paul Churchland is an eliminativist about propositional attitudes (see e.g. Churchland (1981)), and a staunch critic of anti-reductive theories of phenomenal consciousness, but is not an eliminativist about consciousness. For his views on the subject, see e.g. Churchland (1995), ch.8. See Dennett (2003) for a partial defence of his methodology against eliminativism. 7 See e.g. Dennett (1991). Humphrey (2000) is also a good example of the deflationist strategy. 6


107

Before continuing, I should be clear about the scope of this paper’s aim. The arguments of some anti-reductivists, such as Chalmers, aim not merely at the conclusion that we lack a satisfactory explanation of phenomenal consciousness, but that we necessarily lack such an explanation. The characters of explanandum and explanans are such that we simply cannot see how one could explain the other—this is not a result of lack of clarity on our part, but an appreciation of a necessary inexplicability. It is this strong conclusion I want to take issue with— what follows is not to be understood as a satisfactory explanation of phenomenal consciousness, but an attempt at a framework that might house such an explanation, and thus a rebuttal of the claim that we are necessarily unable to see how such an explanation might go.

2 The obligations of a deflationist theory To begin with, we need some understanding of exactly what is required of a deflationary theory of qualia. It is difficult to give an exhaustive list of criteria, since qualia, the explanatory target of the theory, are ill-defined. However, we can draw up a tentative list of the special properties qualia are supposed to have—those that are thought to make them features of experience that resist reductive explanation.8 Intrinsicality: Qualia are sometimes referred to as “intrinsic features of experience”.9 However, it is difficult to pin down the exact sense in which “intrinsic” is being used here—presumably not the straightforward sense in which mass is an intrinsic feature of an object. The general thought seems to be that qualia are features of experience which ground our beliefs and judgments about our experience—the various reactions and dispositions that arise from our qualia are (in principle) separable from the quale that grounds them. On this interpretation, the thing which is supposed to be intrinsic is the phenomenal character of the quale in question. It is intrinsic in the sense of being non-relational—the phenomenal character it has is not possessed in virtue of any facts about the typical effects or dispositions it gives rise to, and is thus in principle dissociable from them. The above gloss on intrinsicality also captures the idea of irreducibility that often seems implicit in qualophile arguments. To explicitly claim irreducibility as a feature of qualia would beg the question against any reductive attempt. With the apparently more innocent feature of intrinsicality, qualophiles are close to 8 The following list is basically the same as that of Dennett (1988), but with slightly different glosses on some of the items, and the addition of “simplicity”. 9 E.g. Harman (1990), though this is in the context of an attempt to argue against such a construal of qualia.

108


introducing irreducibility from the outset. It seems uncharitable for a deflationist to deny that there is something intuitive that qualophiles are getting at when they claim intrinsicality to be a feature of qualia. However, when the notion is unpacked as above, irreducibility practically falls out of it; if qualia have their phenomenal character (which the deflationist is trying to explain reductively) intrinsically (in the above sense), then this is tantamount to the claim that there is nothing in virtue of which the phenomenal character is as it is. If this is so, the game is up for the deflationist. This explains why irreducibility is often thought to result from our “everyday conception” of qualia—why Chalmers believes that we can see simply from considering the nature of qualia that they are not explicable in terms of structure and function. In light of this, it is tempting for deflationists to reject this feature of qualia out of hand, claiming that to insist on it is tantamount to question-begging. However, this would justifiably lead to accusations of eliminativism. The deflationist must attempt to account for the source of the intuitions that lead to the claim that qualia are in some sense intrinsic, in a way that leaves open the possibility of a reductive explanation. Moreover, the deflationist should attempt, as far as possible, to honour these intuitions; our tendency to characterise our phenomenology as intrinsic, and our feeling that this intrinsicality is such that it suggests irreducibility, is a bona fide fact in need of explanation. Though a deflationist cannot credit intuitions about irreducibility, an account which leaves them looking unmotivated will thus be implausible. So, the strategy here will be to deny that qualia are “intrinsic” in the above sense (of having their qualitative character independently of the various capacities and dispositions that are associated with them) whilst giving an explanation of the introspective evidence which seems to count in favour of intrinsicality. Ineffability: The character qualia have for the subject cannot be precisely expressed in words. We might be able to allude or gesture to the sort of qualia we are having, but any amount of metaphor or verbal description will necessarily fall short of a complete characterisation of the experience. No matter how eloquently I describe the taste of the wine I am drinking, my description will always lack some of the richness and particularity of my experience. I will argue that this is a genuine feature of experience, but one which the deflationist can account for. Privacy: Related to ineffability—qualia can be fully known only to the subject who possesses them. I argue that once ineffability has been explained, we can see why privacy is also a genuine feature of experience.


109

Simplicity: Like intrinsicality, it is difficult to pin down just what is meant by claiming “simplicity” as a feature of qualia. The general intuition seems to be that in contrast to the complex brain states they may be correlated with, or the complex properties in the world they alert us to, qualia are experienced as simple. Our sensation of the colour red seems to have a simplicity incongruous with its complex neural or microphysical basis. The sense in which this is so is difficult to pin down—it is certainly true to say that many of our experiences, such as the sound of an orchestra or the taste of a fine wine, strike us as extremely complex in character. It seems to me that the intuitions behind the positing of this feature are related to the features of direct availability and intrinsicality. Firstly, there is the apparent effortlessness with which we are aware of qualitative features of experience. Secondly, there is the way in which those features do not seem decomposable to component parts—a particular shade of red is determined by properties of hue, saturation and brightness, but we experience the shade as a single, simple property, and are not introspectively aware of the constituents which determine it. My strategy in what follows will be to argue that a deflationist can accommodate the above two points, and that the apparent introspective existence of a sense of simplicity which outruns these two can be explained away. Direct availability to consciousness: There are various ways this putative feature might be cashed out. Dennett (1988) suggests that it amounts to infallibility about qualia on the part of the subject who possesses them, before embarking on a series of thought experiments which undermine the possibility of such infallibility. However, this interpretation is too strong—a qualophile might want to hold that we can misdescribe or mischaracterise the character of our visual experience. For example, Noë (e.g. 2002) argues that though we are prone to describe our visual experience as uniformly rich and detailed, reflection proves that this is not so; rather, we are only aware of small patches of detail at a time. It seems reasonable for a qualophile to hold that they can accommodate this insight whilst retaining some kind of direct accessibility for qualia. A better gloss on “direct availability to consciousness” might be that we are “non-inferentially aware” of the qualitative features of experience. Certain features of experience “force themselves on us”, without us having to do any conscious mental work. Relatedly, it seems that we cannot, generally, miss or radically mistake our own qualia. I will argue that this is a genuine feature of experience, but the task of explaining it can be classed as an “easy” problem of consciousness. Qualophiles may want to add more to this list, but I think the above properties capture all the problems qualia pose for a deflationist. Ineffability and privacy constitute the reason why just what we are attempting to explain is so tricky to pin down. In itself, this seems to rule out the possibility of a satisfying explanation—if

110


there is no way to fully understand the character of qualia except from a 1st-person viewpoint, any reductive 3rd-person story about qualia will necessarily leave something out. Taken by itself, the feature of direct accessibility is the least problematic for a deflationist. However, our privileged 1st-person access to our qualia underscores the existence of the 1st/3rd-person asymmetry that is the deflationist’s main obstacle. As argued above, the feature of intrinsicality, with its undertones of irreducibility, is the most pressing problem for the deflationist to tackle. Clearly, the deflationist cannot succeed by taking these putative properties at face value. However, the qualophile claims that reflection on our qualitative experience suggests that it has the above features—thus a deflationist account that denies these features will be implausible. The emphasis on introspection leaves the deflationist some room to manoeuvre, though. For each putative property, two options are available to them. Firstly, they might argue that introspection does reveal that experience has these features, but leaves open the precise gloss we put on them. Thus, they might concede that there are aspects of our experience that are ineffable, private and intrinsic, but deny that we should interpret these features in a way that thwarts the possibility of a reductive account of qualia. Secondly, they might claim that introspection has led us astray—though it is tempting to posit the above properties, introspection does not force them on us. The introspective evidence might be consistent with a less demanding set of properties of experience. As noted above, in the outline of a deflationist account that follows, both strategies will be employed—arguments will be given to show that ineffability, privacy and direct accessibility can be accounted for by the deflationist, and that the appearance of intrinsicality (at least in the problematic sense at issue) is illusory, and can be explained away. A mixed strategy will be employed for simplicity—I argue that though qualia are not simple in any straightforward sense, we can explain the temptation, upon considering our experience, to claim some such feature for them. Will such a strategy be acceptable to a qualophile? I see no good reason why not. The qualophile claims that introspective evidence rules out the possibility of a deflationist account. If the deflationist can convincingly account for the introspective evidence to which the qualophile appeals, they have met this challenge head on, and surely delivered as much as could be required of them. A qualophile might object to such a deflationist account in two ways. Firstly, they might claim that the deflationist analysis of one or more of the above properties is inadequate in some way. Secondly, they might claim that the introspective evidence forces the qualophile analysis of one or more of the above properties of experience and rules out any deflationist one. In order to see whether either objection is open to the qualophile, let’s look at the type of account I think a deflationist should give.


111

Before we get started, note that as far as possible, the analyses of the properties should be taken together and viewed as a holistic account. The above glosses on the properties showed that several of them seem to overlap, or be motivated by similar intuitions—this is because what they aim to characterise (phenomenal consciousness) cannot be broken down into isolable parts, each corresponding to one of the properties. Rather, it must be seen as a single phenomenon, which we are attempting to explain, with the conjunction of the above properties aiming to do justice to the special features of this phenomenon. Indeed, we do think of phenomenal consciousness in this way, and this fact can make the type of “divide and conquer” strategy employed below seem inadequate—no single treatment of a property does the whole job, and thus after every section there is still the sense of something important having been left out. However, just as the above properties must be taken together to do justice to phenomenal consciousness, so the following accounts must be appreciated as a whole if they are to have their full explanatory purchase.

3 Intrinsicality As noted above, intrinsicality is the most problematic property for a deflationist account of qualia, and my aim is to deny that qualia are intrinsic in the reductively problematic sense outlined in the definition above, but give an explanation of the introspective evidence that leads us to think otherwise. A recent paper by Philip Pettit (2003) suggests a way in which this might be done. Pettit is concerned with the question of what it is for an object to look red (or any colour), and contrasts two possible responses. Firstly, the qualia theory—for a qualophile, an object looks red (has a red look) to a perceiver if it produces a red quale in the perceiver. This quale then enables various capacities in the perceiver, such as the ability to sift that object from differently coloured objects, to sort it into the same category as similarly coloured objects and to track it over time, all on the basis of the sensation of its colour. Furthermore, the quale is perceptually manifest to the perceiver (they are aware of themselves as having it), and manifestly enables the above capacities (they are aware of the quale as so enabling them). Thus, the quale is directly available to consciousness (by being perceptually manifest), and intrinsic (by being that which manifestly enables sifting, sorting and tracking capacities). Pettit does not discuss ineffability and privacy, but we can easily suppose that his qualia theorist claims these properties for the quale in question10. Pettit favours an alternative, dispositional account of what it is for an object to look red, basically consisting in a reversal of the explanatory direction of the qualophile’s strategy. For the qualophile, the red look of the object (the red quale) 10

Pettit does discuss simplicity—see section 6, below.

112


is what explains the subject’s perceptually manifest ability to sift, sort and track on the basis of the object’s look. For Pettit, the object has certain properties (the microphysical properties subserving its colour properties) which are such as to manifestly enable perceivers to sift, sort and track it on the basis of its colour. The sum total of these abilities forms the supervenience base for the red look of the object—the abilities to sift, sort and track are explanatorily prior to, and constitutive of, the red look of the object, not vice-versa.11 On Pettit’s dispositional theory, the red look of an object (which we may continue to call a quale) is not intrinsic in the sense outlined above. On Pettit’s view, the qualophile is exactly wrong in thinking that the red look is what enables our various colour-based capacities. Clearly, Pettit owes us a story of why we should accept this account rather than the (in this case) more intuitive qualophile view of matters. Pettit has a couple of things to say on this score. Firstly, he gives an analogy with a case where the idea of an intrinsic quale that grounds our judgments is implausible. Consider what it is for a ball to look as if it’s moving fast. Can we think of the visual sensation we get when perceiving a fast-moving ball as effectindependent—as a visual sensation that we might have whilst perceiving a slowmoving ball, being inclined to judge that the ball is moving slowly and disposed to act accordingly? Or alternatively, can we think of perceiving a fast-moving ball, being disposed to judge that it is moving fast, to act in ways appropriate to being confronted with a fast-moving ball, but having a visual sensation as of its moving slowly? In these cases, such effect-independence seems implausible.12 The look of going fast is “essentially tied to the effect of inducing in you the judgment “It’s going fast!” and, no doubt more primitively, certain reaching and ducking responses.”13 If it is plausible that the visual sensation of speed is constitutively 11

Note that this means Pettit’s account is not quite a reversal of the qualophile’s—the qualophile holds that the red quale causally underlies the subject’s various perceptual capacities. For Pettit, the capacities do not cause the object to have a red look. Rather, the red look supervenes on these capacities. Keeping this in mind lets us see that Pettit’s title— “Looks as powers”—might be somewhat misleading. The red look of an object is not a power in the object, or even a power enabled in the subject, though it is closely tied to both of these. Rather the look supervenes on the discriminatory powers of the subject, enabled by features of the object. 12 It might be thought that cases such as the Muller-Lyer illusion constitute examples of a phenomenal look being independent of dispositions to judge. Pettit might respond that our subpersonal perceptual mechanisms still dispose us to judge that the lines in the illusion are of different length, but that this disposition is vetoed by a higher-level disposition that has arisen as a result of previous encounters with the illusion. Pettit must hold that if the correction of the dispositions really does take place wholly at the subpersonal level, then the lines will look the same length to the subject. I think this is a plausible contention. 13 Pettit (2003), p. 229.


113

linked to the disposition to make certain judgments and responses, why should it be implausible that there is an analogous constitutive link between the visual sensation of a colour and the various dispositions, judgments and abilities associated with it? There does seem to be a disanalogy between the cases, however, and I think it comes down to the apparent ineffability of colour experience, something that is not echoed in our experience of “fastness”. But Pettit’s account (perhaps with some help from other accounts) may have the resources to explain away the apparent ineffability, as I shall argue below. Secondly, it helps to emphasise that the enabling of these abilities will occur at the subpersonal level. The alternative to this would be that the exercise of our sorting, sifting and tracking capacities would involve some kind of inference or deduction on our part. If we were to take such a view of experience, it seems we would need to posit some quale on the basis of which we made our inferences— otherwise the story would be that we made conscious inferences on the basis of no visual sensation, and that visual sensations somehow materialised in their wake. Clearly, Pettit does not have either of the above in mind. Rather, “The idea is that perceptual exposure to a red object tunes us in a way that enables us to sift, sort, track and so on, and that this tuning occurs “behind our back”; it occurs in virtue of the brain’s responses to the perceptual input at a subpersonal level.”14 Emphasising the phenomenological implications of this point is important—in virtue of various subpersonal goings on (the nature of which are discussed in more detail below) we find ourselves with a broad and complex range of discriminatory abilities, all manifestly in virtue of our visual exposure to the object. Consider what this process might feel like “from the inside”. We see an object, and as a result we manifestly have certain abilities. We have no privileged access to how these abilities arise— we just see them as manifestly there, on the basis of our visual experience.15 On Pettit’s account, the performance of the necessary subpersonal tasks is what constitutes the look of the object—so for the perceiver the look and the enabling of the abilities will occur simultaneously.16 Reflecting on this, we might make the 14

Ibid. p. 232. With the exception of this quote, the importance of the subpersonal/personal divide is underemphasised by Pettit. However, it is at the heart of Clark (2000): “A case where access implies qualia?” The above discussion draws heavily on that article. 16 To be clear, the discriminations at the subpersonal level are held to be constitutive of the phenomenology at the personal level. This might provoke the worry that the claim is too strong—not all the discriminations made at the subpersonal level are phenomenally conscious. To address this worry, the account must be supplemented with a theory of what it takes for a particular (subpersonal) brain event to become conscious. Just what this theory will be is largely an empirical matter, and beyond the scope of this paper. However, note that the “consciousness” of which it will be a theory consists merely in information’s arriving at the personal-level, rather than the phenomenal consciousness at issue. Thus the deflationist is not begging any questions by postponing a definite account here. 15

114


following point: all we get in experience are the look of the object and various discriminatory capacities. It does not seem that introspection can tell us reliably whether the look causes the capacities, or supervenes on them—both Pettit and the qualophile’s story are consistent with the introspective evidence.17 To recap, I have glossed the property of intrinsicality roughly as “possible effect-independence”. Pettit’s position denies the possibility of such effectindependence, by constitutively tying the qualitative look to the various capacities that are enabled at the subpersonal level. My contention above is that the introspective evidence is compatible with both Pettit’s theory and the qualophile’s. So, on the above reasoning, Pettit’s denial of intrinsicality is as plausible as the qualophile's assertion of it—which account to prefer will have to be decided by other considerations.

4 Ineffability So far, our story states that when perceiving a coloured object, subpersonal processes give rise to complex discriminatory capacities, upon which the look, or quale, of the object supervenes. The natural worry at this point is that the capacities Pettit appeals to do not seem adequate to support the richness of our colour experience—a package of discriminatory capacities sounds like just the sort of thing a qualophile will deny that qualia can be explained in terms of. At this point, most sympathies will doubtless be with the qualophile. However, Pettit is aware of this, and attempts to erode this intuition by examining in more detail just what these capacities will entail. In this section I argue that his arguments, combined with some similar insights from Dennett, provide us with the resources to explain ineffability. The basic idea will be that ineffability arises from deliverances of subpersonal processes which are sufficiently complex to outrun our personal-level capacities to describe them. To address the concern that the capacities he considered are not suitable candidates for the grounding of qualia, Pettit argues that our intuitive conception of sifting, sorting and tracking as simple capacities is mistaken if we believe that the simplicity extends to the cognitive capacities which underlie them. When we appreciate just how much is required to ensure that these capacities are in place, it becomes implausible that the look of an object and the discriminatory effects it is associated with could come apart. Citing accepted neuropsychological research,18 17

For Pettit’s story to be plausible, the sifting, sorting and tracking capacities he appeals to should be rich enough that it is implausible that they could leave the visual sensation underdetermined, allowing the possibility of the capacities staying the same while the look varied. Pettit’s strategy to ensure this is outlined in the following section on ineffability. 18 Which he takes from Byrne and Hilbert (1997).


115

he lists three capacities required to be able to discriminate coloured objects in the way his list of abilities requires: 1) The subject must be sensitive to variations in illuminance, such as the difference in light levels between dawn and dusk, and be able to compensate accordingly in its discriminations. 2) The subject must be sensitive to different wavelengths of light, as these pertain to colour, and be able to correct for differences in intensity of the light. 3) The subject must be sensitive to differences in the surface spectral reflectance profiles of different objects—the “differential powers of different surfaces to reflect different wavelengths of light in different ratios.”19 Each of these abilities will in turn require a different set of abilities to subserve it. Pettit focuses on 3), which will involve: I) The ability to detect edge contrasts in surface spectral reflectance. This is achieved by the operation of double opponent cells which are excited by one colour and inhibited by its complementary colour. II) The ability to compare the colours of two objects on either side of another. This is accomplished by a cognitive mechanism which integrates information about the edge contrasts in surface spectral reflectance. III) The ability to standardise different contexts (for example correcting for differences in light and shade) when comparing the colours of two objects on either side of another. Thus discriminability requires more than our intuitive folk psychology gives it credit for: “Colours meet the eye only in virtue of a highly complex registering of intensities and wavelengths of incident and reflected light, and in virtue of a highly complex registering, integration and standardisation of the contrasts displayed by various objects in the ratios at which they reflect different wavelengths.”20

To press home the point that folk psychology can lead us astray here, Pettit draws attention to the possibility of selective impairment of these abilities. Intuitively, it seems that if someone is perceiving three objects side by side and perceives the edge between the first and second as being a red/yellow contrast, and the edge between the second and third as being a yellow/red contrast, they can easily detect that they are looking at a yellow object flanked by two red ones. However, this is not necessarily so—selective impairment of one or more of the above mechanisms can result in the subject’s having an intact contrast-detecting mechanism, but also an inability to use the information from it to (for example) contrast the colours of separated objects. 19 20

Pettit (2003), p. 233. Ibid. p. 234.

116


The complexity of the deliverances of our colour-discriminating mechanisms is further brought home by some of Dennett’s remarks on colour vision. Dennett precedes his discussion of colour phenomenology with an evolutionary story about colour vision. Dennett argues that colour vision evolved as a way of discriminating and categorising features of the environment. According to Dennett, evolutionary biological orthodoxy has it that colours in nature co-evolved with the colour perception of certain creatures. For example: apples are not red only because of chemical changes occurring within them when they ripen. It is also due to the fact that without apple-eating seed-spreaders, there would be no apples. Thus, if appleeating seed-spreaders have good red-discriminating colour vision, apples which are red when they ripen are evolutionarily better off than those which are not. Dennett and the theorists he cites believe that all colour vision arose this way, by symbiotic relationships between organisms. The upshot of this view is that all inorganic (and presumably many organic) objects are seen to have colour only derivatively—once a creature can see red apples it can also see red sunsets, but this is a coincidental bonus with no evolutionary advantage. “Why is the sky blue? Because apples are red and grapes are purple.”21 Our contemporary situation is very different from the ones in which our colour vision evolved—we retain many discriminatory capacities and reactive biases that were evolutionarily essential to our ancestors, but not to us. Additionally, another Darwinist tenet is that there is always random variation, which may become evolutionarily relevant at a later date—thus there might be features of our colour experience that no longer have, or never have had, any adaptive advantage. Taken together, the above considerations might constitute a partial explanation of what we find mysterious, or difficult to explain about colour experience. Qualophiles are fond of asking why experience does not go on “in the dark”—why should we have qualia at all? However, we can begin to address this question by noting that there was clear reason for our distant ancestors to be able to discriminate ripe from unripe fruits.22 If we add to this Pettit’s case against intrinsicality, above, we can start to see the possibility that these discriminatory capacities could be the basis of what we take to be qualia. The knock-on effects of colour vision (that once a colour can be discriminated in one place, it can be seen wherever the same microphysical conditions occur), the likelihood of discriminatory capacities and biases that are no longer evolutionarily important, and the existence of random variation can all contribute to a diagnosis of the 21

Dennett (1994), p.378 It’s tempting to wonder whether this story hits its mark—why is the information about the world felt rather than simply registered? I think this question is less tempting when we bear in mind my initial plea to bear in mind one part of the account when assessing another—the information is felt due to our limited access to the subpersonal mechanisms which register it. I try to emphasise and clarify this point in the following section. 22


117

intuition that we cannot give plausible evolutionary stories about the character of our colour experiences—attending to the above points lets us see that it is simply true that many of the features of our way of experiencing the world are as they are for no good reason. A strength of this model is its ability to account for our difficulty in tying colours to particular primary properties in the world—historically every attempt to do this in a precise way has failed. We can diagnose this as due to the fact that colours, as we experience them, “are the product of biological evolution, which has a tolerance for sloppy boundaries when making categories.”23 In contrast to the simple way we perceive them, colours are complex and loosely demarcated properties of objects. This complexity is what Dennett believes to account for the apparent ineffability of our experience.24 Dennett draws our attention to the password system improvised by Soviet spies, the Rosenbergs25—two agents who needed to identify each other would each be given one half of a torn piece of cardboard. The ragged edge the tear produced was sufficiently complex that only the other edge would fit it—each piece becomes a unique “shape detector” for the other. The complexity of each edge allows it to defy description; the only feasible way of describing the shape it instantiates is by pointing to its partner and saying “the shape which that thing detects.” This ineffability is due to complexity, rather than some more mysterious source. So it is with colour perception—our colour experiences are practically ineffable due to the complexity of the property detected and property-detecting-mechanism within us. It seems the only way to describe it fully is to “point inward” to your experience. However, as Pettit argued above, we need not assume that that which we are ostending to is a quale in any problematic, intrinsic sense. Rather it is the activation of some property-detector and the discriminatory effects it enables. Again, the fact that we have limited personal-level access to the deliveries of these property-detectors is an important point, emphasised by Clark (2000)—we effortlessly find ourselves with rich and reliable information about complex properties in the world, oblivious to the subpersonal hard graft that makes this so. The complexity and informational richness of the subpersonal deliverances outruns our personal-level capacities to adequately 23

Ibid., p. 382n. Clearly, complexity does not always imply ineffability—the brain is a complex system, but there is nothing particularly ineffable about it. Additionally, all cognitive states are subserved by a complex neural architecture, but not all have the ineffability we find in the case of colour experience. Dennett’s story only makes sense when coupled with a story about the access we have to this subpersonal (and worldly) complexity. Keeping within the scope of this paper, I shall assume that some such story could be given in functional/computational terms, and thus that spelling out such a story counts as an “easy” problem, rather than part of the putative “hard problem” which I’m concerned with here. 25 Dennett (1988), p. 422. 24

118


describe it, and this richness, coupled with the fact that the subpersonal deliverances are manifest to us without knowledge of their aetiology, gives rise to ineffability. On this interpretation, we should see the ineffability as expected and explicable, not as any kind of mysterious obstacle to a deflationary account.

5 Privacy An account of privacy falls out of the above characterisation of ineffability. For any subject, the property-detectors which subserve the qualia are the subject’s own, giving rise to experience only for them, rather than anyone else. Perhaps, were ineffability not a feature of experience, privacy (in the qualophile’s sense) would not be an issue; though the experiences would still belong only to the subject, their character would be fully communicable, removing any apparent mystery from the privacy of experience. However, since the deliverances of the subpersonal mechanisms involved are rich enough to outrun the subject’s capacity to describe them, the resulting qualia are necessarily knowable only to the subject. Thus, the aspect of privacy which makes it appear problematic for a reductive theory is its link to ineffability. If we accept the above deflationist account of ineffability, the problematic aspect of privacy is thus taken care of.

6 Simplicity In section two I noted that it was hard to pin down just what the feature of “simplicity” might amount to. One reason for thinking that a feature we might call “simplicity” presents a puzzle for a deflationist theory is that despite the richness and complexity of the suite of effects of each colour, we still regard different instances, shades and hues of the same colour as being instances of one and the same property. Why should we experience colours as grouped in this way, rather than experiencing each different colour as a discrete property? Pettit (2003) addresses this by appealing to our implicit knowledge of the relations colours bear to each other: “In every instance where something looks red, the subject will […] instantiate a particular discriminatory effect. […] But in addition to this discriminatory effect, the subject who is experienced in exposure to colour will presumably form an expectancy that the object will give rise to corresponding discriminatory sensitivities in cases of different kinds.”26 Thus, the requisite unity is secured by what Pettit calls a “meta-cognitive commitment” on the part of the observer—the observer sees the colour as belonging to a family of surrounding colours by (implicitly) understanding the different way that the appearance of the colour of the object will change in different circumstances, and 26

Ibid. p. 237.


119

understanding its similarities to and differences from other objects both in the same and different contexts. This aspect of Pettit’s approach is reminiscent of Alva Noë’s treatment of “the problem of perceptual presence”27—we perceive a sphere as wholly spherical despite only being presented with its facing side, in virtue of our implicit knowledge that were we to walk around it, or turn it around, we would see the rest of its spherical shape. If we expect it to have a non-spherical shape behind its facing side, our phenomenal experience of it will not be as of a sphere. Pettit is appealing to our implicit knowledge of colour space to account for the phenomenal unity of colours in the same way. In the spirit of the Dennettian evolutionary story told above, we might also speculate that there is some evolutionary utility in perceiving colours in this way— perhaps a creature who perceived each different permutation of hue, saturation and brightness as an instance of a separate property would be less able to swiftly act upon visual information, or would require far more cognitive power, than us and our more parsimonious way of perceiving colours. It might even be the case that the “meta-cognitive commitment” Pettit appeals to is a necessary product of our type of perception—it seems that the subpersonal capacities upon which our colour vision is based include sensitivity to differences and similarities between colours at a very low level. Given that this is so, is it really intelligible to suppose that we could experience a specific shade of red as an instance of a property discrete from all its nearby shades, whilst still being implicitly aware (in virtue of our subpersonal constitution) of its intimate relation with its neighbouring shades? Another aspect of the simplicity of qualia was their apparent lack of decomposability to constituent parts. Pettit’s discussion of the subpersonal components of colour perception (and the possibility of their dissociation) should have swayed our conviction in the importance of this intuition, but another example will serve to emphasise the point. Dennett (1991) gives the example of playing harmonics on a guitar. When an open string is plucked, the resulting note sounds “apparently atomic and homogenous”.28 However, pressing lightly on the twelfth fret of the same string when sounding the note results in a different kind of tone— purer, bell-like and one octave higher. Playing the two notes in succession you can clearly hear that the harmonic is “contained” within the original note: “The homogeneity and ineffability of the first experience is gone, replaced by a duality as directly apprehensible and clearly describable as that of any chord.”29 There are other instances of apparently simple experiences which can be decomposed in this way—think of a wine-taster learning to tease apart the different notes in a fine

27

E.g. Noë (2004, ch. 4). Dennett (1991, p. 49). 29 Ibid., p. 49. 28

120


vintage, or, in the case of colour, the way an experienced artist can learn to see a colour as composed of those which she needs to mix it on her palette. So, the sort of deflationist account I recommend can account for simplicity in one sense (the sense in which different instances of the same colour are experienced as one and the same property), and claim that it is not a genuine feature of experience in the other (many of our experiences seem indecomposable, but in reality are not). Is there more to the feature of simplicity than this? In section 2 I suggested that simplicity may be related to intrinsicality and directavailability—our qualia are immediately and effortlessly present in our experience and do not seem decomposable to component parts. The discussion of intrinsicality in section 3, and against decomposability, above, aim to show that neither aspect is a barrier to reductive explanation. If the qualophile wishes to claim that simplicity is a feature of experience in some sense other than those accommodated here, the onus is on them to spell out what this could be.

7 Direct availability I have saved the treatment of direct availability until last, since this is one feature of qualia about which I do not have a positive story to suggest. This might seem significant, especially since almost all of the foregoing discussion has assumed and relied upon the direct availability of the fruits of subpersonal processing to the personal-level. However, I do not think my account suffers from this lack—direct availability is unique among the original list of properties in that it is held (even by qualophiles) to be characterisable in structural and functional terms. The reductively problematic features of qualia were the various features of the items we have direct access to in experience, rather than the access itself. To see this, consider the fact that the distinction between availability and phenomenology is precisely where Block (1995) chooses to draw the line between his two versions of consciousness—access consciousness being reductively benign, phenomenal consciousness reductively problematic. Problems of availability of content for verbal report/control of behaviour/access to working memory are perennial favourites on the “easy” list of Chalmers’s problems of consciousness. And Chalmers responds to Clark (2000)’s attempted explanation of consciousness in terms of patterns of access by claiming that a zombie can have all the access you like, but still lack phenomenal consciousness.30 Thus, although an explanation of the workings of direct availability is ultimately desirable, I don’t think it’s necessary for my present purpose—arguing that the “Hard Question” of consciousness can be answered by the easy ones.

30

Clark (2000, p. 35).


121

This might flag a closely related worry—about the way in which I have helped myself to a robust version of the personal/subpersonal distinction, which plays a crucial role in the foregoing account. In doing so, might I have illicitly smuggled in phenomenology, unexplained, by the back door? I think I can make a similar response here—although I don’t have an account of the distinction between the personal and subpersonal levels, this is something that clearly belongs on the “easy” side of Chalmers taxonomy. Again, to see this, we can consider one of Chalmers’s zombies. Chalmers allows that they are physically, functionally and psychologically identical to us. Thus, they can have all the same (personal-level) beliefs about, and degrees of access to, all the same (subpersonal) information registering and processing mechanisms as us. These conditions are all that are required to support the kind of personal/subpersonal separation required for the above account. Chalmers claims that they are compatible with the absence of qualia. The above account has aimed to show that they are all that is required for qualia.

8 Conclusion The above account attempts to undermine the intuition that qualia are “intrinsic” in a way problematic for the deflationist, by arguing that the phenomenal aspects of experience result from a certain type of access to the workings of subpersonal perceptual mechanisms. In light of this, it can explain the apparent ineffability and complexity of our perceptual experience by highlighting the complexity of the deliverances of those mechanisms. This complexity provides the basis of an account of why the subpersonal processes underlying a specific colour will be sufficiently distinctive that it is implausible that the phenomenal look and subpersonal processes which underlie it could ever come apart. We are left with a story that will make the following predictions about a creature with our discriminatory capacities, subserved by our particular subpersonal architecture (a creature who a qualophile such as Chalmers thinks could conceivably lack qualia)—it will sincerely report that there is something it is like for it to see a particular colour, and that the precise character of this “what it’s like” is not something it has the ability to put into words. For brevity’s sake, it might begin to refer to this “what-it’s-like” as the particular quale of the experience. Reflecting upon the ineffability of this quale, the creature will conclude that it is private to it, since firstly, it is a state of the creature and not of someone else, and secondly, the creature is unable to communicate the character of the quale with any great success—it must resort to “ostending inwards”. Whilst a deflationist will say that this inner ostension is merely to the highly complex results of a subpersonal mechanism, the creature (unless of a deflationist persuasion) will feel that it ostends to the quale of its experience. In light of all this, the creature will probably theorise

122


that the quale is “intrinsic” in some sense; the character of it is so perplexingly ineffable that nothing it can say—particularly no story it can give about the range of discriminations it is able to make, or other abilities its perception gives it—can communicate it, or, it seems, yield a satisfactory explanation of why the sensation is the way it is. It will seem that the sensation must somehow have its phenomenal character simply in virtue of itself, since any explanations in terms of anything less (such as the workings of some brain mechanism or other) seem unsatisfying. Presuming that the creature has a subpersonal architecture that enables it to discriminate various different colours and to appreciate the relations of different shades of colour to each other, and based on the apparent intrinsic and ineffable nature of its experience, it may also decide that its quale is, in some sense, simple. If the above is right, the creature will judge itself to be the locus of ineffable, private, directly-accessible, simple and intrinsic (and perhaps irreducible, depending on its philosophical temper) qualia, and will be sceptical of any reductive story we might try to tell about them. Importantly, the opinions of the creature are not “mere judgments” that simply occur to it. They are the considered and sincere result of its introspection on how it feels for it to have certain perceptual experiences. In a robust sense, they are a true account of how things seem to it. The deflationist cannot be accused of shifting the goal of explanation to verbal reports or judgments here—what is being explained is the phenomenological way things seem to the creature. Importantly, the phenomenological character of its experience is not what is being “deflated”. Rather, the object of deflation is the interpretation that the qualophile puts on its experience—a particular way of thinking about the phenomenological character of experience. Strictly speaking, I am advocating not deflationism about qualia, but merely about some of their putative features.31 The preceding remarks have aimed to show that the qualophiles conception of qualia and phenomenal consciousness is not compulsory, and to provide an alternative. To stress the above one last time—it really will feel to the creature as if it has qualia, if the above account is correct. How can a qualophile respond to this? One way might be to try to show that a deflationist story of the above type does not in fact ensure that the subject will have perceptual experience with the features of ineffability, privacy, direct accessibility and apparent intrinsicality, as the deflationist claims. I have made my case as best I can here—qualophile (or other) arguments to the contrary will have to be assessed as they arrive.

31

Of course, a qualophile might hold that anything without these five features is not worthy of the name qualia. In that case, the above account advocates not merely deflationism, but eliminativism about qualia. The important sidebar is that, here, this would not entail deflationism or eliminativism about phenomenal consciousness, which is what the above account aims to explain.


123

Secondly, the qualophile might hold that the story is still eliminativist in some way—the creature described could conceivably both believe and phenomenologically feel like it had qualia, but be mistaken. I’m not sure if any qualophile would want to take this route, but if they did, the resulting position does not seem tenable. To take this line is allow the possibility of an especially unusual type of zombie—a zombie who is identical to us in its phenomenology (assuming that we understand phenomenology to be the way things feel to the subject), but lacking in qualia. Taking such a tack seems to me to be an instance of the qualophile shifting the goalposts of explanation, and to an unacceptable location— rather than simply referring to an aspect of our phenomenology, the irreducible qualia is now something that somehow stands behind that phenomenology, ensuring it is real phenomenology, and not the inferior sort a zombie might have. This is certainly not the explanatory task with which I began the essay, and would not be one I would be tempted to take seriously, were a qualophile to press this line. What reason do we have to take such an explanandum seriously? Ex hypothesi, it is not something forced on us by introspection. If the qualophile challenge restricts itself, as it should, to the reductive explanation of the way things seem to us, I think the above considerations show that the deflationist has at least the raw materials for an answer.

Bibliography Block, N. (1995). “On a Confusion about a Function of Consciousness”, in Behavioural and Brain Sciences 18: 227-47. Byrne, A., & Hilbert, D.R. (1997). “Colors and Reflectances”, in A. Byrne and D.R. Hilbert (eds.), Readings on Color, Volume 1: The Philosophy of Color. Cambridge, Mass.: MIT Press. Clark, A. (2000). “A Case where Access implies Qualia?”, in Analysis 60: 30-8. Chalmers, D. (1996). The Conscious Mind: In Search of a Fundamental Theory. New York: Oxford University Press. Churchland, P.M. (1981). “Eliminative Materialism and the Propositional Attitudes”, in Journal of Philosophy, 78: 67-90. ——— (1995). The Engine of Reason, the Seat of the Soul. Cambridge, Mass.: MIT Press. Dennett, D. (1988): “Quining Qualia”, in A.J. Marcel and E. Bisiach (eds.), Consciousness in Contemporary Science. Oxford: Clarendon. ——— (1991). Consciousness Explained. London: Allen Lane. ——— (1994). “Instead of Qualia”, in D. Dennett (1998), Brainchildren. Cambridge, Mass.: MIT Press. ——— (2003). “Who’s on First? Heterophenomenology Explained”, in Journal of Consciousness Studies 10: 10-30.

124


Harman, G. (1990). “The Intrinsic Quality of Experience”, in J. Tomberlin (ed.), Philosophical Perspectives 4. Northridge: Ridgeview. Humphrey, N. (2000). “How to Solve the Mind-body Problem”, in Journal of Consciousness Studies 7: 5-20. Jackson, F. (1982). “Epiphenomenal Qualia”, in Philosophical Quarterly 32: 12736. Levine, J. (1983). “Materialism and Qualia: The Explanatory Gap”, in Pacific Philosophical Quarterly 64: 354-61. McGinn, C. (1989). “Can we solve the mind-body problem?”, in Mind 98: 349-66. Noë, A. (2002). “On what we see”, in Pacific Philosophical Quarterly 83: 185-94. ——— (2004). Action in Perception. Cambridge, Mass.: MIT Press. Pettit, P. (2003). “Looks as Powers”, in Philosophical Issues 13: 221-52. Shoemaker, S. (1982). “The Inverted Spectrum”, in Journal of Philosophy 79: 35781.

CHAPTER SEVEN HEIDEGGER AND AI: TECHNOLOGICAL METAPHOR AND SELF-UNDERSTANDING

RICHARD HAMILTON

“[T]he perceiving of what is known is not a process of returning with one’s booty to the ‘cabinet’ of consciousness after one has gone out and grasped it.” (Heidegger 2000, p. 89)

1 Introducing the subject The argument in this paper is directed against the use of Heidegger’s philosophy in speculating about the possibility of Artificial Intelligence (AI), rather than against the possibility of AI as such.1 I argue that Heidegger is not a friend of AI hopefuls, because he holds that it is not possible to recreate thinking by accurately determining the features of the thinking brain, whether it is along the lines of an input/output processor (Good old fashioned AI, or GOFAI), a neural network (modern more sophisticated connectionism), or anything else2. I want to 1

My thanks go to the participants of the Mind2005 conference, in particular those who provided difficult questions for me. Thanks in particular to Julian Kiverstein who presented a particularly incisive criticism of this paper, which caused me to rethink much of it. Needless to say I am entirely responsible for anything objectionable which remains. 2 The term “GOFAI” was coined by John Haugeland (1985). Good Old Fashioned AI is taken to mean symbolic AI, where the mind is conceived (after Hobbes, Haugeland tells us) as an abstract token manipulator. Similarly, the assumption behind GOFAI is that the mind, in the same way as a computer, is an automatic symbol manipulator. GOFAI tells us that in the mind, arbitrary symbols or tokens are utilized in reasoning. According to GOFAI, the symbolic nature of reasoning is in fact the basis for thought and consciousness. GOFAI has come under increasing criticism of late in AI circles, with many favouring more connectionist approaches. Connectionism denotes any system which focuses on neural

126


argue that one often overlooked element from Heidegger’s philosophy of technology speaks against all of this, and that is the idea of reflective selfunderstanding, which I take to involve an attack on any metaphorical conception of the mind. I argue on the basis of this element in Heidegger’s philosophy of technology that anyone who recruits Heidegger to support points about AI must accept “all or nothing”, since, as I will show, the elements of Heidegger’s work which are traditionally hauled out in AI debate are intimately bound up with the elements I will describe. I will argue that not only is Heidegger against a certain model of the mind along the lines of old fashioned AI, but he is against any model of the mind whatsoever. My claim is in response to the thrust of a paper by Mark Okrent, the title of which speaks for itself—“Why The Mind Isn't a Program (But Some Digital Computer Might Have A Mind)” (Okrent 1996). Okrent is one of what we can call the “pragmatist” readers of Heidegger, along with, most notably, Hubert Dreyfus, and a few others. His pragmatist reading of Heidegger explains the first half of his contention. As Dreyfus argues in his book on Heidegger (Being-inthe-World; Dreyfus 1991) and in another book on AI (What Computers Can’t Do, and 2nd edition What Computers Still Can’t Do; Dreyfus 1972, 1992), Heidegger’s account of Being-in-the-World lends him to a pragmatist reading, where the main claim is that, rather than a disinterested isolated subject moving about in a world of objects, Heidegger describes a being for whom meaning is constituted by use, a being who is constantly and concernfully tangled and embedded in a practical network of things. In his book on AI, Dreyfus argues that this pragmatist Heidegger, along with Investigations Wittgenstein, are antagonists to GOFAI research. This sort of strong AI involves the idea of a computer program, equipped with millions of beliefs and assumptions, working on a model of input and output. There are many arguments against the mind as a computer program. Aside from philosophical debate, many researchers in AI are embroiled in an internal debate about which type of research program will be most beneficial. What is of interest to me, however, is that Okrent makes a further claim, that some digital computer might have a mind. His argument is essentially thus: “Strong AI” is the wrong model of the mind, but some day we might come up with the correct one. This is a point which is hard to refute, since it offers us a mere “what if” scenario; that is, what if the mind was describable, then why couldn’t we hit upon a model of it, and why couldn’t that be a digital computer? Who’s to say? I won’t argue against this point, but rather against his claim that that there is nothing in Heidegger that networks rather than symbol manipulation. A neural network is a simplified model of the brain which is composed of large numbers of units (performing the function of neurons) along with weights that measure the strength of the connection between the units. Neural nets have had some success in areas such as face recognition, detection of simple grammatical structure and reading, although the extent to which it is a challenge to GOFAI, and its implication for AI in general is a hot area of debate.


127

precludes this, because the “pragmatist Heidegger” reading is based on arguments that appeal to the ontological structures of being human, or, as Heidegger would say, fundamental ontology. The question about whether this or that could be a mind is an “ontic” question. “Ontic” is an adjective Heidegger uses in opposition to “ontological” in Being and Time, and it is essentially a distinction concerning depth. “Ontic” pertains to mere things, facts, contingent qualities, etc., while “ontological” is supposed to be about deeper essential structures.3 So, in a sense, because the Heideggerian argument against GOFAI is, at some deep level, criticising the very assumptions about what it is to think or be conscious at all, it is unimportant what the “platform” or “substrate” for a thinking being is. I take it, therefore, that Okrent holds that the situating of the critique at the ontological level leaves it open that such a “platform” as a digital computer could be a mind, given the correct model was employed. Okrent’s claim (that, although GOFAI is a failed project, we might still model the mind correctly and as such a digital computer could be a mind) is implicit in the very fact that the philosophy of mind remains one of the most popular areas of contemporary philosophy. Strong AI might be all but dead and its own failure so far “in the field” might mean that those involved in AI research in general are more humble in their approach, but there is still a widely held view that it is only a matter of time. This is illustrated by one of the fathers of AI, Marvin Minsky, who famously claimed that “the mind is a meat machine.”4 The idea this materialism gives rise to is basically that either AI is possible, or we're not (after Herb Simon). From the perspective of philosophy we can put the point like this: as long as philosophers still think that there is such a thing as an adequate description of the 3

By his distinction between the ontic and the ontological, Heidegger means to capture the difference between “Being as such”, and particular entities. This is often referred to, in subsequent discussion as “the ontological difference”, between Being, and beings, between facts about entities (the ontic) and the being of those entities as such (the ontological). For Heidegger in Being and Time (¶10), the natural sciences such as biology, psychology, geology are ontic sciences, because they operate on an already understood conception of their subject; that is to say, biology already knows what life is, and consists of investigation into the entities within its remit, but does not ask about the being of those entities. It defines in advance what it is to be alive (bios). As such, ontology precedes ontical inquiry. If we undertake ontological questioning into what Being itself means, as Heidegger would have us do in Being and Time, then that inquiry precedes any ontic questioning which operates on an understanding of being which is already there. Thus what it means to exist as a thinker, to be conscious, is an ontological question, and Heidegger would say that it could not be answered by a science of thinkers (psychology). What sorts of things could be thinkers on the other hand, as an ontic question could be answered by an ontic science such as psychology, but only on the basis of a predefined understanding of what a thinker is (such as GOFAI or Connectionism provides; see Fn. 1. above). 4 Minsky (1986).

128


mind which accounts for consciousness and thought, there is always hope, the sort of hope expressed by Okrent. As long as intelligence and “minded-ness” are still conceived along the lines of something which a material platform supports, Okrent’s possibility may be maintained. We might note here that in making his claim Okrent is still playing a game that tacitly employs the metaphor of platform/operating system taken straight from computers. I will return to the idea of technological metaphor later. My argument will take the following form: 1. I will outline the thesis of “reflective self-understanding” and show that Heidegger holds it. 2. I will show that this thesis leads to a criticism of self-understanding based on technological metaphor. 3. I will finally show that this thesis is not just a part of the later philosophy that can be ignored, but is already there in Being and Time. 4. I will therefore conclude that Heidegger holds that AI research methods, and our reasons for pursuing them are merely symptomatic of a technological age.

2 Reflective self-understanding The main engine of this paper is the idea of reflective self-understanding. This is a thesis I will outline in a moment, which Heidegger certainly adheres to, most notably in The Question Concerning Technology, but also, as I will show, in Being and Time. I will also take examples from the phenomenologist Don Ihde, who outlines the thesis more clearly than Heidegger does. Similar and supporting points are made by Richard Lewontin (the dialectical biologist) and the late David Edge, former director of science studies at Edinburgh. Reflective self-understanding is the thesis that our world and the way we view it shapes the way we conceive of ourselves. Don Ihde (discussed below) addresses it as a question of how humans understand and interpret themselves in a given culture. He asserts that our culture is a technological one. This is an assertion dominant in Heidegger, in fact it is the basis of his philosophy of technology, that is, modern technology is the defining feature of the modern age. Before looking at this thesis in Heidegger, I will develop it by looking at two more clearly written accounts of some of the issues: first Ihde, then Lewontin. The place where this thesis of reflective self-understanding is developed most is in Don Ihde’s paper, “Technology and Human-Conception” (Ihde 1990), in Larry Hickman’s Technology as a Human Affair. Ihde’s paper makes the claim that “technology supplies the dominant basis for an understanding both of the world and ourselves” (Ihde 1990, p. 126). This links the question, “how do humans understand themselves?”, with the question, “how does technology affect this understanding?”


129

The structure of Ihde’s thesis is traceable to Husserlian phenomenology. Phenomenologically, world and self are equally certain for Husserl. It is from the world that I understand myself. Thus, it is in interaction with the world that I come to any form of self-understanding. In contrast to Cartesian subjectivism, phenomenology emerges as a new kind of relational objectivism. In other words, if I am already in a world, and it is by means of the world that I came to understand myself, then there is an essential sense in which self-understanding is always tied to an understanding of a world. On this view, to know oneself becomes a task and a process, not a Cartesian given. Being true to his phenomenological background, Ihde proceeds to examine what are the essential structures of reflective self-understanding by means of the Husserlian technique of variation. He examines three somewhat stylised versions of differing self-understanding in differing worlds. Running over them briefly then, the first variation is a somewhat idealised American-Indian partly nomadic culture. What sort of world is projected? One in which knowledge of animal and plant life are important. Genuine knowledge is had, for example migratory routes, what plants have what qualities, etc. Turning to the reflective side, how does their world affect their self conception? We can see the answer to this, claims Ihde, in cults of various hunted animals, such as buffalo and snake. His point is that one factor stands out in all this; the initiate into this ritual identifies with or becomes like that to which he is primarily related. He thinks himself brother to the wolf, or, in general, brother to the animal kingdom. In the second variation, Ihde takes an idealised agricultural society. Animals and plants have been domesticated and life now revolves around rhythms or life cycles of grain or animal production, fertility, birth, maturity, slaughter, planting, growth and harvest. Increasingly these people project this form onto a wider set of phenomena, such as the projection of repeated cycles into the heavens themselves. In the ritual of this society, again we see the need to repeat and be like the basic pattern of life and the world that is projected. The rites of an agricultural society are “great cycle” rites that emphasise fertility, and identification with plant or animal life. The third variation is the contemporary world. In this case the world is a technological one. There can be little doubt of the dominant interpretation of our world. It has been de-animated of both the nomadic and agricultural projections. But in positive terms it has been mechanised. For a thorough history of this I refer the reader to Lewis Mumford’s Technics and Civilisation (Mumford 1934). Now we imagine ourselves as contrivances of pumps, levers and electrical systems. What is this but a technological projection of world? So Ihde’s contention is that our world and self-understanding in fact reflect the very form of existential praxis in which we have been engaged for several centuries (ibid., p. 131). He suggests that, far from clearly showing that twentieth-century life

130


is superior, regarding self-understanding, to all other forms of human life, the form of self-interpretation is invariant through all three examples. We project in terms of some focused form of existential praxis which influences, if not sets, the selected forms of knowledge which we will regard as central. Richard Lewontin opens his book The Triple Helix (Lewontin 1998) with a discussion of the role of metaphor in understanding science. Although he is specifically speaking about metaphor in biology, we can easily see his general point. He writes for example that physicists talk of waves and particles even though there is no medium that those waves move in, and no solidity to those particles. Biologists speak of genes as blueprints, and DNA as information. Indeed the entire body of modern science rests on Descartes’ metaphor of the world as a machine: “I have hitherto described this earth and generally the whole visible world, as if it were merely a machine in which there was nothing at all to consider except the shapes and motions of its parts.” (Descartes 1983, Part IV)

While we cannot dispense with metaphors in thinking about nature, says Lewontin, there is a great risk of confusing the metaphor with the thing of real interest. We cease to see the world as if it were like a machine, and take it to be a machine. The result is that the properties we ascribe to our object of interest and the questions we ask about it reinforce the original metaphorical image, and we miss the aspects of the system that do not fit the metaphorical approximation.

3 Heidegger The thesis of “reflective self-understanding” occurs most notably in Heidegger’s essay “The Question Concerning Technology” (Heidegger 1977). I will give a brief summary of some of the ideas in it, to allow me to approach his formulation of reflective self-understanding. His main thesis is that in the modern world, defined as a technological one, the character of technology is such that it demands a view of all things as resource, that is, in terms of what they can be used for. But not mere resource. For Heidegger, what is characteristic is not just that people view all things in terms of what can be done with them, since this is a feature of humans since the dawn of time. Rather, he feels that a special type of resource is at work here, and he calls it Bestand or “standing-reserve”. The difference here is that instead of seeing something as useful for something else, things are seen as useful per se, that is, things are stored up as resources which are now “to-hand” and ready for any use whatsoever. As an aside, and for the purposes of better understanding this concept, let’s compare it to the idea of Capital in Marx. Capital is like Bestand in that it demands that all things be conceived of in terms of


131

it. Everything is translatable into monetary terms for Marx. In “The Power of Money in Bourgeois Society” in Marx’s Manuscripts of 1844,5 or Paris Manuscripts, money is described as the overturner and inverter of all qualities.6 It is the similarity in structure which is interesting here. As in Capital, and also in Marx’s concept of reification as developed by Lukacs, so Bestand is a universal and totalizing way of understanding our world. This is similar to Lewontin’s point, above, about metaphor, that is, the conception of all things in terms of their resource value is totalising, and so it becomes less and less possible to view things aside from their value in terms of standing reserve. This even affects what we come to see as real and not real, according to Heidegger. To quote Max Planck, “that is real which can be measured”.7 Now we are in a position to understand Heidegger’s formulation of the principle of reflective self-understanding. Keeping Ihde’s earlier thesis in mind, we can see that for Heidegger, if we have limited our understanding of the world to that of everything as resource, then we risk viewing first each other, and then even ourselves as resource: human resources! This is what Heidegger describes as the danger at the heart of the essence of modern technology. One of the outcomes of the reflective self-understanding found in the age of technology, according to Heidegger, is that the conception of the mind and thinking, and indeed man in general, becomes one based on the machines and technologies that populate our world. The fate of modern philosophy, for Heidegger, is to become nothing but what he calls “cybernetics”. Consider the following passage from the interview 5

Marx (1971). Marx quotes Goethe among others to illustrate. If I am lame, but I own six stallions, then their strength is my property. Strength can be translated into money. “Six stallions, say, I can afford, Is not their strength my property? I tear along, a sporting lord, as if their legs belonged to me” (Goethe, cited in Marx 1971).The universality of the property of money is its omnipotence. Marx writes “That for which I can pay, that am I. The extent of the power of money is the extent of my power.” (Ibid., p. 167.) In this way, money is the abstract and alienated ability of mankind. He quotes Shakespeare to show that money is the bond of all bonds, and the dissolution of all ties. It is the universal agent of binding and separation. It is the Galvano-chemical power of society. All social relations, be they good, and helpful to advancement, or bad, and a constraint on advancement, are made and dissolved by money. Money makes brothers of impossibilities. If a man is unattractive, he can pay for a partner. If he is dishonest or badly thought of, this can be reversed since in modern human society monetary success carries with it respect and reverence. “Money is the supreme good, therefore its possessor is good.” (Ibid., 167.) If he is stupid or dull, or untalented, he can control many intelligent and talented people. And surely, argues Marx, if one can control intelligent and talented people, one is therefore more talented and intelligent then them. After all, the extent of the power of my money is the extent of my power. Money is one thing for Marx, a universal operator. 7 Cited in Heidegger (1977, p. 169). 6

132


published in the German magazine Der Spiegel on May 31, 1976, ten years after it was conducted:8 HEIDEGGER: The role philosophy has played up to now has been taken over by the sciences today. […] Philosophy dissolves into the individual sciences: psychology, logic, political science. SPIEGEL: And what takes the place of philosophy now? HEIDEGGER: Cybernetics. The above claim is situated within a discussion the place of philosophy in an age defined by technology. Heidegger proposes that philosophy is in fact over, and it has been replaced by cybernetics. But what might Heidegger mean by “cybernetics”? Without going into the topic too much, cybernetics is generally a field that studies systems, networks and control. In particular, it includes theories of connectionism, popular with current AI research. Connectionism is an approach in the fields of cognitive science, neuroscience, psychology and philosophy of mind. Connectionism models mental or behavioural phenomena as the emergent processes of interconnected networks of simple units. There are many different forms of connectionism, but the most common forms utilize neural network models. We can view connectionism as an alternative to computationalism, the thesis that the mind is a sort of Turing machine, and its work is the manipulation of symbols. This is the thesis behind what Dreyfus calls “Strong AI” or “GOFAI”, and it is the model that he attacks based on his reading of Heidegger. What Heidegger’s criticism of cybernetics amounts to, is a thesis that all our attempts to understand ourselves nowadays are imposed by a technological worldview, which, as we saw in “The Question Concerning Technology”, forces all things to be viewed as part of a vast network of technological resource, even ourselves. This “enframing”, which Heidegger calls “Gestell”, decides in advance what is real and what is not. So in an essay accompanying “The Question Concerning Technology”, Heidegger claims that “Science challenges things to take their place as part of a network of cause and effect” (Heidegger 1977, p. 167). Furthermore, in “The Age of the World Picture” Heidegger says that each age has a world picture which determines how things are viewed. “Wherever we have the world picture, an essential decision takes place regarding what is, in its entirety. The being of whatever is, is sought and found in the representedness of the latter” (Heidegger 1977, p. 130).

8

Reprinted in Wolin (ed.) (1993, p. 108).


133

Here we have the philosophical position that, based on the reflective selfunderstanding thesis, all our attempts to understand ourselves are based on technological metaphor. We talk of nervous system and brain “wiring” and “information storage”. Even the most sophisticated accounts of the mind that have moved away from the input/output processor model now talk about “massively parallel processors”. But this has been going on for a long time, as is argued by David Bolter (1984) in Turing's Man. Bolter argues that the computer, as a defining technology of our age, is a technology which captures the imagination of poets and philosophers and in doing so helps redefine how an age sees itself, rather as the steam engine in the Industrial Revolution, the clock in the Renaissance and the potter's wheel in ancient Greece did. Again, based on Ihde’s thesis, the dominant technology of an age conditions our self-conception. One has only to work intimately and for extended periods of time with a computer to see that in doing so new metaphors and ways of seeing suggested by the computer become internalised. Our metaphor for the mind has changed a number of times even over this century. As Rodney Brooks, director of the MIT AI lab, and one of the speakers in Steve Reich’s recent composition Three Tales, remarks: “We’ve always thought about brains in terms of our latest technology. At one point our brains were steam engines. When I was a kid, they were telephone switching networks. Then they became digital computers. Then massively parallel digital computers. Probably out there now there are kids’ books which say that our mind is the World Wide Web.” (Reich 2002.)

This work of music (Three Tales) is a mine of wonderful sound-bytes concerning the topic at hand. At the same time as the speaker above is talking about metaphor, he is cut in with Richard Dawkins who maintains that “we and all other animals are machines created by our genes”, and: “DNA is a molecule which carries coded information in exactly the same way as the computer tape. If you want to understand life, don’t think of warm breath, or a spark of life, think about information technology” (ibid.).

Finally, he confesses that: “I don’t think there is anything that we are, that is in principle deeply different from what computers are” (ibid.). Writing about technological metaphor, David Edge says: “The process whereby we construe an uncertain, obscure, or puzzling area of experience in terms of one both familiar and apparently (in at least some respects) similar, the displaced pattern acting as a metaphorical redescription of the unfamiliar, has been shown to be central to many key scientific innovations. Technological devices are significant components of our familiar, everyday world: they therefore share in this process, forming the literal basis of metaphors which give

134

Content, Consciousness and Perception implicit, tacit structures to our thought and feeling—that “fill our consciousness”. Indeed, many of the most influential of the theories of modern science have an explicit origin in such “technological metaphor.” (The pervasive metaphors of cybernetics, reconceptualising the brain and society in terms of the behaviour of computers and other electrical networks, offer one striking example: how often do you hear the term “feedback”?) We would do well to explore the extent and the dynamics of this process by which our imagination comes to be dominated by those very devices which we devise in order to dominate and control our environment and human society”. (Edge 1990, p. 79.)

This way of thinking is so pervasive that even when such technological metaphors break down, the problem is still put in terms of the inadequate metaphor. So the technological cybernetic way of looking at brain function derives from the hardware of control technology. And the problems with it are ones such as the brain’s “impenetrable black box”, and the “mysterious nature of human encoding and decoding”. Judith Schlanger (1970) argues (in a paper on metaphor in molecular biology) that the successful metaphor does not merely provide answers to pre-existing questions, but by radically restructuring our perception of the situation it creates new questions, while also largely determining the answers. “The Cybernetic analogue provokes and investigates its own theoretical elaboration” (Schlanger 1970, p. 21). As we have already seen, this is a theme that is shared by Heidegger, who frequently asserts that modern science and technology predetermine what questions are asked and what answers are possible. All in all, we have a thesis in Heidegger that goes as follows. Based on the theory of reflective self-understanding, we understand ourselves technologically as a by-product of the technological age. This is what Heidegger calls the ultimate danger, in “The Question Concerning Technology”, that Man’s essence will be determined by technology. His criticism of the transformation of philosophy into cybernetics can be taken to be symptomatic of this also. Therefore, I take it that Heidegger’s views are incompatible with Okrent. What Heidegger’s philosophy implies about the AI project is not that it is impossible, but that it embodies the history of a mistake; we have been misled, we have followed a model of selfunderstanding that we did not create, but were given by our age. The AI project, according to Heidegger, has not engaged in any sort of authentic questioning about the being of man, and what understanding, intelligence and the mind really are. Heidegger would reject the idea of any digital computer having a mind, because it is merely an understanding of things that is foisted upon us by a technological age that forces us to think of all things in its terms. But can’t we just keep the original ammunition that Dreyfus and Okrent have drawn from Heidegger’s philosophy, that is, the phenomenon of Being-in-theWorld, and reject all this stuff about reflective self-understanding? Perhaps we could run the common assertion that Heidegger (like Wittgenstein) is a sort of


135

Jekyll and Hyde character; we want Being and Time, but not all the strange later stuff about technology, language and thinking. No. The thesis of reflective self-understanding that occurs clearly in the work on technology (firmly part of Heidegger’s “later” philosophy) also occurs right at the heart of Division One of Being and Time, in the very passages that the pragmatists appeal to, to make their arguments against Strong AI. It is bound up in Heidegger’s idea of Being-in-the-World itself. The most striking example is early in Being and Time, when, speaking about history and tradition, Heidegger talks about Dasein falling back on its world, and interpreting itself in the reflected light of that world.10 This, of course, calls our attention to the later chapters on “falling” and the everyday character of Dasein, where the main thesis is that the individual understands himself in terms of a self that is formed by the expectations of others, a self that is properly that of the “herd” around him (noting Nietzsche and Kierkegaard’s influence). It is the self of Das Man, that is, the “they” self. Later in the book, at the end of Division One, I would also argue that Heidegger’s section on truth gives us the bones of the later thesis of reflective self-understanding. Without getting too far into what is an extremely difficult excursus on the history of the concept of truth since Aristotle, and the argument that all of it presupposes his own phenomenological account of truth as disclosure, we find in this section the thesis that one’s own constitution determines the how of this disclosure, that is, how entities appear to us. This is the basis of Heidegger’s hermeneutics of foreknowledge, or preconceptions conditioning understanding. The idea is that something of us goes into how we uncover entities in the world. If we skip back a few sections in Being and Time, deep into the section which the pragmatists make most use of, the section on “Being in the world as the basic state of Dasein”, Heidegger talks about the history of philosophy, and mistaken conceptions of the self. Although he just talks about the form this takes, and how it happens, without giving examples, it is safe to assume that the usual suspects, namely Descartes and anyone who has a mechanistic account of things, is under attack here. We might also include, then, those people that insist on understanding the mind, consciousness, etc., by means of mechanical metaphor. He writes: “On the other hand, this ‘seeing in a certain way and yet for the most part wrongly explaining’ is itself based upon nothing else than this very state of Dasein’s Being, which is such that Dasein itself—and this means also its Being-in-the-world—gets 10

Heidegger (2000, pp. 42-43): “Our preparatory Interpretation of the fundamental structures of Dasein with regard to the average kind of Being which is closest to it (a kind of being in which it is therefore proximally historical aswell), will make manifest, however, not only that Dasein is inclined to fall back upon its world (the world in which it is) and interpret itself in terms of that world by its reflected light, but also that Dasein simultaneously falls prey to the tradition of which it has more or less explicitly taken hold”.

136

Content, Consciousness and Perception its ontological understanding of itself in the first instance from those entities which it itself is not but which it encounters ‘within’ its world, and from the Being which they possess.” (Heidegger 2000, p. 85.)

In this section Heidegger is trying to advance the thesis that earlier attempts to understand the mind, the being of humans, etc., have all made the mistake of understanding themselves in terms of other non-human things, which for Heidegger have a being which is totally and essentially other to that of Dasein. This is one of the main theses of Being and Time. I think this passage, which blames the misconception of the self on the propensity of Dasein to interpret itself in terms of its world, and also the section on truth, whereby Dasein’s modes, or ways of interpreting, or again, preconceptions, all go in to how the object is disclosed to it, can be brought together to show that the idea of reflective self-understanding is not just a relic of the later philosophy that can be ignored by the pragmatists, but rather it is nascent at the very heart of the parts of Heidegger upon which they base their claims. Therefore I conclude that if a reading of Heidegger is to be used to attack AI projects for conceiving of people as input/output processing programs rather than conceiving of them as being embedded, or practically involved, as Dreyfus does, then there is a further implication from Heidegger for AI, that is the prevalence of the technical-metaphorical image of humans, thrust upon us by our age. I conclude that, far from being separate occurrences, the second is a follow-on from Heidegger’s conception of Being-in-the-World, thus it is not open to the pragmatists to take one half of Heidegger, and not the other. If the technological metaphors used to describe human thought are a symptom of the all-encompassing force of technology on the way we view the world, then we must conclude that Heidegger would reject all attempts to model the mind (where “model” is taken as being technological), and in particular Okrent’s second claim, that is, that some digital computer could have a mind. I have not of course shown that it is impossible for a digital computer to have a mind, and in that sense, my paper is perhaps misleading. But I do not believe that Okrent et al. are merely claiming that a computer could satisfy the conditions for being minded, since this is a claim about basic possibility, a mere “what if” claim, which, without holding out for some mystical properties of the mind, it would be hard to deny. The argument is more sophisticated. It wants to say that Heideggerian concerns have no bearing on the project of AI along the lines of a computer model. Okrent wants to say not just that a computer could be a mind, but that despite Dreyfus’s attack, AI on the digital computer model remains essentially on the right track. Nor is Heidegger’s position genetically fallacious. He would not claim that the computer model of the mind is wrong because it came about via reflective selfunderstanding, but that our self-conception itself is misguided, being based on the dominant feature of our modern world, technology. This itself does not prove that a digital computer could not be a mind, but it explains why we might come to


137

imagine it that way, and base our research projects on such imaginings. In that sense, this criticism of Heidegger’s is genetic, but not fallacious. Many of Heidegger’s criticisms of the philosophical tradition consist of a genealogy whereby he claims that such and such a commonplace philosophical concept (be it time, truth, the self, death, being) is really a confusion based on a mistaken conception somewhere in the tradition. Of course, showing that a certain concept is the history of a mistake does not prove its impossibility; as much Gettier-example discussion has shown, we may still hit on the truth not only accidentally but because of a mistake. But those who want to keep the computer metaphor of the mind and yet subscribe to Heideggerian insights to help AI on its way need to say that they have the right conception, for the right reasons. That Heidegger holds a theory of reflective self-understanding with respect to technological metaphor shows that even if they are on the right track, it is for the wrong reasons, and this contribution of Heidegger’s to AI is intimately bound up with the more often quoted contributions.

Bibliography Bolter, D. (1984). Turing's Man: Western Culture in the Computer Age. London: Duckworth. Dreyfus, H. (1972). What Computers Can’t Do: A Critique of Artificial Reason. New York: Harper and Rowe. ——— (1991). Being-In-The-World. Cambridge, Mass.: MIT Press. ——— (1992). What Computers Still Can't Do: A Critique of Artificial Reason. Cambridge, Mass.: MIT Press. Descartes, R. (1983). The Principles of Philosophy. London: Reidel. Edge, D.O. (1990). “Technological Metaphor and Social Control”, in L. Hickman (ed.), Technology as a Human Affair. London: McGraw-Hill. Haugeland, J. (1985). Artificial Intelligence: The Very Idea. Cambridge, Mass.: MIT/Bradford Press. Heidegger, M. (1977) The Question Concerning Technology and Other Essays. London: Harper & Row. ——— (1993). “Only a God Can Save Us”, in R. Wolin (ed.), The Heidegger Controversy: A Critical Reader. Cambridge, Mass.: MIT Press. ——— (2000). Being and Time. Oxford: Blackwell. Ihde, D. (1990). “Technology and Human Self-Conception”, in L. Hickman (ed.), Technology as a Human Affair. London: McGraw-Hill. Lewontin, R. (1998). The Triple Helix: Gene, Organism and Environment. Cambridge, Mass.: Harvard University Press. Marx, K. (1971). The Economic and Philosophic Manuscripts of 1844. New York: International Publishers.

138


Minsky, M. (1986). The Society of Mind. New York: Simon and Schuster. Mumford, L. (1934). Technics and Civilisation. New York: Harcourt Brace. Okrent, M. (1996). “Why the Mind isn’t a Computer Program (But some Digital Computer could have a Mind”, in The Electronic Journal of Analytic Philosophy, 4. Reich, S. (2002). Three Tales, Hindenburg+Bikini+Dolly. Accompanying Video by Beryl Korot. Nonesuch Records. Schlanger, J. (1970). “Metaphor and Invention”, in Diogenes 69: 12-27. Wolin, R. (ed.) (1993). The Heidegger Controversy: A Critical Reader. Cambridge, Mass.: MIT Press.

CHAPTER EIGHT CAUSAL EXCLUSION AND OVERDETERMINATION

MARKUS E. SCHLOSSER

This paper is about the causal exclusion argument against non-reductive physicalism.1 Many philosophers think that this argument poses a serious problem for non-reductive theories of the mind—some think that it is decisive against them. In the first part I will outline non-reductive physicalism and the exclusion argument. Then I will distinguish between three versions of the argument that address three different versions of non-reductive physicalism. According to the first, the relation between mental and physical events is token-identity. According to the second, mental events are distinct from physical events, but the latter metaphysically include and determine the former. And on the third version, mental and physical events are entirely distinct. I will argue that the causal exclusion argument is not decisive against non-reductive physicalism in any of the three versions. According to non-reductive physicalism, mental events are dependent on physical events. Causal exclusion and overdetermination, however, requires distinct and independent causes. I will argue that the burden of proof lies with the opponents of non-reductive physicalism, who have to explain how metaphysically dependent events can possibly overdetermine an effect or exclude each other from being causally efficacious.

1 Non-reductive physicalism Assume that there is a true theory of the mind that employs intentional vocabulary, and call that theory psychology. Non-reductive physicalism is the conjunction of physicalism and the claim that psychology is not reducible. 1

I would like to thank my respondent Zoë Payne, the audience of the Mind2005 conference and Sarah Broadie for their helpful comments and remarks.

140


Physicalism, one may think, says that everything is physical. It is common to distinguish physicalism from claims such as that there are only material objects or that everything is composed of matter. Physicalism refers to physical things, rather than material ones, and it leaves it up to the physical sciences to specify what counts as a physical entity. Further, the view concerns not only objects and what they are composed of, but also events and properties. Now, if non-reductive physicalism is to be a consistent position, then physicalism cannot be construed as the view that all objects, events and properties are physical. In order to see why, let us have a brief look at the notion of reduction involved. In the philosophy of mind it is common to talk about the reducibility of mental types, properties or kinds. However, it is also common to point out that reduction is an issue that concerns theories and their explanatory power or autonomy. How do these claims hang together? On the standard model of scientific reduction, whether a theory T is reducible to the theory T* depends on whether the properties that T ranges over can be identified with some of the properties T* ranges over. That is, T is reducible to T* if and only if T-types are identical with T*-types. On that view, psychology is reducible to some non-mental theory T* if and only if mental types are identical with (non-mental) T*-types. That claim, however, has been contested as too strong. Some philosophers think that type-identity is not necessary for reduction, since weaker bridge-laws—laws that do not claim type-identities—are sufficient. What is uncontested, though, is that if psychology is not reducible, then mental kinds cannot be identified with non-mental kinds. So, according to non-reductive positions, there are mental properties that are not identical with physical properties. Hence, non-reductive theories are incompatible with a view that says that all properties are physical. If it is to be compatible with non-reductionism, physicalism must be specified accordingly. A common suggestion goes along the following lines. Physicalism says, firstly, that all concrete objects are composed of, or constituted by, physical entities, and, secondly, that all properties and events are dependent on physical properties and physical events.2 Physicalism says, in other words, that physical particulars and properties are the basic or fundamental particulars and properties.3 2

Compare, for instance, Beckermann (1992, pp. 1-2) and Crane (1995, pp. 211-212). Note that the notion of asymmetric dependence is itself in need of explanation. Some have therefore suggested construing the relation in terms of composition or constitution only. The problem with that strategy is that it is difficult to see how it can be applied to objects as well as events and properties (compare Crane, ibid., p. 212). Others have suggested an account of the dependence involved in terms of realisation of mental properties by physical properties (compare Kim, 2000, pp. 19-24). 3 That is the standard account of non-reductive physicalism, which faces the problem of causal exclusion. There is, however, an alternative construal that might avoid the exclusion problem altogether. Arguably, we can distinguish between two different conceptions of


141

2 The causal exclusion argument The so-called causal exclusion argument has been formulated in different ways. A first difference concerns the kinds of entities that are said to exclude each other. On some formulations the exclusion concerns mental and physical properties, others talk about mental and physical events, and some formulate the argument simply in terms of mental and physical causes and effects.4 A second relevant difference concerns the mode of causal exclusion. Some philosophers insist that what is at stake is the causal efficacy of mental events or properties. Others, however, say that the efficacy of physical events excludes the causal relevance of mental events or properties. What is common and central to all formulations of the argument, though, is the following intuition concerning causal exclusion and causal overdetermination. Suppose that c is sufficient to cause the occurrence of e, and that e has another cause c*, which is distinct from and not part of c. Let us say that an event e1 is a sufficient cause of the event e2, if e1’s occurrence is, in the circumstances, sufficient for the occurrence of e2. And e1 is a partial cause of e2, if e1 is a cause of e2 in the sense that e1 is itself not sufficient for the occurrence of e2, but it is part of a complex event, which is sufficient for the occurrence of e2. Now, c*, the additional cause of e, is either a sufficient or a partial cause of e. If c* is sufficient, then c and c* exclude each other from being the cause of e, because they overdetermine its occurrence. If c* is a partial cause of e, then c excludes c* from being causally efficacious, since c is already causally sufficient for the occurrence of e. (I will call causes that exclude each other or overdetermine their effects in that sense, rival causes). These intuitions concerning exclusion and overdetermination are considerably strong and straightforward only insofar as they are formulated in terms of causes reduction. On the first, reduction is an ontological issue insofar as it concerns the relation between mental and physical properties. On the second conception, reduction concerns the relation between theories (their vocabularies and explanatory powers). Non-reductive physicalism is clearly committed to non-reductionism in the second sense. It is not clear, though, whether it is also committed to ontological non-reductionism. If it is not, then it is not committed to the claim that there are irreducible mental properties. I can agree that reduction is, first and foremost, a relation between theories. The question, however, whether the reduction of a theory presupposes ontological reduction is beyond the scope of this paper. I will assume, throughout, that the standard construal of non-reductive physicalism is correct. 4 Formulations in terms of causes and effects can be found, for instance, in Lowe (2003) and Merricks (2001). For formulations in terms of instantiations of properties compare Kim (1993 and 2000), Crane (1995) and Menzies (2003). Kim and Yablo (1992) think that it is of no significance whether the argument is put in terms of causes, events or property instantiations.

142


and effects. Our intuitions are far less straightforward with respect to causally relevant properties. It is not obvious whether instantiations of properties can exclude each other, or overdetermine effects, in the same way as causes. To assume that they can is a substantial—and controversial—additional assumption. That is why I introduce the causal exclusion argument purely in extensional terms: formulated in terms of causes and effects only. The three basic assumptions behind the argument are the following. (1) Mental Causation: Mental phenomena cause physical phenomena. (2) Causal Closure of the Physical: Every physical effect has a causally sufficient physical cause.5 (3) Exclusion of Causal Overdetermination: Causal effects are, usually, not causally overdetermined. The causal exclusion argument goes as follows. Every version of physicalism is committed to the three claims just presented. A non-reductive version of physicalism, however, is incompatible with the conjunction of them. Given that only events can be causes, (1) says that some mental events have physical effects. Assume that the mental event m is a cause of the physical event p; m is either a partial or a sufficient cause of p. Applying (3), we exclude that m overdetermines the occurrence of p (we assume that p has only one sufficient cause, if it has a sufficient cause). So, if p has a sufficient cause, c, then m either is c, or m is not a sufficient cause of p. According to (2), p has a sufficient physical cause. Hence, c is a sufficient physical cause of p. Given that, m cannot be a sufficient cause, but it must be a partial cause of p. Partial causes are parts of sufficient causes. Since c is the only sufficient cause of p, m must be part of c. But since c is a physical—that is, non-mental—cause, m must be a physical cause of p, contrary to the assumption that m is a mental cause of p. The argument shows that the assumptions (2) and (3) exclude the causal efficacy of mental events, contrary to (1). The exclusion argument is generally considered to be a very powerful argument that constitutes a serious problem for non-reductive physicalism.6 It is 5

Generally, by causally sufficient I mean sufficient either for the occurrence of the effect or sufficient to determine its chance. In this paper, though, I have to restrict my considerations to the deterministic case. So, in what follows causally sufficient will mean sufficient for the occurrence of the effect. 6 A well-known response on behalf of non-reductive physicalism is a reductio ad absurdum to the conclusion that there must be something wrong the argument. There is no obvious reason to deny that the argument applies to the special sciences in general. If the argument can be generalised, it entails that, for instance, the efficacy of chemical, physiological and biological events is excluded by the efficacy of physical events—which is absurd. Proponents of the exclusion argument dismiss this reductio as evasive. What is required,


143

acknowledged that there are different versions of non-reductive physicalism that require different versions of the argument. But it is usually thought that differences with respect to the details do not diminish the main thrust of the argument. In the following I will have a closer look at three different versions of the argument. I will show that the differences between them are significant, and I will argue that the causal exclusion argument is not decisive against any version of non-reductive physicalism.

3 Events and property instantiations I assumed that the relevant mental phenomena are mental events. Philosophers of mind, though, often talk about mental states and properties. It is common to use the term mental events in a broad sense that includes mental states.7 We would, then, obtain two versions of the argument; one in terms of mental events, and the other one in terms of the instantiations of mental properties (for no one should expect the properties themselves to have a causal role). Little significance has been given to this distinction. It has been assumed that the argument is equally compelling in both versions. Let us first consider the version in terms of mental events. The causal exclusion problem would dissolve under the assumption that mental kinds are identical with physical kinds. But this solution is obviously not an option for non-reductive physicalism, which is committed to the rejection of typeidentities. Another possibility, though, is identity between mental and physical event-tokens. There are two ways to distinguish events as mental events. An event is a mental event just in case it has a mental description, or, alternatively, just in case it has a mental property. I shall assume that the two definitions are equivalent. Given that, it is certainly possible that mental events are identical with physical events, even though mental properties cannot be identified with physical properties. For it is possible that one and the same event has both a mental and a physical description (or property).8 Further, it is possible that the physical properties of such an event include those properties that realise and determine the event’s mental properties. they insist, is a metaphysical solution to the problem of mental causation. Or, at the very least, non-reductive physicalists must be able to spell out what is wrong with the argument. Compare Kim (2000) and Crane (1995). 7 Compare for instance Horgan and Tye (1988), who say that they are following a “frequent recent practice” by using “event” in a sense that includes “states, process, and the like” (p. 427). 8 Davidson has famously argued that this not only possible, but actually the case: every event that has a mental description has a physical description. See Davidson (1980), essays 11 and 12.

144


That is, it is possible that it is one and the same event that has both the mental properties and the physical properties on which they depend. Suppose that this is generally the case. Then the non-reductive physicalist can say that the mental and physical events are not rival causes, because they are one and the same event— being token identical, they share causal powers. If we consider property instantiations instead of events, that move—the appeal to token-identity—is not available. For what goes hand and hand with that alternative version is a particular view on the nature and individuation of events, according to which events are property instantiations. On that view, events are properties instantiated by a substance at a time. That rules token-identity out, since the instantiation of a mental property and the instantiation of the underlying physical property constitute distinct event-tokens. It seems that we have identified a significant difference between the two versions of the exclusion argument. But it merely seems so, the opponents insist. According to Davidson’s theory, which is the token-identity theory, the events in question instantiate a causal law only under their physical description. We are warranted in regarding them as cause and effect only insofar as they are covered by a physical law. But that means that they are causally efficacious only in virtue of their physical properties; they cause what they cause not in virtue of being mental events. Therefore, appeal to token-identity does not help, and the difference between the two versions is of no significance. Consider, for instance, how Stephen Yablo has put that point. “To reply with the majority that mental events just are certain physical events, whose causal powers they therefore share, only relocates the problem from the particulars to their universal features […]. Mental events are effective, maybe, but not by way of their mental properties; any causal role that the latter might have 9 hoped to play is occupied already by their physical rivals.”

A few things, however, have been overlooked in this diagnosis. In my alternative analysis I will distinguish between three versions of the argument, whereby each version results from a combination of non-reductive physicalism with a particular view on the individuation of events.

3.1 Version one The first version of the argument is directed at the already mentioned tokenidentity theory. On that view, the mental events and the underlying physical events, which realise and determine the mental events, are token-identical. The charge against this view is that such events cause their effects only in virtue of their 9

Yablo (1992, pp. 248-249).


145

physical properties—the events in question have physical effects, but not in virtue of being mental events. Let us assume, for the sake of the argument, that events cause their effects in virtue of some of their properties.10 Why do the events in question have their effects not in virtue of their mental properties? How can the opponent justify the claim that they are efficacious only in virtue of their physical properties? We can distinguish between two arguments for that claim. The first argument appeals to a connection between the causal role of properties and causal laws. It is assumed, firstly, that an event c causes the occurrence of the event e in virtue of having the property P if and only if P figures in the causal law that covers e and c (more precisely, if and only if there is a causal law according to which an event’s being P is nomologically sufficient or relevant for the occurrence of e). Further, if c causes e in virtue of having P, P is said to be a causally relevant property with respect to e’s causing c. And it is assumed, secondly, that there are no psychological laws or regularities that ground causal claims about mental events; it is assumed, in other words, that psychological anomalism is true. From that it follows that no event causes an effect in virtue of having a mental property. That first argument, however, is not decisive for the following reasons. Firstly, the second assumption—psychological anomalism—is rather controversial. It is in need of independent justification and cannot just be assumed. The problem is that most philosophers who deploy the exclusion argument against non-reductive physicalism also argue for some form of reductive physicalism.11 But it is hard to see what independent reasons a reductive physicalist might have to hold the second assumption. In fact, reductive physicalism seems to be committed to the claim that there are psychological laws or regularities that ground causal claims about intentional states and events. Secondly, the argument shifts the focus of the debate in a way that is problematic for the opponent of non-reductive physicalism. Typically, opponents of non-reductive physicalism argue that the problem of mental causation requires a metaphysical solution, and that all proposals that establish merely the explanatory relevance of mental events or properties are inadequate or beside the point. The problem, they insist, concerns causal efficacy, 10

Davidson’s own response is that, given his view on causation and the individuation of events, it simply does not make any sense to say that a cause is efficacious in virtue of some of its properties. Causation is an extensional relation between events. Surely, in order to obtain a causal explanation we have to refer to the events using the right descriptions. But it does not follow that they are efficacious in virtue of some property (compare Davidson 1993, especially pp. 12-13). Opponents will either insist on the principle that causes are efficacious in virtue of some of their properties, or they will try to show that Davidson’s response merely relocates the problem in a way that does not help to save the view. My task, however, is not to decide on the tenability of Davidson’s response. 11 The most prominent proponent of that strategy is Kim (1997 and 2000).

146


not causal explanatory relevance. The argument, though, shifts the focus from the causal efficacy of mental events to the question whether mental properties figure in causal laws and the question whether there are psychological laws; questions which concern the explanatory relevance of the mental, rather than their causal efficacy.12 According to a second line of argument, it becomes obvious that the events in question have their effects only in virtue of their physical properties once the principle of the causal closure of the physical is understood in the right way. According to Kim, the basic idea behind that principle is that, for every physical event, one will never “leave the physical domain”, if one traces out its complete causal history.13 The causal history of any physical event consists only of other physical events. Events are physical events in virtue of having physical properties. So, the causal history of any physical event can be given in terms of other events and their physical properties only. And that means, it seems obvious, that all the causal relations that constitute that history hold only in virtue of physical properties; there is no room left for mental properties to do any additional causal work. The closure principle, however, is a purely metaphysical and extensional principle; it talks about causes and effects only. Assume, once more, that events and only events can be causes and effects. Consider a physical event e that is caused by c, and assume that e is both a physical event in virtue of having the physical property P and a mental event in virtue of having the mental property M. What licences the claim that c causes e only in virtue of having P? I fail to see why and how the closure principle, by itself, rules out the possibility that c causes e in virtue of having M. Note, firstly, that in order to hold that c causes e in virtue of having M one does not have to “leave” the physical domain. Metaphysically speaking, we cannot leave the physical domain, since each mental event is, by assumption, identical with a physical event. Secondly, the thought behind the argument cannot be that the causal relevance of mental properties is excluded by the causal relevance of physical properties. Without the introduction and justification of further metaphysical principles, the notion of exclusion—just as the notion of overdetermination—applies only to causes (that is, events).14 The 12

Further below I will say more about causal efficacy and causally relevant properties. Compare also note 14. 13 Compare Kim (2000, p. 40): “One way of stating the principle of physical causal closure is this: If you pick any physical event and trace out its causal ancestry or posterity, that will never take you outside the physical domain.” 14 Stephen Yablo says that although “causes and effects are events, properties as well as events can be causally relevant or sufficient” (Yablo 1992, p. 247, note 5). It is correct that we can talk about causally relevant events as well as causally relevant properties. But is not obvious that the sense of causal relevance is the same in both cases. An event can be causally relevant in the extensional sense of being a partial cause. In that sense, events are


147

closure principle does not entail anything with respect to the causal relevance of properties. It does not exclude the relevance of mental properties, since it does not say that physical properties are sufficient—whatever that might mean. It says that every physical event has a sufficient physical cause. But it is an event that is sufficient, not an event’s having a certain property rather than another.15 In conclusion, we can say that the causal exclusion argument is not decisive against the token-identity version of non-reductive physicalism. Opponents, however, will object that even if the response is correct, it shows only that it is possible that some events have their effects in virtue of their mental properties—it shows only that mental properties can be causally relevant. What needs to be shown, though, is that the mental is causally efficacious. What kinds of things can be causally efficacious? Causes, I suggest, and only causes are causally efficacious. And as the standard view has it, events and only events are causes. Instantiations of properties, on the other hand, can have two different causal roles. Firstly, as being the property of an event—as being instantiated by an event—a property can be causally relevant in the sense that the event causes the effect in virtue of having the property. Secondly, as being an event the instantiation of a property—by a substance, at a time—can be causally efficacious. On the present version of the argument, mental event tokens are physical event tokens. That is, the instantiation of the mental property and the instantiation of the physical property do not constitute two distinct events, since they are instantiated by one and the same event. Given the distinctions just made, it follows that instantiations of mental properties can at best be causally relevant—they are not and cannot be causes. And that means that they cannot be causally efficacious. Further, instantiations of physical and mental properties cannot be causal rivals, since causal exclusion or overdetermination presupposes distinct causes. Now, what does not follow from all that is that mental events are not, or cannot be, causally efficacious. causally relevant as causes. If one denies, as Yablo does, that properties can themselves be causes, it remains to be explained in what sense properties can be causally relevant. And whatever that sense is, it is different from the one in which events are relevant, and it is, therefore, not clear at all that the exclusion argument can be restated simply by substituting property for event, as Yablo suggests. 15 Some may construe the closure principle as saying that the occurrence of any physical event can be explained in physical terms only (only in terms of events and their physical properties). But by reading the argument in that way, the opponent is again shifting the focus from causation to causal explanation. The closure principle is a metaphysical principle. It is about causes, not about causal explanations. It does not say that physical causes have their effect only in virtue of their physical causes, nor does it say that everything can be causally explained in terms of physical properties.

148


3.2 Version two On the first version, mental and physical properties are properties of one and the same event. If we construe events as property instantiations, we obtain two further versions of the argument. On both versions, the properties are partly constitutive of events, rather than being instantiated by events. Kim, for instance, says that “the properties an event exemplifies must be sharply distinguished from its constitutive property (which is exemplified, not by the event, but by the constitutive substance of the event).”16 According to that, properties can be related to events in at least two different ways. Either the property is instantiated by the event, or its instantiation by a substance at a time constitutes the event. In the former case the property modifies the event, in latter case it modifies the involved substance. In some cases, however, it will be controversial whether a given property is constitutive of an event or not. In an example Kim compares the event Sebastian’s stroll with Sebastian’s leisurely strolling. Are there two events happening at the same time, or is there only one event has that more than one true description? On what is known as the Anscombe-Davidson view on the individuation of events, there is only one event that can correctly be described as a stroll and as a leisurely stroll. According to the view that events are property instantiations, however, whether there is only one event depends on whether strolling and strolling leisurely are constitutive properties. Further, if two properties constitute events, we must distinguish between two ways in which the events are related. For two events can be distinct, without being entirely distinct, as Kim says.17 Sebastian’s stroll and Sebastian’s leisurely stroll, for instance, are distinct events. But not entirely distinct, since the latter event metaphysically includes the former. Given that, we can now formulate the second version of non-reductive physicalism, according to which the instantiation of mental and physical properties constitute events that are distinct, but not entirely. Rather, physical events include mental events. Kim did not try to spell out how the relation of inclusion between distinct events has to be understood; he says that the notion is intuitively plausible enough.18 Stephen Yablo, though, has suggested the following. According to Yablo, the relation between mental and physical events is best understood as one between determinables and determinates, which, in turn, is best understood as an event’s essence subsuming the other event’s essence. What does that mean? 16

Kim (1993, p. 43). The distinction between distinct and entirely distinct events is supposed to block a counterintuitive proliferation of events. Compare ibid., pp. 42-46. 18 Ibid., p. 45. 17


149

Consider the determinate being red of the determinable being coloured. Being red is a specific—or determinate—way of being coloured. Similarly, Sebastian’s leisurely strolling is a determinate of Sebastian’s stroll, and the former subsumes the latter. The important thing to note is that the relation of subsumption—of one event’s subsuming another event—is precisely the relation of inclusion. I will not go into any further details of Yablo’s account.19 Rather, let us see how it can be applied to the problem of mental causation. The lesson to be learned, according to Yablo, is that mental and physical events cannot causally exclude each other, if they stand in the suggested metaphysical relation of inclusion.20 For we know, as Yablo says, that determinates and determinables are not causal rivals. Surely, by citing an object’s being red we may be able to causally explain an event, or an event’s having a certain property, which cannot be explained by referring to that object’s being coloured. But it cannot plausibly be suggested that these properties are rival causes, simply because the particular instance of being red just is the particular instance of being coloured. If we apply this lesson to the problem of mental causation, we can then say with Yablo that “any credible reconstruction of the [problem] must respect the truism that determinates do not contend with their determinables for causal influence.”21 It does not matter whether or not Yablo was successful in analysing the relation of metaphysical inclusion correctly. For, on any account of that relation, if the physical event determines the mental event by including it, the two events cannot plausibly be causal rivals. To say they cannot be causal rivals is to say that they can neither exclude each other, nor overdetermine an effect, and we can conclude that the causal exclusion argument does not apply to the second version of nonreductive physicalism.

19

As indicated, Yablo suggests construing the relation of inclusion as a relation between the event’s essences: an event e is said to subsume or include another event e*, if the essence of e* is a subset of the essence of e. Compare Yablo, 1992, section 5, especially pp. 261-262. 20 One may wonder how a physical event can include a mental event. The best way to think about it is in terms of multiple realisation. Assume that systems of type S and type T can be in mental states of type M, and that systems of the two types realise M in different ways. Then the states that realise M in systems of type S and T must be alike in certain respects, because they both realise M. But given that M is multiply realisable, we can assume that those states also differ in some respects; they realise M in different ways. Given that, we can say that the way in which the two systems realise M are specific or determinate ways of being in M. 21 Ibid, p. 259.

150


3.3 Version three On the third version of the argument, instantiations of the mental and the physical properties constitute entirely distinct events. None of the responses that have been given so far apply. The mental and physical properties in question are constitutive of metaphysically distinct events. Instantiations of them can therefore be efficacious as causes. Furthermore, it seems that they can be causal rivals, since they constitute entirely distinct events.22 Yablo argued that, if two events are not entirely distinct, in the sense that one includes the other, then they cannot be causal rivals. It does not follow that two events can be causal rivals, if they are entirely distinct. Nor is it obvious that all causes that are entirely distinct causes of one and the same effect are, or can be, rival causes. In order to decide whether they are or can be rival causes, we need to have a closer look at some of the issues involved. Intuitively, for causal rivalry, exclusion or overdetermination to occur, there have to be two or more independent causes of one and the same effect. If that is not the case, if the causes in question are not independent in the relevant sense, the exclusion argument does not apply, because the notions of causal rivalry, exclusion and overdetermination do not apply. In other words, whatever is necessary for causal rivalry, exclusion and overdetermination, is necessary for an application of the argument. In the following I will suggest a characterisation of causal overdetermination and of the involved notion of independence. I will argue that mental and physical events, construed as entirely distinct events, cannot overdetermine their effects, because they are not independent in the required sense. Given that, it is then the opponent’s burden to clarify why and how the exclusion argument applies to non-

22

Non-reductive physicalists may argue that this way of construing the relation between mental and physical events is not compatible with non-reductive physicalism. According to non-reductive physicalism, mental events depend on physical events. This dependence, one may think, is incompatible with the claim that the events are entirely distinct. But is it impossible that two events are entirely distinct and dependent? Consider the following. Brutus killed Caesar by stabbing him with a knife. At one point, Kim suggests that both Brutus’ killing Caesar and Brutus’ stabbing Caesar are generic—hence, entirely distinct— events, whereas Brutus’ stabbing Caesar with a knife merely modifies Brutus’ stabbing Caesar (compare Kim, 1993, p. 44). Intuitively, though, there is a sense in which the particular killing of Caesar depends on Brutus’ stabbing him. To decide whether the third version is consistent would require a detailed discussion of the involved notions of metaphysical distinctness and dependence, which is beyond the scope of this paper. Rather, I will assume, for the sake of the argument, that it is consistent. If it is not consistent, so much the worse for the proponents of the exclusion argument—for then the third version collapses into either version one or two.


151

reductive physicalism and to the problem of mental causation in general. Let me begin with two observations. Let us consider a clear case of causal exclusion; a case in which two causes overdetermine an effect. An example that is often used to illustrate causal overdetermination is the case in which two sharpshooters kill their victim, which happens to be one and the same person, at exactly the same time. Each shot is sufficient to cause the victim’s death, which is overdetermined by two distinct sufficient and independent causes. The first observation is that the events in the case of mental causation are not distinct in the same way as the relevant events in the sharpshooter case. In the sharpshooter case, the events are not only distinct with respect to the instantiation of properties, but they are distinct in the further sense that the properties are instantiated by distinct substances. This, of course, is not the case for mental causation; the mental and the physical property are instantiated by one and the same substance. That does not show, of course, that being instantiated by distinct substances is necessary or sufficient for causal rivalry. But it suggests that the event’s being entirely distinct is not sufficient. What is required, in addition, is that they are independent, in some sense. This brings us to the second observation. What we are assuming, and what the opponents of non-reductive physicalism usually assume as well, is that the two purported causes - the mental and the physical event—stand in some intimate relationship. All versions of non-reductive physicalism assume that mental events supervene, in some sense, on physical events, and that mental events are realised by them. It is assumed, in other words, that mental events (or properties) are, in some sense, dependent on physical events (or properties). By hypothesis, that is, the mental and their underlying physical events (or properties) are not independent. This independence, however, concerns the existence of mental events and properties. Whenever there is a mental event, there is some physical event that realises it. It is possible that there are physical events and no mental events, but it is impossible that there are mental events and no physical events. The mental depends on the physical and is realised by it, but not vice versa. What we are interested in, however, is independence of causes as a criterion for the application of the causal exclusion argument. Recall that the third assumption of the causal exclusion argument concerns causal overdetermination. The argument applies only if the notion of causal overdetermination applies. That is, for the argument to apply, there must be causes that can overdetermine an effect. I suggest having a closer look at the notion of causal overdetermination in order to get better a grasp of the relevant notion of independence. Let us begin with a

152


characterisation of overdetermination, which is adopted from Trenton Merricks’ Objects and Persons.23 Some effect e is causally overdetermined if: (a) e is caused by c, (b) c is causally irrelevant as to whether some other numerically distinct cause, c*, is a cause of e, (c) e is caused by c*, and c* is numerically distinct from c. Whereby c is causally irrelevant as to whether c* is a cause of e if:24 (d) c is not a cause of c* (that is, c is neither a sufficient nor a partial cause of c*, and c is not an intermediate in a causal chain that runs from a cause of c* to c*), or (e) c does not cause c* to cause e (nor does it cause any members of c* to cause e, if c* consists of more than one cause jointly causing e). One can agree with Merricks that this extensional construal of overdetermination is straightforward. But, unfortunately, it does not help us any further for two reasons. Firstly, Merricks talks about objects as being causes.25 For overdetermination to occur, two distinct causes have to cause the effect, and distinct causes are, on that view, distinct objects. That means that rival causes—causes that can overdetermine an effect—are numerically distinct substances. That reflects my intuition that we get a clear sense of the phenomenon of overdetermination, if the causes are associated with distinct substances. But since it does not cover the case in which two distinct events occurring in one and the same substance, it is of no help. Secondly, the definition recognises only causal relations between the two potentially rival causes; what matters, according to clause (b), is whether one cause is causally relevant or irrelevant as to whether the other cause brings about the effect. What we are looking for is a characterisation of the way in which the two causes of the effect, c and c*, can said to be dependent or independent causes. The presented definition captures the dependence between them in terms of causal relevance; that is, the dependence is construed as causal dependence. Some philosophers have tried to understand the relation between mental states and their physical realisations in causal terms. Most philosophers, however, think 23

Compare Merricks (2001, p. 58), who thinks that a definition along such lines is ‘the most literal, straightforward and natural’ definition of causal overdetermination. 24 Compare ibid, p. 57. Note that the involved notion of causal relevance is different from the notion of causal relevance as it applies to properties. Compare also note 14. 25 It is not obvious whether the view is committed to substance-causation. One might hold, for instance, that causation by objects is causation by change in objects or by the object’s having of a certain property.


153

that the dependence between the mental and the physical is of a different kind. The notion of supervenience, as many think, provides merely a minimal constraint for that relation, which is itself in need of explanation (given that it holds). A popular solution is to say that supervenience holds, because the physical states realise the mental states.26 What is needed, then, are dependence-conditions that can be applied to different relations between the physical and the mental, such as causation, supervenience, determination and realisation. An obvious candidate for that is counterfactual dependence. It is commonly assumed that there is a close connection between causal dependence and counterfactual dependence. Causal relations are said to entail or support relations of counterfactual dependence. On Merricks’ definition, c and c* do not overdetermine e, if c is causally relevant as to whether c* causes e. If c is causally relevant in that way, then whether c* causes e depends causally on c. This causal dependence entails or supports counterfactuals of the following form: (CF)

Given relevantly similar circumstances, had c not occurred, c* would not have caused e.

My suggestion is to replace the causal dependence employed in Merricks’ definition by counterfactual dependence. For what is crucial, it seems, is not the causal connection itself, but the entailed counterfactual. Consider again the case of the two sharpshooters. It is a case of overdetermination, because there are two sufficient and independent causes of the same effect. Specified in causal terms, the independence consists in the absence of causal connections between the two causes. Why is that relevant to the question of whether the effect is overdetermined? If one shooting depends causally on the other shooting, as one might say, then the two shootings are not independent causes of the victim’s death. But that is plainly circular, given that we want know in virtue of what causes are independent causes. A better answer is the following. If the two shootings are causally dependent, then one of the two sharpshooters would not have killed the victim, had the other one not done so. The case of the two sharpshooters is a case in which it is a coincidence that two causes are sufficient to cause the same effect. A causal connection would render the two shootings non-coincidental. But so would a counterfactual connection. That is why I suggest replacing the causal condition on overdetermination by a counterfactual one. Accordingly, the two causes, c and c*, do not overdetermine the effect e, if c*’s causing e is counterfactually dependent on c; that is, if CF holds.

26

Compare Kim (2000, p. 23-24).

154


We found Merricks’ definition of causal overdetermination to be too narrow, since it covers only cases in which two potentially rival causes are causally dependent. To construe the kind of dependence that is necessary for causal rivalry and overdetermination in terms of counterfactual dependence has the advantage that it covers causal as well as other relationships that might hold between events. In order to see whether the counterfactual approach will get us any further with our problem, let us see whether the relation of realisation—as it is usually employed by non-reductive theories—supports counterfactuals of the right sort. Suppose that p is the physical realisation of the mental event m, and that both p and m are causes of the event e. Is it true that had p not occurred, m would not have caused e? Let us assume that m is multiply realisable.27 We have then to distinguish between two cases. In the first case we assume that each system—or agent—falls under a certain type—or species—such that for all mental event-tokens m of type M: for any agent of a certain type, tokens of M are realised by physical eventtokens p of type P. In that case, we can limit our considerations to agents of certain types, and restricted counterfactuals of following form will hold. (CF’) For all agents s of type S: had the s-involving event p not occurred, the sinvolving event m would not have caused e, given relevantly similar circumstances. That counterfactual holds, because the occurrence of m depends counterfactually on p. If the antecedent holds, then the mental event m does not occur—and if m does not occur, then, trivially, m does not cause e. In the second case, we do not assume that mental events of a certain type are realised by all agents of a certain type in the same way—realised by physical events of the same type. In that case we have to consider the set {P1, P2, P3…} of all possible realisations of the mental state type—across all individuals of all agenttypes. In that case, counterfactuals of the following form will hold. (CF’’) For any agent s: had none of the s-involving events pi (of type Pi) occurred, the s-involving event m would not have caused e, given relevantly similar circumstances. Again, that counterfactual holds, because the occurrence of m depends counterfactually on pi. If the antecedent holds, then the mental event m does not 27

Note that non-reductive physicalism is typically motivated by the possibility of multiple realisation of mental states. It is therefore safe to make that assumption. (Moreover, if relevant counterfactuals hold in case mental states are multiply realisable, they certainly hold in case they are not.)


155

occur—and if m does not occur, then, trivially, m does not cause e. What we get, in both cases, is a counterfactual dependence between p, m, and m’s causing e. Given all that, we can conclude that mental events and their physical realisations do not overdetermine their effects since they are not independent causes. The mental event is not an independent cause, because its occurrence and its causing the effect depends counterfactually on its physical realisation. No dependency of that sort holds for standard examples of causal overdetermination. The two shootings do not depend on each other—neither causally nor counterfactually. What makes the sharpshooter case a case of causal overdetermination is precisely the fact that had one sharpshooter not killed the victim, the other one would have. They kill their victim independently of each other—and that they do so at the same time, with the same success, is a mere coincidence.

4 Conclusion I argued that the causal exclusion argument is not decisive against nonreductive physicalism, no matter whether the mental and physical events in question are construed as token-identical, distinct or entirely distinct. What is important to note is that the response to the last version applies to all three versions, since it relies on the claim that mental events are dependent on physical events—a claim that all versions of non-reductive physicalism are committed to. What causal exclusion and rivalry amount to is fairly straightforward in case there are distinct and independent causes. But the case of mental causation is not of that sort. It is not clear at all what causal exclusion and rivalry amount to, given that the mental depends on the physical. There would have to be a kind of causal exclusion that is either not tied to the outlined notion of overdetermination, or one that is not tied to any notion of overdetermination at all. In any case, it remains to be shown what sort of causal exclusion that is and how it has to be understood. We can conclude that the opponents of non-reductive physicalism have yet to show how the causal exclusion argument applies to mental causation, even if it is assumed that mental events and their physical realisations are entirely distinct.

Bibliography Beckermann, A. (1992). “Introduction—Reductive and Nonreductive Physicalism”, in A. Beckermann, H. Flohr, and J. Kim (eds.), Emergence or Reduction? New York : Walter de Gruyter. Crane, T. (1995). “The Mental Causation Debate”, in Proceedings of the Aristotelian Society, Supplementary Volume LXIX: 211-236.

156


Davidson, D. (1980). Essays on Action and Events. Oxford: Oxford University Press. ——— (1993). “Thinking Causes”, in J. Heil and A. Mele (eds.), Mental Causation. Oxford: Clarendon. Horgan, T., and Tye, M. (1988). “Against the Token Identity Theory”, in E. Lepore and B. McLaughlin (eds.), Actions and Events: Perspectives on the Philosophy of Donald Davidson. Oxford: Blackwell. Kim, J. (2000). Mind in a Physical World: An Essay on the Mind-Body Problem and Mental Causation. Cambridge: MIT Press. ——— (1993). Supervenience and Mind: Selected Philosophical Essays. Cambridge: Cambridge Studies in Philosophy. Lowe, E. J. (2003). “Physical Causal Closure and the Invisibility of Mental Causation”, in S. Walter and H-D. Heckman (eds.), Physicalism and Mental Causation: The Metaphysics of Mind and Action. Exeter: Imprint Academic. Menzies, P. (2003). “The Causal Efficacy Of Mental States”, in S. Walter and H-D. Heckman (eds.), Physicalism and Mental Causation: The Metaphysics of Mind and Action. Exeter: Imprint Academic. Merricks, T. (2001). Objects and Persons. Oxford: Clarendon. Yablo, S. (1992). “Mental Causation”, in Philosophical Review 101: 245-280.

PART III: PERCEPTION

CHAPTER NINE METAPHYSICAL DISJUNCTIVISM AND THE INTENTIONAL THEORY OF PERCEPTION

FRANCISCO PEREIRA

1 Introduction The intentional theory of perception acknowledges that our perceptual states have the ability to be about physical objects and properties that characterise our environment.1 This ability possessed by perceptual states by virtue of which they can be correct or incorrect about the world is commonly taken to be a symptom of these states having a representational nature. What is for a mental state, like a visual experience to be “representational”? There are many available theories of mental representation, but on the face of empirical research in the cognitive sciences and evolutionary considerations, it seems reasonable to argue that mental representations are best understood as neurophysiological states of the brain whose proper function is to provide information about worldly objects and the properties they instantiate. Visual experiences and the way they represent the world must be regarded as biological functions, that is, as contentful states that have been biologically selected for enhancing our survival. This is certainly something definitional and fundamental at the moment of understanding their nature.2 1

I would like to thank Professor David Papineau, Dr. Mark Textor and Christopher Bartel at King's College London, for their invaluable feedback. I am also very grateful to Matthew Conduct who provided excellent comments on this paper during the Mind2005 Conference at the University of Edinburgh. Finally I want express my gratitude to the audiences of the Mind2005 Conference in Edinburgh and the ECAP5 Conference in Lisbon. 2 I will use the expression “visual experience” to characterise those mental states that take place indistinctively when we veridically perceive, when we have illusions and when we hallucinate. I will use “perceptual experiences” or “perceptions” to pick up those episodes— veridical or illusory––in which there are particular objects or properties in the world being


159

Bearing this characterisation in mind, it is reasonable to argue that unlike theories located at the opposite end of the theoretical spectrum, such as the sensedata view, the distinctive feature of the intentional theory of perception is that it successfully acknowledges the fact that the immediate objects of experience do not need to be actual or real entities. Like any other system whose proper function is to represent how things are in the environment, our perceptual mechanisms sometimes do not represent the mind-independent world correctly. However, misrepresentation and perceptual error do not entail the actual existence of the objects that are phenomenologically manifest during an experience. Sometimes we just have visual experiences about objects and properties that do not exist. According to the interpretation I am suggesting, the minimal commitment that any intentional theory should accept involves the rejection of what we might call the “reality of appearances” principle according to which for every phenomenal property P, manifest to a subject S when undergoing a visual experience, there is always some real object or real qualitative-element apprehended by S in virtue of which P is individuated. This principle, sometimes uncritically accepted as a necessary premise in the formulation of sceptical arguments based on the possibility of perceptual error, such as the argument from hallucination or the argument from illusion, is something that the intentional theory necessarily rejects. The denial of the “reality of appearances” takes the following form: [Intentional Theory] Visual experiences represent the world as being a certain way F, whether or not there is actually something F.

The intentional content of a particular visual experience specifies the way the world must be for that experience to be veridical or non-veridical. However, there seem to be at least two different representational ways to understand perception and the intentional nature of visual experiences. On the one hand, we have theorists insisting that the actual satisfaction of the veridicality conditions is something that must be settled independently of how the visual experiences represent the world to be. In this view, the content or correctness conditions of the experiences are considered as fully independent from the actual worldly objects and properties that they represent. Genuine instances of seeing a particular object are explained by what we might call the instantiation of an experiential act of awareness plus extra causal considerations. Using Johnston’s terminology, let us call this traditional version of the intentional theory the “conjunctive” view, as a way of emphasising successfully represented by the subject. I will use “hallucinatory experience” or simply “hallucination” when there are no real particular objects or properties in the world being represented by the subject. Finally, I will use the expression “phenomenal property” as a way of referring to those aspects that characterise what it is like for the subject to have a particular visual experience of a certain type.

160


an approach that defines perception conjunctively, that is, by means of “an analysis of seeing by genus and differentia” (Johnston 2004, p. 121).3 On the other hand, we find theorists that acknowledge the intentional nature of visual experiences, but that want to deny the alleged independence of mental content from the objects and properties being represented at least in perceptual cases. Some of these theorists appeal to causal self-referential aspects involved in the content of experience; others simply rely on the so-called “singularity” of visual experiences and the role that perceptual demonstratives play in individuating the content of experience.4 Despite the differences, these theorists share the motivation of denying that the actual satisfaction of the correctness conditions need to be considered independently of whether a subject actually sees or not. This view involves the idea that genuine perceptual experiences, that is, those visual experiences that represent actual objects/properties in the world should not be analysed, but considered unitarily as being constituted by those objects and properties they represent. In short, those who take this strategy generally claim that the contents of genuine perceptual episodes are object-involving. What really interests me is that those who take the conjunctive path and those who deny it seem to agree that any intentional theory of perception has to acknowledge and somehow explain at least the following two aspects of visual experience: [Phenomenal Indiscriminability] For every veridical experience, there is a possible non-veridical experience that is phenomenologically indiscriminable from it on the basis of introspective reflection. [Commonality Thesis] Genuine perceptual experiences and hallucinatory experiences have something in common, allegedly their being mental states of the “same kind”. 3

This particular view of the intentional theory is said to be “conjunctive” because it analyses seeing as involving the combination or conjunction of two different and independent aspects, namely the instantiation of an internal visual experiential episode with content and an appropriate causal link between the objects allegedly represented and the contentful experience. Hallucinations and perceptual states are said to be of the same genus because in both cases the subject visually entertains a content of the same kind. The difference or differentia rests on how these contentful experiences are causally brought about and whether the contents actually match the environment or not. I think that in different ways Tim Crane (2001), Paul Coates (1998) and Michael Huemer (2001) are good examples of the intentionalist conjunctive approach. 4 For a theory acknowledging causal self-referential aspects as part of the content of visual experience see John Searle (1983). For theories about visual content that appeal to the “singularity” of visual content and the role that perceptual demonstratives play in perception see Paul Snowdon (1988), Matthew Soteriou (2000) and Mark Sainsbury (2005).


161

Almost everyone seems to agree that the first claim is just a fact about the limits and constrains of our subjective discriminatory abilities. The situation is not similar in the case of the Commonality Thesis, a subject in which there is widespread disagreement. Some philosophers insist that what is common to perceptual, illusory and hallucinatory visual episodes should be taken as the instantiation of an experience of the same ontological kind. Others have reacted against this thesis, raising the voice on behalf of a new philosophical standpoint about perception that simply denies that there is a “single state of affairs” obtaining across perceptual and hallucinatory instances of visual experience. This later approach is said to be “disjunctive” because according to the original formulation of the theory, any visual experience reportable using the traditional language of appearances in the form “It seems to me as if there is an F” or “It looks to me as if there is an F” can be also stated disjunctively as “either I perceive an F or I hallucinate an F”. There are several considerations that might motivate one’s acceptance of disjunctivism about visual experience. One might endorse disjunctivism as a thesis regarding the phenomenal character of experience insisting that experiences related to distinct particular objects, or experiences that do not denote any objects at all, cannot have the same phenomenal character. One could also take an epistemic version of disjunctivism, according to which perceptual episodes justify empirical beliefs in a way that cannot be equated with the role played by hallucinatory states. Alternatively, one could take disjunctivism only as a thesis regarding the representational content of visual states arguing that experiences that appear to be the same can indeed have different contents. However, all these versions may or may not follow from what I take to be the crucial motivation underlying the works of Hinton (1973), Snowdon (1988) and Martin (2002; 2004) among others. I take their motivation to be a metaphysical one according to which perceptual and hallucinatory instances should not be considered from an ontological point of view as visual experiences of the same kind. Disjunctivism in its metaphysical version could be characterised in the following way: [Metaphysical Disjunctivism] Perceptions and Hallucinations are experiences of essentially different kinds.

At first it seems that Metaphysical Disjunctivism is in tension with the basic commitments of the intentional theory, particularly with the Commonality Thesis, that tell us that there is something “common” to perceptual experiences and hallucinations. Indeed, the philosophical literature is full of discussions and arguments in which disjunctivism is presented as a theory incompatible or opposed

162


to the intentional understanding of visual experience.5 In this paper I will argue that, if we carefully reflect about the metaphysics of visual experience, we are not forced to accept this apparent antagonism between the intentional/representational view and disjunctivism. If visual experiences are mental representations with contents whose proper function is to provide information about objects and properties instantiated in our environment in order to ensure and enhance our survival, then genuine perceptual episodes and hallucinatory episodes are essentially different as Metaphysical Disjunctivism states. However, I think that this difference does not threaten the idea that there is a common psychological aspect that preserves and explains the intuitive thought according to which perceptions and hallucinations are genuine “experiential” episodes. In what follows I will consider the argument from hallucination and I will point out that, when interpreted in a certain way, the argument involves the acceptance of what we might call a Highest Common Factor account of visual experience. This Highest Common Factor view motivates the intentionalist conjunctive point of view, according to which perception must be analysed in virtue of a common internal experience with content (that is of the same kind across veridical and nonveridical instances) and extra causal conditions. Intentionalists usually accept the Highest Common Factor on the basis of causal and phenomenological grounds that seem to entail a symmetrical metaphysical understanding of those contentful visual experiences that we undergo when perceiving and those we undergo when we hallucinate. However, I will argue that this strategy rests on a very strong interpretation of the role played by proximate neurophysiological causes in visual experience and some dubious remarks about the powers of our discriminatory abilities. My strategy will consist in considering those features or arguments that are said to be decisive at the moment of embracing the conjunctive strategy in order to evaluate whether or not we can give an alternative account of perception in a way that does not require us to accept the Highest Common Factor. First, I will discuss the assumption according to which if the relevant proximate causes of a hallucinatory experience and a veridical perception are in principle the same, then the experiential outcome of the activation of those proximate causes must be 5 For example, some intentionalists do not hesitate to say that the disjunctive theory should be considered as “a major rival to the causal theory” (Lowe 2000, p. 158), or that the dispute about whether perceptual states are genuinely relational is a dispute “between ‘intentionalists’ and ‘disjunctivists’” (Crane 2006, p. 128). In the same fashion, some theorists have favoured proposals that could be considered disjunctive precisely because “in any event, perception is not representational” (Travis 2004, p. 57), or because the similarities and differences that we may find between perceptual states and sensory imagination “gives us reason to endorse the non-representational conception of experience associated with these disjunctive theories of perception” (Martin 2002, p. 380).


163

essentially the same as well. Secondly, I will make some comments regarding the limits of our introspective discriminatory abilities. It is said that if what it is like to hallucinate O is exactly like what it is like to veridically perceive O, then these experiences must be of the same essential kind. In the first case I will argue that there is room for an alternative account of the “same proximate cause, same effect” principle according to which distal objects do play an important non-causal role determining the kinds of experiences that we have. In the second case, I will remind the reader that there are some epistemic constraints with regard to what we can or cannot tell on the basis of introspective discrimination. The existence of these constraints will show us that phenomenal indistinguishability cannot be a decisive guide for typing experiential kinds. Finally, on the basis of these considerations and an adequate distinction between epistemic and ontological aspects involved in visual experience I will claim that there is a fundamental ontological asymmetry between perceiving and hallucinating that demands the rejection of the Highest Common Factor. The conclusion I want to establish is that Disjunctivism at least in its metaphysical version is fully compatible with the Intentional Theory.

2 Considering the argument from hallucination Consider the following formulation of the argument from hallucination: (i) Subject S veridically perceives a physical object O at t. (ii) When S veridically perceives O at t, S undergoes an experience P. (iii) By activating the relevant neural causes in S’s brain it is possible to produce in S a perfectly matching hallucinatory experience H, which is phenomenologically indistinguishable from P, but where there is no physical object O being perceived. (iv) Whatever kind of experience S undergoes when veridically perceiving O, the same kind of experience could occur if S were hallucinating O. (v) P and H are experiences of the same kind. (vi) If P and H are experiences of the same kind, then the naïve view that considers perceptual experiences as essentially related to mind-independent physical objects and properties is false. (vii) The naïve view about perceptual experiences is false.

The argument from hallucination as it is usually presented seems to involve two important claims: (1) that the activation of the same neural proximate causes is sufficient to bring about an experience that is phenomenologically indistinguishable from some veridical perception and (2) that there is an experiential kind which is common to veridical and non-veridical cases. The first claim seems uncontroversial especially if we consider that visual experiences just as any other states or events in the natural world, are subject to the causal laws of nature. It is not only logically, but in theory empirically possible to manipulate the causal order, the visual cortex

164


for example, in order to induce a perfectly matching hallucination. I think indeed of visual experiences as the outcome of physical processes or as mental states that even if not reducible to physical processes at least supervene on such processes. It is the second claim––clearly an interpretation of the Commonality Thesis––that I find problematic due to its different consequences depending on the way we use to establish what it is for two experiences to be experiences of the “same kind”. Some philosophers argue that if two experiences are produced by the same proximate neural causes, then these experiences must be essentially––metaphysically––of the same kind: “It is necessary to give the same account of both hallucinating and perceptual experience when they have the same neural cause. Thus, it is not, for example, plausible to say that the hallucinatory experience involves a mental image or sense datum, but that the perception does not, if the two have the same proximate—that is, neural—cause.” (Robinson 1994, p. 156.)

The thesis that perceptual and hallucinatory experiences are of the same fundamental kind, involves a particularly narrow interpretation of the “same proximate cause, same effect” principle, according to which distal causes do not play any role constituting the kind of experiences that we undergo. As a result, some theorists think that if we manipulate the causal order in order to produce a perfectly matching hallucination, then we should accept that the outcome of the activation of the relevant neural proximate causes must be the same for both hallucinations and perceptions. Notice that for this particular interpretation the idea is not just that we can have an experience that is phenomenologically indiscriminable from a veridical perception, but a stronger thesis according to which perceptual and hallucinatory states are of the same ontological kind. In other words, the commonality is in this case explained metaphysically and demands an acceptance of what we might call the Highest Common Factor view of visual experience. In order to distinguish this view from its purely epistemic formulations I define it in the following way: [Highest Common Factor] For any subject S who veridically perceives O at time t, S could in principle have the same fundamental kind of experience she is having at t, if she were hallucinating O.

Should we accept openly the Highest Common Factor View (HFC)? If we accept it, then we commit ourselves to a form of intentionalism according to which you could be in the same fundamental kind of experiential state you are in now (assuming that you are veridically perceiving) if you were hallucinating. The connection present in this interpretation between the Highest Common Factor and the Intentional Theory provides a positive understanding of both perceptions and


165

hallucinations as experiential states that share a common ontology. I am not only taking into account how these experiences are instantiated at the subpersonal level, but also and primarily how they represent the world from the subject’s point of view. According to the Highest Common Factor view, our experiences represent the world when veridically perceiving and when hallucinating in essentially the same way, given that in both scenarios subjects have an experiential outcome of essentially the same kind. The resulting Intentional Theory is of course conjunctive, in the sense that hallucinatory and perceptual experiences are of the same genus, differing only in the way they are causally brought about. If the experiences that you have when hallucinating and the experiences that you have when perceiving are fundamentally of the same kind or genus, then objects and properties in the world do not play a role constituting those experiences. The “conjunctive” version of the Intentional Theory is therefore committed to the idea that perceptual states are not essentially relational. At the beginning of this paper I said that the heart of the intentional theory rests on the denial of the “reality of appearances” by virtue of understanding perceptual experiences as states that represent the world as being a certain way F, whether or not there is actually something F. I also pointed out that the traditional “conjunctive” understanding of the intentional theory involves the idea that veridicality is something independent from how the experience represents the world to be. Indeed, this is why the conjunctive intentionalist usually appeals to a version of the causal theory of perception in order to rule out veridical hallucinations and decide whether or not the relevant experiences can be considered strictly speaking perceptual states or not. Some intentionalists do indeed accept a version of the Highest Common Factor and hold that contentful perceptual experiences are not constituted by the objects and properties instantiated in the perceiver’s environment. After all, this is why they consider seeing and hallucinating as experiences of essentially the same kind. However, I also mentioned that some advocates of the Intentional Theory do not accept this and insist that mind-independent objects and properties do constitute contentful perceptual (veridical or illusory) states. They argue that the actual satisfaction of the correctness conditions of a visual experience is not something that has to be considered independently of whether a subject really sees or not. They of course accept that both successful perceptual experiences and hallucinations are mental representations brought about by the relevant neural proximate causes. Nevertheless, they deny that this is all that is involved at least in the perceptual case. I think that they are right in claiming that perceptual states–– unlike hallucinations––should not be regarded simply as the contentful experiential effects of the activation of proximate causes reliably correlated with the environment. Indeed, theorists like Soteriou (2000), Thau (2004) and Snowdon (1988) have insisted that perceptions are in fact constituted by the worldly objects

166


and properties represented in the experience. Despite their differences they seem to agree that in those cases in which we represent actual worldly objects and properties the interesting metaphysical question about visual experience is “what more is involved in their being mental representations in addition to being reliable effects?” (McCulloch 2003, p. 7). In the following sections I will try to make sense of the view according to which some intentional visual experiences are essentially relational while others––hallucinatory states––are not.

3 Perceptual experience and distal objects One of the reasons traditionally taken to ground the conjunctive approach that accepts the Highest Common Factor view is the idea according to which the activation of the same neural causes must produce the same fundamental effects across perceptual and non-perceptual cases. If the activation of the same neural proximate causes is sufficient to bring about an experience of the same fundamental, that is, constitutive ontological kind, then it seems that the traditional conjunctive interpretation of the intentional theory is right and actual distal objects and properties do not play any role determining and constituting the kind of experience a subject undergoes when perceiving. Worldly objects and properties might play an extra role at the moment of ruling out misrepresentation on external grounds, but not as essentially determining the kind of experiences we have. Some theorists think that this is an unavoidable consequence of the already introduced “same proximate cause, same effect” principle. However, I will argue that mindindependent features do not constitute the kind of experiences we have only if we accept a particularly restrictive interpretation of the “same cause, same effect” principle. I think that the conjunctive interpretation of the “same cause, same effect” principle is not only too restrictive but also fails to include what I take to be essential for any proper understanding of perception, that is, that seeing is a cognitive achievement in which the environment is revealed and not the solitary work of the brain or the visual mechanisms. In simple terms, seeing is not a worldindependent enterprise. The challenge is to make sense of the causal aspects of perception without denying that perceiving is something that goes all the way down to the objects and properties instantiated in the environment. This would help us to understand why perceiving is a mental act essentially different from hallucinating. A possible strategy to emphasise the cognitive significance of perception is to point out the inadequacy involved in thinking that the link between physical natural processes and perceptual awareness rests merely on process causation. Johnston has remarked on this aspect, pointing out that we should understand those mental states that we have when we perceive as materially constituted by the mind-independent world and not merely as the blind outcome of physical processes.


167

“The constitutional basis for the act of awareness involved in hallucination is the state of the hallucinator’s visual system, while the constitutional basis for seeing is (as it actually turns out) the state of the visual system plus the appropriate causal influence by external things. The right causal connection does not itself cause seeing. The right causal connection guarantees that the visual channel is open so that direct visual awareness of external things takes place. There are thus distinctively different acts of awareness involved in hallucinating and seeing, individuated by different objects of awareness.” (Johnston 2004, p. 139.)

The main thought underlying Johnston’s claim is that there isn’t something like the “last” blind neurological state responsible for the act of seeing. On the contrary, seeing is for him something materially constituted by the complete dynamic process that connects the object seen to our final state of visual consciousness. What is really interesting about Johnston’s idea is that his view allows for the constitutional basis of hallucinatory states to be “part” of the constitutional basis of genuine perceptual states to the extent that in both cases the visual system is turned up in a phenomenologically indiscriminable fashion.6 Therefore, perceptual and hallucinatory states may have something in common, but this commonality according to the interpretation I am suggesting does not constitute an ontological identity of the experiential kinds. The view I am pursuing requires the denial of the restrictive interpretation of the “same proximate cause, same effect” principle, according to which if the relevant proximate causes of a hallucinatory experience and a genuine perception are in principle the same, then the experiential outcome of the activation of those proximate causes must be ontologically speaking of the same kind as well. A strategy that seems to be fully in touch with this project has been endorsed by Martin, who thinks that we should allow the objects of experience to 6 In Johnston’s particular theory, this common aspect is due to the fact that during hallucinatory experiences we are aware of sensible profiles, which are not actually instantiated in the environment. Ironically, Johnston (2004) argues against a disjunctive account of perceptual experience and insists that disjunctivism––as he understands it––has nothing to say in order to answer the following question: “What kinds of things can visual experience be a relation to so that in a transition from a case of visual hallucination to a case of seeing there need be no difference which the subject can discern?” (Johnston 2004, p. 124.) He explains the seamless transitions from hallucinatory to perceptual experiences by appealing to these sensible profiles. My attempt here is not to discuss Johnston’s account of sensible profiles or his particular epistemological characterisation of disjunctivism. Rather, I want to claim that, given that he accepts a distinction between causal processes and constitutive elements, his theory seems perfectly compatible with the metaphysical version of disjunctivism that I am suggesting. Indeed, Johnson explicitly acknowledges that “to be an act of seeing is to be a visual experience … that is not an act of hallucination” and is sceptical about the analysis of perception because “it does not set out what it is intrinsically and essentially to be an act of seeing” (2004, p. 173).

168


play a non-causal role in determining whether an experience is a perception or merely an hallucination. “The only sense in which we can account for the role for the object of perception as a constituent of the sensory episode is acting as a necessary condition on the occurrence of the perceptual event. Mere presence of a candidate object will not be sufficient for the perceiving of it, that is true, but its absence is sufficient for the nonoccurrence of such an event. The connection here is not one of degrees of influence but that of a constitutive or essential condition of a kind of event.” (Martin 2004, pp. 56-57.)

There is a non-causal difference between hallucinations and veridical perceptions that play a constitutive/ontological role at the moment of typing experiential kinds. What happens in the hallucinatory experience––the absence of the perceptual object––simply does not happen in the perceptual case. With regard to the first idea that could motivate the intentionalist to accept the Highest Common Factor, I want to claim that we can perfectly acknowledge that perceptual experiences are indeed part of the natural causal order, emphasising at the same time that there are non-causal aspects, which are constitutive of the kind of experiences we have. The intentionalist is not forced to accept the restrictive interpretation of the “same proximate cause, same effect” principle. It is not in virtue of the representational content considered independently of the mindindependent world that perceptual experiences have a functional explanatory role in our mental economy. On the contrary, it is only in virtue of those experiential contents considered as constituted by the objects themselves that they can play that role. As we will see in the following sections, this disparity present at the ontological/metaphysical level does not force us to deny the fact that genuine perceptions and their perfectly matching hallucinations can indeed appear to be the same from the subject’s introspective standpoint.

4 Phenomenal indiscriminability In addition to the restrictive interpretation of the “same proximate cause, same effect" principle that assumes that the experiential outcome of the activation of the same proximate causes must necessarily be of the same ontological kind, those who argue in favour of a Highest Common Factor view have traditionally appealed to introspective phenomenological considerations. According to the argument from hallucination, we cannot tell apart solely by means of introspection a veridical perception from its perfectly matching hallucinatory counterpart. Allegedly, sameness of experiential kind follows from the fact that we cannot discriminate on the basis of introspection between what it is like to have a veridical perception of O and merely seeming to have a perception of O. If we cannot discriminate between


169

them on the basis of introspection, it must be because the experiences have exactly the same phenomenal character. And if they have the same phenomenal character, then these experiences must be of the same fundamental kind. According to this line of interpretation phenomenological indiscriminability entails fundamental sameness of experiential kind and consequently the Highest Common Factor view of visual experiences. The argument––as used by Highest Common Factor theorists––takes the following general form: (i)

If two experiences are subjectively indiscriminable, then they have the same phenomenal character. (ii) If two experiences have the same phenomenal character, then they are experiences of the same fundamental kind. (iii) Therefore, subjectively indiscriminable experiences are experiences of the same fundamental kind.

I think it is very important to keep apart what are the relevant ontological and epistemological claims involved in this argument. First of all, we have the epistemic claim that we cannot tell apart solely on the basis of introspective reflection the phenomenology of a veridical perception from the phenomenology of a perfectly matching hallucination. The claim is indeed epistemic because discrimination is an epistemic notion connected with what we can and cannot tell on the basis of our introspective evidence. Secondly, we have an ontological claim regarding experiential kinds, specifically concerning what it is for a particular experience to be identical to or of the same kind as another experience. In this sense, I think the Highest Common Factor view of perception and its denial are fundamentally metaphysical or ontological proposals, and in this sense the conclusion of this section will also be an ontological one. First of all, I do not see why we need to accept premise (i), mainly because I see no conclusive reason to reject the plausible assumption that often, experiences with different phenomenal characters can seem the same from the subject’s standpoint. As Harold Langsam (1997) has pointed out, the traditional reading of the argument from hallucination assumes that introspective indistinguishability has to be explained exclusively by virtue of the instantiation of essentially the same phenomenal features. However, this interpretation is too strong. All we need to say is that perceptual experiences and hallucinations may seem to be the same. What matters for indistinguishability is sameness at the level of how these experiences seem from the subject’s standpoint. Perceptual states and hallucinatory states have something in common only to the extent that they seem to be the same because the subject cannot discriminate between them. However, epistemic considerations with regard to what we can or cannot discriminate from the introspective point of view are not necessarily tied to the basic metaphysics of the visual experiences in question.

170


“Perceptual and hallucinatory experiences… need not share the same ontological character; in particular, the phenomenal features of perceptual experiences can be the instantiations of relations between material objects and minds even though the phenomenal features of hallucinations are not. There is no reason to think that two phenomenal features cannot share the same appearance yet differ in ontological character. So the conclusion of the Argument from Hallucination that the phenomenal features of perceptual experience are not relations between material objects and minds is avoided.” (Langsam 1997, p. 39.)

If discrimination is an epistemic notion regarding what we can judge on the basis of introspective evidence I think that we should consider seriously Williamson’s advice when he says that “granted that knowing is a mental state, one should therefore not be surprised that one can fail to know something without being in a position to know that one fails to know it” (2000, p. 15). In perceptual terms this means that one can fail to perceive a particular object without being in a position to know that one is actually failing to perceive it. Indeed, if we assume that veridical perceptions and their perfectly matching hallucinations are phenomenologically indiscriminable, we should also admit a fallible view about the nature of our introspective judgments. Why? Well, simply because phenomenological indiscriminability puts us in a disjunctive situation regarding how things are with us. Any statement usually expressed using the language of appearances, such as “It seems to me that there is a red carpet”, can be alternatively reported by the disjunction “either I perceive a red carpet or I am hallucinating a red carpet”. The crucial aspect is of course that from the subject’s point of view we are not in a position to privilege one section of the disjunction over the other. This idea is central to what I have called Metaphysical Disjunctivism and also to any intentional account of perceptual experience that attempts to preserve a relational understanding of successful experiential states as involving actual physical objects in the physical world. Indeed, a modest or limited view of our discriminatory abilities and the sort of ontological considerations that we can draw from them has been present since the beginning of what we might call the “disjunctive turn” in philosophy of perception with Hinton (1973) and his early followers. The suggestion is that our judgments regarding how things look introspectively and what we can or cannot tell on the basis of these “looks” are not infallible. Indeed, there is empirical evidence that has advanced the thought that sometimes we are cognitively homeless with regard to the essential constitutive nature of our experiential states. Part of the evidence that leads us to argue in favour of a fallible view about our discriminatory abilities, and in general about what can we settle on the basis of introspection, comes from the so-called non-transitivity of indiscriminability for some observable properties, like shades of colours. The idea that indistinguishability in appearance is not transitive has been supported by


171

several experiments, like the one performed by Rohit Parikh that Hilary Putnam mentions in his Dewey Lectures: “He took a can of white paint and a pack of a hundred 3" by 5" cards and painted one card with some of the paint. Then he added a single drop of red paint to the can and stirred well. He painted the next card with the mixture. (It was absolutely indistinguishable from the first card, “as far as the eye could tell”). He continued in this way, adding one drop of red paint to the mixture after each card was painted, using the resulting mixture to paint the next card and so on. The result was a pack of one hundred cards such that if one looked at two successive cards one could not distinguish them at all (with respect to colour), but if you looked at two cards eighteen or nineteen apart in the pack you could see that the card that was later in the pack was slightly more pinkish than the pack that was earlier in the pack.” (Putnam 1999, p. 130)

The relevant philosophical point that we can extract from this experiment is that if we consider a pack that contains one hundred cards (let C1, C2, C3…, C100 be the cards), then according to the story C1 and C2 are indistinguishable for the subject on the basis of introspection. If this is the case, according to the Highest Common Factor view C1 and C2 not only have the same phenomenal character, but also the experiences that the subject undergoes when perceiving C1 and C2 are essentially of the same kind. We can call the phenomenal character essentially common to C1 and C2 simply “quale1/2” and similarly the phenomenal character common to C2 and C3 “quale 2/3”. According to the Highest Common Factor view of perceptual experience, if C1 and C2 are phenomenologically indiscriminable and C2 and C3 are also phenomenologically indiscriminable, then C1, C2 and C3 must have something fundamental in common, namely they must be experiences of the same kind. The question that drives Putnam’s attention is of course: Are quale1/2 and quale2/3 identical? If they are non-identical, then we arrive to the odd conclusion that one card, C2 in this case, has two subjective colours and we violate the idea that underlies the highest common factor view, that is, that C1, C2 and C3 have the same phenomenal character. On the other hand, if quale1/2 and quale2/3 are indeed identical then the indistinguishable phenomenal qualia that correspond to the rest of the cards, quale3/4, quale4/5, quale5/6, quale6/7… quale99/100 must be also identical, an idea that clearly is at odds with the experimental data according to which C1 and C19 are actually different colours and indeed the subject can perfectly discriminate between them. The philosophical lesson we must take from the non-transitivity of colour indiscriminability is that we are not forced to accept the premise according to which if two experiences are subjectively indiscriminable, then they must have the same phenomenal character, especially if we think––as I think we should––that phenomenal character has to be explained in terms of the external objects and properties that figure in our experience. Subjective indiscriminable experiences are

172


not necessarily experiences of the same essential ontological kind as the Highest Common Factor view entails.7

5 Fundamental ontological asymmetry In the previous section I argued that subjective indiscriminability is not sufficient for the identity of phenomenal character and therefore that it is not necessarily the right criterion for typing experiential kinds. The ideas discussed acknowledge that we can be wrong about the phenomenal character of the objects represented by our experience and that the Highest Common Factor interpretation of what actually follows from subjective indiscriminability is also too strong or restrictive. All I have pointed out up to now is that, given the fallibility of our introspective discriminatory abilities, we should not move so easily from epistemic considerations to ontological ones. However, there is something deeply problematic here. At the beginning of the paper I said that an intentional theory is necessarily committed not only to phenomenal indistinguishability but also to what I have been calling the “Commonality Thesis”: [Commonality Thesis] Genuine perceptual experiences and hallucinatory experiences have something in common, allegedly a mental state of the “same kind”.

On the one hand, the Commonality Thesis seems to be one of the main motivations for endorsing the ontology of the Highest Common Factor view. However, on the other hand, we have been describing Metaphysical Disjunctivism essentially as an ontological view according to which there is nothing fundamental obtaining across perceptual and hallucinatory instances of visual experience. This may suggest that Metaphysical Disjunctivism necessarily entails the rejection of the Commonality Thesis and a denial of the main features of the Intentional Theory. On what remains of this section I want to make clear that this is not the case and that Metaphysical Disjunctivism can perfectly acknowledge the Commonality Thesis granting that the kind of experience that a subject undergoes when veridically perceiving can indeed be considered as having something in common with the matching hallucinatory case. Following Pitcher (1971) I want to argue that we can indeed accept the Commonality Thesis, but not under every description. A neural state will always cause a “seeming to see something F” which is true under this particular description. Nonetheless, there are different descriptions like “hallucinating something F” or “seeing something F” which are certainly different and with different conditions of satisfaction. 7

In the disjunctive account of perception that I am suggesting I assume that the phenomenal/qualitative character of perceptual experiences is constituted and determined by the way physical objects and properties are independently of the perceiver.


173

“Consider the following example: light is reflected from Sarah Jones and ultimately produces a certain brain event, w, in a visual perceiver, Q. Imagine that w is described in purely physiological terms. It may well be true that w has the following effect in this case: Q sees Sarah Jones. Now imagine that Sarah Jones place is taken by her identical twin, Mary, who is dressed and made up just as Sarah was, down to the smallest detail. Light rays are reflected from Mary Jones and produce in Q a “qualitatively identical” brain event w as before. But this time, we cannot say that w has the effect that Q sees Sarah Jones, for he sees Mary Jones.” (Pitcher 1971, p. 57.)

In order to deny that hallucinations and perceptual experiences are of the same ontological/fundamental experiential kind it is reasonable to appeal to similar intuitions and argue that mental events can satisfy more than one description, constituting consequently different kinds. Martin (2004) for example, has distinguished between fundamental kinds and psychological kinds. Perceptual events are of a fundamental kind that cannot occur while hallucinating and, at the same time, they are of a psychological kind that is present during a causally matching hallucination. We have then, a psychological level of description that is common to both perceptions and hallucinations and a fundamental or constitutive level that makes the ontological difference. There is what I have called a constitutive or “fundamental” asymmetry between perceptual and hallucinatory states not only because the contents of genuine perceptual experiences are actually constituted by the worldly objects and properties which they represent, but also because––given that particular objects and properties are really constitutive of the kind of experiences we have––these contentful neurophysiological mental states are performing the task or goal for which they were fundamentally or essentially designed by nature. In order to clarify the last aspect we should remember that according to the interpretation that I suggested when introducing the Intentional Theory, visual mental representations are better defined in terms of biological proper functions. I think that if visual experiences are defined in this way, then they must be fundamentally evaluated and typed in terms of what these representations are supposed to or should do according to their evolutionary history, that is, ultimately to ensure and enhance our survival. As we know, the idea of visual experiences being biological functions does not rule out the possibility that these mechanisms may fail to perform according to design. That is why we have to distinguish between mental representations functioning properly (as they were biologically selected for) and not doing so. It is important to emphasise, though, that the possibility of misrepresentation in general and the occurrence of hallucinations and illusions in particular, does not imply that there is something wrong with the system which is actually producing contentful mental representations. Usually the failure

174


of a biological system to perform its functions properly is due to an “uncooperative environment” and not to internal failures. On the basis of these teleological considerations, I suggest we should think about perceptual experiences as essentially different from hallucinatory experiences, without forgetting that a biological system can fail to perform as designed, producing false or empty representations that could be indistinguishable from the subject’s introspective point of view. Even though hallucinations and perceptual experiences may play similar narrow psychological roles given that their contents are phenomenologically indiscriminable, the essential asymmetry of these experiences can be teleologically explained in terms of what we might call their fundamentally different broad world-involving results. Despite the fact that perceptual experiences and hallucinations cannot be introspectively discriminated by the subject of experience, it is clear that the results in terms of action and how these mental states actually lead us to succeed in the world and enhance our survival are radically different. If you disagree in this respect just think about the non-introspective elements that determine our perceptual and hallucinatory states and the different practical consequences that these visual experiences may have. If it seems to you that there is a “stop” traffic signal at a busy road intersection, or a “zebra” crossing, on the basis of which you confidently cross the street without looking both ways, you better be actually perceiving and not merely hallucinating. Accordingly, we should always keep in mind that apart from the constitutive role played by worldly objects and properties in perceptual experiences, hallucinations are also essentially different from perceptions because the world-involving results of these experiences are crucially different. Some theorists argue that it is only on the basis of epistemic indiscriminability that hallucinations are linked to veridical perceptions. M.G.F. Martin (2004), for example, insists that both mental states share the property of “being indiscriminable from” a veridical perception and claims that apart from this negative epistemic characterisation, hallucinations have nothing else in common with successful cases of perception. However, this thought is not necessary or even part of the sort of Metaphysical Disjunctivism I have in mind.8 Someone could in principle hold a view in which hallucinations are considered genuine experiences with content–– although a fundamentally different one––on the basis of the metaphysical/teleological considerations we have been discussing, or simply on the basis of reasons connected with alleged “particularity” exhibited by the content of 8 Matthew Conduct advised me to specify this point during the Mind2005 Conference at the University of Edinburgh. As he suggested, my view––unlike other disjunctive accounts–– involves the claim that hallucinations are indeed visual experiences with content, although a fundamentally different one. I want to thank him for reminding me of this point, which is certainly important to explain the psychological relevance of hallucinations and their connection with other intentional states of the mental network, such as beliefs and desires.


175

visual experiences. Mark Sainsbury, for example, has recently argued for a singularist position of perceptual content according to which “non-empty singular content is essentially non-empty, and empty singular content is essentially empty” suggesting therefore that “no singular content could be common to perception and hallucination” (2005, p. 253). Keeping this in mind, the general strategy does require the distinction between a fundamental and a psychological level of description, but it does not require an exclusively negative characterisation of hallucinatory states like the one held by disjunctivists who think that hallucinatory states are not considered contentful experiential episodes. I think that we can integrate in our teleological approach Martin’s (2004) distinction between what is fundamental or constitutively relevant for a visual state and what is merely a psychological or epistemic attribute of it, without endorsing or committing ourselves to the extreme position, according to which the only positive characteristic of hallucinations is the negative epistemological fact that they cannot be discriminated from perceptions by the subject. All we need to accept from Martin’s (2004) view are the following two claims, which indeed are fully compatible with Metaphysical Disjunctivism and with the teleological nature of mental representation. (1)

What we might call the fundamental mental property of hallucinations is the property of being indiscriminable from veridical perceptions.

(2)

However, unlike hallucinations, the fundamental mental property of a veridical perception of x is crucially its being a veridical perception of x and not its being indiscriminable from being such a veridical perception.

It is in this sense that hallucinatory experiences can be described only derivatively in connection with veridical perception. Perceptual error is, according to the interpretation that I am taking, parasitic in those cases of successful representation, but the relation that exists between perceptions and hallucinations is better described as a fundamental ontological asymmetry. In order to explain the fundamental nature of hallucinations, we need to appeal to veridical perceptions. However, in order to explain the fundamental nature of veridical perceptions we do not need a derivative tactic, in spite of the fact that both of them––perceptual and hallucinatory experiences––may share a psychological property. We can make this assumption only if we keep in mind that there are different levels of description–– an epistemic/psychological one on the one hand, and an essential/ontological one on the other. The denial of the Highest Common Factor view does not necessarily entail the denial of the Commonality Thesis as I stated it. In summary, hallucinations and perceptions are contentful experiential states that represent the world as being a certain way, even if the world actually is not the way it is being represented. In addition, hallucinatory experiences are


176

phenomenologically indiscriminable from perceptual episodes solely on the basis of introspection. However, given that perceptual experiences (illusory and veridical) are actually constituted by worldly objects and/or properties, they are essentially different from hallucinations. After considering the importance of noncausal elements, the non-transitivity of indiscriminability for some observable properties, and the metaphysical/teleological reasons for accepting a fundamental ontological asymmetry between successful instances of perception and hallucinatory states, we have enough grounds to argue for an alternative nonrestrictive interpretation of the argument from hallucination that allows us to preserve our pre-theoretical intuition according to which perceptions are essentially different from hallucinations, despite the fact that from the subject’s introspective point of view it is not possible to discriminate between them. Let us consider then for the last time the argument from hallucination, but now integrating the previous remarks:9 (i) (ii)

Subject S veridically perceives a physical object O. When subject S sees O, there is some type of proximate cause C in S’s brain that determines the occurrence of the veridical experience VP that S undergoes. (Veridical Scenario) (iii) It is possible for C to occur in S, even if there is no relevant object O in the environment being perceived. (Hallucinatory Scenario) (iv) When two scenarios involve the activation of the same proximate causes and do not differ in any non-causal way for the occurrence of some experience, then the possibility for the occurrence of such experiences are the same in both scenarios. (v) No non-causal conditions are present in the Hallucinatory Scenario, which are not also present in the Veridical Scenario. (vi) For whatever kind of experience that occurs in the Hallucinatory Scenario, the same kind of experience could occur in the Veridical Scenario.

Notice that (iv) is a different version of the “same proximate cause, same effect” principle that allows the possibility that some effects––in this case psychological states––can be individuated in the manner suggested by externalists like Tyler Burge (1991) or Hilary Putnam (1999). This weakened version of the “same proximate cause, same effect” principle acknowledges that the activation of the same proximate causes is not sufficient to bring about an experience of essentially the same kind. Notice also that (v) is compatible with the idea that the 9

I owe this interpretation of the argument from hallucination to similar versions proposed by Mike Martin, both in his writings and in his seminars on perception at the University of London. Despite the influence of Martin’s work in my paper it is crucial to remind the reader that Martin’s views on Disjunctivism and the Intentional Theory are clearly different from the ones I have been suggesting.


177

presence of the relevant object of perception is not sufficient for a particular experience to count as a veridical perception, and also consistent with the claim that the absence of the object is indeed sufficient for the non-occurrence of that veridical experience. It is the absence of the non-causal conditions that makes the difference. Finally, we can accept conclusion (vi) acknowledging that veridical perceptions and hallucinations are of the same psychological kind without accepting the ontological identity given the different ways in which these experiences are in fact externally individuated. More importantly, the Commonality Thesis interpreted in this epistemic/psychological way does not involve the acceptance of the Highest Common Factor and does not necessarily entail the denial of the naïve pre-theoretical view that holds that perceptual experiences are object-involving. Metaphysical Disjunctivism, that is, disjunctivism considered strictly as a metaphysical thesis regarding perceptual experiential kinds, does not involve the denial of the essential claims that characterise the Intentional Theory. Rather, I think that the Metaphysical/Ontological Disjunctive approach I am suggesting can be better characterised as a particular way of being an intentionalist, a way that makes sense of our pre-theoretical intuitions regarding visual experience.

Bibliography Brewer, B. (2004). “Realism and the Nature of Perceptual Experience”, in Philosophical Issues 14: 61-77. Burge, T. (1991). “Vision and Intentional Content”, in R. Van Gulick and E. LePore (eds.), John Searle and his critics. Oxford, Blackwell. Coates, P. (1998). “Perception and Metaphysical Scepticism”, in Supplementary Proceedings of the Aristotelian Society 72: 1-28. Crane, T. (2001). Elements of Mind: An Introduction to the Philosophy of Mind. Oxford: Oxford University Press. ——— (2006). “Is there a Perceptual Relation?”, in T.S. Gendler and J. Hawthorne (eds.), Perceptual Experience. Oxford: Oxford University Press. Hinton, J. M. (1973). Experiences: An Inquiry into some Ambiguities. Oxford: Clarendon. Huemer, M. (2001). Skepticism and the Veil of Perception. Oxford: Rowman & Littlefield. Johnston, M. (2004). “The Obscure Object of Hallucination”, in Philosophical Studies, 120: 113-182. Langsam, H. (1997). “The Theory of Appearing Defended”, in Philosophical Studies, 87: 33-59. Lewis, D. (1988). “Veridical Hallucination and Prosthetic Vision”, in J. Dancy (ed.), Perceptual Knowledge. Oxford: Oxford University Press.

178


Lowe, E. J. (2000). An Introduction to the Philosophy of Mind. Cambridge: Cambridge University Press. Martin, M. G. F. (2002). “The Transparency of Experience”, in Mind and Language, 17: 376-425. ——— (2004). “The Limits of Self-Awareness”, in Philosophical Studies, 120: 3789. McCulloch, G. (2003). The Life of the Mind: An Essay on Phenomenological Externalism. London: Routledge. McDowell, J. (1988). “Criteria, Defeasibility, and Knowledge”, in J. Dancy (ed.), Perceptual Knowledge. Oxford: Oxford University Press. Pitcher, G. (1971). A Theory of Perception. Princeton: Princeton University Press. Putnam, H. (1999). The Threefold Cord: Mind, Body, and World. New York: Columbia University Press. Robinson, H. (1994). Perception. London: Routledge. Sainsbury, R. M. (2005). Reference Without Referents. Oxford: Oxford University Press. Searle, J. (1983). Intentionality. Oxford: Oxford University Press. Snowdon, P. (1988). “Perception, Vision, and Causation”, in J. Dancy (ed.), Perceptual Knowledge. Oxford: Oxford University Press. Soteriou, M. (2000). “The Particularity of Visual Perception”, in European Journal of Philosophy 8: 173-189. Thau, M. (2004). “What is Disjunctivism?”, in Philosophical Studies 120: 193-253. Travis, C. (2004). “The Silence of the Senses”, in Mind 113:57–94. Williamson, T. (2000). Knowledge and its Limits. Oxford: Oxford University Press.

CHAPTER TEN CHANGE BLINDNESS AND COUNTERFACTUAL DEPENDENCE

NICK JONES

In this paper I am concerned with a familiar philosophical question—what does it mean to see something?—with a particular answer to that question offered by David Lewis, and with recent empirical work in perception that has implications for the sort of answer that Lewis proposes.1 Let’s begin with another familiar question: how should we distinguish seeing from hallucination? At first sight, we might think that the distinction is as follows. When we see something that object is actually present; in hallucination the apparent object is not present. We can refine this a little: in hallucination, no object closely resembling the object presented by the seeming experience is present and in clear view. This close resemblance condition means that minor misperceptions count as instances of seeing, rather than as instances of hallucination. This condition alone cannot deal with the possibility of veridical hallucination—the possibility that hallucination might present actual objects that are in clear view. Grice (1961) famously used this possibility to motivate his causal theory of perception. Our concept of perception must include causation, he claimed, since the difference between perception and veridical hallucination is that one’s perception of O is caused by the presence of O; the same is not true of an hallucination of O. It looks as though the close resemblance condition should be supplemented by a causation condition. By considering a number of what he called “extraordinary circumstances”, David Lewis (1980) showed that whether a causation condition is effective depends 1

I am grateful to the attendees at Mind2005 for their comments and particularly to my respondent, Christoph Kelp. Bill Fish and Stephen Barker kindly read and commented on an earlier draft of this paper.

180


on what we mean by causation. Lewis’s counterexamples are well known, but it will do no harm to rehearse them here. •

In his first example I am hallucinating at random, and have an hallucination of my brain, which is (of course) caused by my brain. Unknown to me, my brain has been removed under anaesthetic, the connecting fibres have been stretched, and my brain is actually lying in front of me, in just the place my hallucination suggests.

•

In the second example, a magician stands in front of me and casts a spell that causes me to hallucinate at random. My hallucination happens to match the scene before my eyes, of the magician casting a spell.

•

The third illustration is more problematic, because the causal mechanism seems to be closer to that of true perception. In this example I am blind, but my brain is wired to a light meter, so that a certain light intensity will cause me to hallucinate a particular landscape. I happen to be facing just that landscape, and the light reflected from it is of sufficient intensity to trigger a matching hallucination.

In each of these cases an object in clear view causes me to have a seeming experience of just that object, yet our intuition is that this seeming experience is an hallucination, not an instance of seeing. Why should this be? Perhaps it is because we feel that these cases do not involve the normal process of perceptual causation: surely normal perception involves light reflected from the object striking our eyes, stimulating the retina, and this in turn producing a signal that travels along the optic nerve to the appropriate areas of the brain, which are responsible for our perceptual experience. In other words, these problem cases are not instances of seeing because they do not involve the standard causal process. Anticipating this response, Lewis asks whether this standard process is essential to seeing: what would be the case, for example, if for a minority of people visual experience were generated by some quite different means? What should we say if someone had a prosthetic eye fitted, which worked rather like a TV camera? Does this not count as seeing? It seems that it should.2 Of course, in Lewis’s examples the hallucination matches the scene only by chance. Even in the light meter case, where the light reflected from the scene is responsible for my visual experience, any number of scenes might have produced the appropriate level of light reflection. This suggests a more robust way of distinguishing perception from veridical hallucination. To truly see a scene, according to Lewis, it is necessary not just that the scene causes us to have a closely resembling visual experience, but that had the scene been different, the content of our experience would then have closely resembled that alternative scene. 2

I share Lewis’s intuition, but don’t intend to defend it here. I am concerned with a quite different objection to his claim about what is essential to seeing.


181

In other words, in order to see it is not sufficient merely that the scene before us causes experience with matching content; there must also be a similar dependence across a range of counterfactual cases. According to Lewis, these are not separate requirements: the appropriate pattern of counterfactual dependence amounts to causal dependence (1980, p. 247; see also Lewis, 1986). So perception is indeed distinguished from hallucination by being caused by the apparent object, provided causation is understood as counterfactual dependence. The counterfactual dependence condition serves to rule out Lewis’s counterexamples as instances of seeing. In these cases my hallucinations are entirely random. Had the scene been different, then it is highly unlikely that my experience would have tracked this difference. (Unlikely but not impossible: my random hallucination might also be different and so might, by chance, match the alternative scene, but as the number of possible counterfactuals grows, the odds against this continuing coincidence become very high indeed). The counterfactual dependence condition is offered as a way of distinguishing veridical hallucination from seeing. However, it is more than this. Lewis is clear that all instances of true visual perception should show this pattern of dependence: he claims that counterfactual dependence is a distinguishing mark of seeing (1980, p. 245). Yet there are some problem cases that suggest we may need to revise this counterfactual dependence condition. These are cases of what has come to be called change blindness, or inattentional blindness. In brief, change blindness describes our failure to recognise significant changes to a scene if these are very gradual, or if the moment of change coincides with some discontinuity in perception—for example, if it is masked in some way. In many of the computer-based demonstrations of the phenomenon the moment of change coincides with a flicker in the picture displayed; it may equally well be masked by blinking or by a rapid saccade (eye movement), or in the case of demonstrations using movies, a change in point of view.3 The changes that take place can be quite dramatic, and if we do become aware of them (which may take some minutes) it seems inconceivable that we could ever have failed to do so. Typical pictures for static demonstrations include examples in which a lakeside scene loses the reflection of buildings in the water, a rocky outcrop behind the main subject disappears, and a church jumps from one part of the scene to another. Other experimental materials include brief movies in which the plates in a restaurant scene change colour; an actor’s colourful scarf appears and disappears; one actor substitutes for another. In the study from which these last examples are taken, none of the participants noticed the change (Levin and Simons 3

See e.g. McConkie and Currie (1996) and Rensink et al. (1997). There is also earlier work of relevance by Neisser and others: Simons and Chabris (1999) includes an excellent brief review of research. The implications for philosophical views of consciousness were anticipated in Dennett (1991) and are explored in Noë (2002).

182


1997). The effect seems to apply to real life situations as well. In one study, the experimenter stopped a passer-by to ask for directions. As they spoke, workers carrying a door passed in between them. While the experimenter was concealed, he was replaced by an accomplice, but only 33% of the subjects noticed the switch (Simons and Levin 1998).4 In the closely related phenomenon of inattentional blindness, participants carrying out an attention-demanding task fail to notice major unexpected events. In one of the best known experiments, participants were shown a videotape of two teams of players passing a basketball, and were asked to keep count of the number of passes carried out by one of the teams. About half way through the tape, either a woman carrying an umbrella or an actor in a gorilla suit was clearly visible walking through the midst of the action; this sequence lasted for about 5sec of the 75sec recording. At the end of the experiment, participants were asked if they had noticed anything unusual. Even without any disturbance in the continuity of the scene, just half (54%) of these who had carried out their task with reasonable accuracy noticed the unexpected event (Simons and Chabris 1999). The experiment demonstrates, in a particularly dramatic way, the extent to which we can fail to notice major changes in a scene if we are not attending to them. There is a growing body of research showing similar effects.5 Much of the philosophical attention that this work has attracted has been concerned with the implications for our view of perceptual experience. These results have been thought to support the claim that in perception we never form a detailed representation of our surroundings. To quote the title of a collection of papers edited by Alva Noë (2002), the apparent detail and richness of our visual experience may be no more than “a grand illusion”. This is clearly an important claim, but it is not what I am primarily concerned with here. Neither am I too worried about whether change blindness and inattentional blindness are distinct phenomena, or what the relationship between them might be. I am more interested in the simple challenge that these experiments, taken as a whole, present to the idea that seeing necessarily involves counterfactual dependence. This is what Lewis says about seeing:

4

To allow for ambiguous responses, the 33% figure was later adjusted by the experimenters to 46% (Levin, Momen, Drivdahl and Simons 2000). The still pictures referred to in this paragraph, together with other demonstrations, can be found on Kevin O’Regan’s website at the University of Paris (http://nivea.psycho.univ-paris5.fr/).and Ronald Rensink’s at the University of British Columbia: (www.psych.ubc.ca/%7Erensink/flicker/download/index.html). 5 Simons and Chabris carried out several variants of the test: the 64% figure is an average across all variants. For one influential strand of research in inattentional blindness, see Mack and Rock (1998).


183

“… if the scene before the eyes causes matching visual experience as part of a suitable pattern of counterfactual dependence, then the subject sees; if the scene before the eyes causes matching visual experience without a suitable pattern of counterfactual dependence, then the subject does not see.” (1980, p. 245)

We can take this to mean that if our experience is to count as seeing, it must be capable of matching the present scene, and of tracking actual and counterfactual changes to the scene. If the scene had been different the content of our experience would have been different. According to Lewis, the content of our experience “… is, roughly, the content of the belief it tends to produce” (1980, p. 239), and such belief is largely self-ascriptive: one ascribes to oneself the property of being in a certain perceptual situation. To say that a subject’s experience correctly matches the scene, then, is to say that her beliefs about her perceptual situation correctly match her perceptual situation. This is just what we don’t find in the change blindness experiments; the subjects show no evidence of any change of belief. I should say exactly how change blindness challenges Lewis’s claim. The challenge does not arise from cases where our experience fails to track an actual change in the scene. There are quite proper questions about whether we can claim to be seeing when the content of our experience differs dramatically from the actual scene before us, and Lewis’s requirement that the actual scene must cause matching visual experience seems unobjectionable (it is a rather stricter variant on what I earlier called the close resemblance condition). But Lewis’s claim is not just about changes to the actual scene, but about possible counterfactuals. According to the dependence rule, we are truly seeing only if our experience matches the scene and had the scene been different, our experience would have been appropriately different. So, consider this case. We have a visual experience—say, of some people passing a basketball—that presents the actual scene before us: in Lewis’s terms we have matching visual experience. In the counterfactual case, if the scene had been significantly different—say, if it included someone dressed as a gorilla—then we would expect our visual experience to be appropriately different. But in the light of the change blindness cases we have reason to believe that it might not be. In a counterfactual case involving change blindness, our visual experience would fail to track the changes in the scene. As a result, since counterfactual dependence fails, we cannot claim to be seeing the scene. At one point Lewis says that he “claim[s] only sufficiency and not necessity” for these two conditions; that scene-matching and counterfactual dependence are sufficient but not necessary conditions for seeing (1980, p. 241).6 If so, then cases 6

It might seem strange to claim that cases that do not involve scene-matching, at least, could ever be instances of seeing. Lewis has in mind cases where the failure of match is systematic (e.g. spectrum inversion), and could be interpreted appropriately by the subject.

184


that satisfy both scene-matching and counterfactual dependence will count as instances of seeing, yet there could also be instances of seeing that do not satisfy these conditions. That might allow the example I have just given to count as an instance of seeing. Yet the claim that these are merely sufficient conditions seems odd, in the light of the section I quoted earlier. In that section Lewis says: “… if the scene before the eyes causes matching visual experience without a suitable pattern of counterfactual dependence, then the subject does not see.” (1980, p. 245)

Expressed in that way, it sounds as though counterfactual dependence is also a necessary condition for seeing. Indeed, it is hard to see how counterfactual dependence could do the job of distinguishing true perception from veridical hallucination if it were not a necessary condition. If counterfactual dependence is the property that all possible instances of seeing possess, and which no instance of veridical hallucination possesses, it must be a necessary property of seeing. If this is so then cases like the one I have just described, in which change blindness is among the counterfactuals, could not be instances of seeing since they fail to demonstrate counterfactual dependence. But it is not just individual cases that are put in doubt: if seeing cannot take place without both scene-matching and counterfactual dependence, then the possibility of change blindness means that we could never see anything. This conclusion may have already occurred to you, but I should spell out how I arrive at it. Consider this: whatever scene currently lies before us, the evidence for change blindness gives us reason to think that we would fail to notice many different changes in that scene. There are therefore many counterfactual cases in which such a change has taken place, and in which our experience has failed to track those changes. As a result, whatever scene currently lies before us, there will never be a good enough pattern of counterfactual dependence to satisfy the dependence condition. If counterfactual dependence is necessary to seeing, then there are no instances in which we can claim to see. I think we can assume, despite what he says, that Lewis intends counterfactual dependence to be a necessary condition for seeing. I suggest that if it is a necessary condition, there will be no cases in which we can claim to see. Should we therefore treat it as merely a sufficient condition, and look elsewhere for necessary conditions that might enable us to discriminate seeing from hallucination? Let’s consider what follows if we take counterfactual dependence to be only a sufficient condition for seeing. Instances of counterfactual dependence would count as instances of seeing, but the way would be open for other experiences, that lacked this dependence, to also count as instances of seeing—cases such as veridical hallucination. And change blindness still presents a problem. For sufficient conditions to be of any value, there must be at least some cases that might fall under them. Yet as I have just said, it seems that there are no cases where change


185

blindness is not among the possible counterfactuals. If so, then there are no cases that could satisfy the counterfactual dependence condition. I have deliberately stated the problem as forcefully as I can. Perhaps I have misapplied the counterfactual dependence condition: here are four possible objections to the case I have just put forward. First, it might be objected that the changes involved in change blindness are too insignificant to count against scene-matching. After all, Lewis recognises that for his proposal to be plausible the close resemblance rule must still apply. As he puts it, “one can see even if the match, actual and counterfactual, is close but imperfect.” (1980, p. 245, my emphasis). Perhaps we can argue that in change blindness the match between scene and experience should count as being close but imperfect; close enough at least for us to claim counterfactual dependence. This requires a judgment about how close a match we need, and admittedly judgments may vary. In my view, however, the match between scene and experience in many of these cases is not “close but imperfect”, it fails massively: a scene that includes a gorilla is not a close but imperfect match to a visual impression of the scene that omits the gorilla. A friend of Lewis might respond that while the gorilla is an important element of the scene, what matters to the observer are changes to those details of the scene that they are attending to, and that it is the match between those details and the experience of those details that matters. I have some sympathy with this objection, for reasons that will become clear, but it is not an adequate defence of Lewis. He requires a close match between target and visual experience at the level of the whole scene. Second, although Lewis requires that there should be matching across a range of possible counterfactuals, he recognises that we can see even if there are some alternative scenes that fail to produce matching visual experience (see his “Laser Beam” and “Hypnotic Suggestion” cases (1980, p. 245)). If change blindness affects some counterfactual cases, but the dependence holds in every other case, is this not good enough for us to claim to be seeing? The problem is that there are not just a few possible change blindness cases. Remember my earlier point: for any given scene, there are a large number of changes that, given the evidence for change blindness, we would fail to notice. Exactly how many, and whether they outnumber the changes that we would notice, is an empirical question that could prove very difficult to resolve, but at the very least there will be no shortage of change blindness counterfactuals. Lewis’s example of The Censor presents a rough parallel to the change blindness case, which is of some relevance to the present discussion. This is what he says: “… the scene before my eyes causes matching visual experience. But if the scene were any different my visual experience would be just the same. For there is a censor standing by, ready to see to it that I have precisely that visual experience and no

186

Content, Consciousness and Perception other, whatever the scene may be … So long as the scene is such as to cause the right experience, the censor does nothing. But if the scene were any different, the censor would intervene and cause the same experience by other means.” (1980, p. 248)

Lewis agrees that this is not a case of seeing, since the censor’s victim has no capacity to discriminate by sight; however the scene differed, their experience would never change. Clearly, the parallel between the censor and the change blindness cases is not exact. In the censor case my experience always remains unchanged. In the change blindness case certain changes might translate into matching visual experience: my experience will not fail to track every counterfactual. So how many counterfactuals must we be capable of tracking in order to count ourselves as seeing, according to the dependence condition? I’m not sure how we might answer that question, or if it even has a precise answer; as Lewis recognised, the boundaries between seeing and not seeing are fuzzy (1980, p. 246). We can reasonably assume that the higher the proportion of counterfactual cases where our experience fails to match the scene, the more uncertain it becomes whether we can claim to see. I claim, on the basis of the experimental results cited earlier, that the proportion of such cases will be high enough to cause considerable uncertainty. The third objection I anticipate is this. If the change is not masked, or if it occurs at sufficient speed at the focus of attention, it will often be immediately obvious. If the two pictures used in a change blindness test are placed side by side, then in many cases the difference is easy to spot. Perhaps we should only consider cases in which we are attending to the scene, our perception is uninterrupted, and the change is rapid enough to attract attention. Yet this would exclude many, if not most, cases of seeing. There is good reason to think that our perception is never uninterrupted; at the very least, our eyes saccade several times a second, and a saccade is quite enough to mask the sort of changes we are considering. This gives rise to the fourth objection. If the changes that arise in change blindness cases are not normally visible, we cannot be expected to discriminate them. We might argue that since change blindness is just a feature of human perception, like our inability to see in infrared, changes due to change blindness are effectively indiscriminable, and shouldn’t count against the close resemblance condition. If my experience cannot register a change, then failure to do so shouldn’t count as failing to provide a close but imperfect match to the scene before me. Lewis addresses the question of perceptual discrimination, in particular the question of changes too fine-grained to be visible, and concludes that: “What can be required analytically is that there be plenty of visible differences of one sort or another; that is, plenty of different alternative scenes that would produce different visual experience and thus be visually discriminable.” (1980, pp. 245/6)


187

So, the objection runs, the change blindness cases fall foul of the requirement to provide “visible differences”, and we can exclude them from the range of counterfactuals that we need to consider. Yet the changes involved in these cases are not generally invisible: they are perfectly visible in certain circumstances, for example when pictures displaying the change are laid side by side, or once the difference has been pointed out. Certainly, there are some circumstances in which they are hard to discriminate—this is why there is such a phenomenon as change blindness—but still, they will be seen by some subjects at some times, and by others at other times. It is not that they are of a sort to be generally invisible, like the fine-grained changes considered by Lewis; rather that there are occasions— admittedly very many occasions—when they will fail to be noticed. Again, I am sympathetic to this objection, but it seems to strike at the heart of the scene-matching condition for both actual and counterfactual cases. If you notice a change in the scene and I don’t, then we want to say that you see it and I do not. Yet if we have excluded this sort of change from the range of changes for which we require matching, how then do we make this distinction? I don’t believe that any of these objections succeeds in defeating the challenge posed by change blindness. If scene-matching and counterfactual dependence are necessary conditions for seeing, then the evidence for change blindness suggests that there are no circumstances in which we can claim to see. Assuming that we want to preserve a causal account of seeing, and think that something like Lewis’s counterfactual analysis is the right way to understand causation, can we amend the conditions so as to overcome the change blindness problem? I’m not certain of the right answer, but here, at least, are two possible approaches which I think point in the right direction. 1Relax the close resemblance condition I’ve already argued that the changes involved in change blindness can be too significant to allow us to claim that our experience presents a close but imperfect match for the scene before us. We might take a different approach. Rather than argue that these changes are de minimis, we might relax the close resemblance condition, which already allows for some slight mismatch between scene and experience, to allow for a more significant mismatch in both actual and counterfactual cases. The problem here, it seems to me, is that if we allow the match between scene and experience to be loose enough for cases of change blindness to count as matching cases, then we are adopting a very lax definition of what it means to see. In order to allow counterfactual cases of change blindness to satisfy the dependence condition, we must allow actual cases of change blindness to satisfy the scenematching condition, and this cannot be right. If I do not see a gorilla, then I do not

188


see a scene that contains that gorilla. In addition, weakening the close resemblance condition in this way allows in other cases that we would wish to exclude. If we allow that there is an acceptable match between scene and experience in the change blindness cases, what other cases (of hallucination or major visual disturbance for example) might we also have to allow to count as instances of seeing? 2 Treat seeing as a matter of seeing details, rather than seeing scenes This approach will probably already have occurred to you. Most, if not all of the problems that I have outlined arise because of the condition that there should be a match between entire scene and experience across both actual and counterfactual cases. Perhaps it is not the close resemblance condition that we should relax, but the requirement that this condition applies across the entire scene. The change blindness cases are typically cases in which our experience correctly matches much of the scene before us, and differs only in detail, albeit sometimes quite significant detail. Perhaps we should say that the unit of seeing is not the scene as a whole, but each detail of that scene. Rather than say that we fail to see the scene, we would say that we succeeded in seeing certain details of the scene and failed to see certain others. How exactly would this modify the counterfactual dependence condition? Well, it seems that we should say that the appropriate dependence (actual and counterfactual), is a dependence of the individual details of our experience on individual details of the scene. Where there is such dependence, we can claim to see just those details; where there is not, we will not see those details, although we may well see many of the other details of the scene. If the dependence holds for enough details of a particular scene, then we may be able to speak, loosely, of seeing that scene, but the unit of seeing is each detail, not the scene as a whole. The level of detail at which we require a match will be important. How might this apply in a particular case? Say the scene before me contains five pine trees. In a range of counterfactual change blindness cases it contains only three, but of course I fail to register any change. So counterfactual dependence fails at that level of detail: can I claim to see the five trees in the actual scene? It seems not. But I can claim to have seen some trees—an indeterminate number of trees—since the informational content of my experience in both the actual and the counterfactual case consist of just this level of detail. This seems to give us the right result. In the actual case there may have been five trees in the scene, but that was not what my experience presented. It presented only an indeterminate number of trees, and this is all that I can claim to have seen. This has necessarily been a very brief sketch of a possible response, but I tentatively favour this second proposal. It enables us to respect the insight expressed in the first objection above; that what matters is not, as Lewis claims, our


189

beliefs about the scene as a whole, but those about the details of the scene to which we are attending. It is also in tune with the conclusion many draw from the change blindness evidence—that our awareness of our surroundings is far less detailed than we often believe.

Bibliography Dennett, D. C. (1991). Consciousness Explained. London: Penguin. Grice, P. (1961). “The Causal Theory of Perception”, in Proceedings of the Aristotelian Society Supp. Vol. XXXV: 121-68. Levin, D.T., Momen, N., Drivdahl, S. B. and Simons, D. J. (2000). “Change Blindness Blindness: the Metacognitive Error of Over-Estimating ChangeDetection Ability”, in Visual Cognition 7 (1/2/3): 397–412. Levin, D.T. and Simons, D.J. (1997). “Failure to Detect Changes to Attended Objects in Motion Pictures”, in Psychonomic Bulletin and Review 4: 501–506. Lewis, D. K. (1980). “Veridical Hallucination and Prosthetic Vision”, in Australasian Journal of Philosophy 58(3): 239-49. ——— (1986). “Causation”, in his Philosophical Papers, Vol. II. Oxford: Oxford University Press. Mack, A. and Rock, I. (1998). Inattentional Blindness. Cambridge, Mass.: MIT Press. McConkie, G. W., and Currie, C. B. (1996). “Visual Stability across Saccades while Viewing Complex Pictures”, in Journal of Experimental Psychology: Human Perception & Performance 22: 563-581. Noë, A. (ed.) (2002). Is the Visual World a Grand Illusion? Thorverton: Imprint Academic. O’Regan, K. Online at http://nivea.psycho.univ-paris5.fr/ (accessed 2/3/06). Rensink, R.A. Online at: www.psych.ubc.ca/%7Erensink/flicker/download/index.html (accessed 2/3/06). Rensink, R. A., O’Regan, J. K., and Clark, J. J. (1997). “To See or Not to See: the need for Attention to Perceive Changes in Scenes”, in Psychological Science 8: 368-373. Simons, D. J. and Chabris, C. F. (1999). “Gorillas in our Midst”, in Perception 28: 1059-74. Simons, D. J. and Levin D. T. (1998). “Failure to Detect Changes to People in a Real-World Interaction”, in Psychonomic Bulletin and Review, 5: 644-649.

CHAPTER ELEVEN AN ENACTIVE THEORY OF PHENOMENAL INTENTIONALITY

JULIAN KIVERSTEIN

Naturalistic philosophers of mind have typically assumed that the problem of finding a place in the natural order for intentionality is a soluble one.1 They have taken the hardest questions for a naturalistic account of the mind to concern phenomenal consciousness—the kind of consciousness in virtue of which there is something it is like for a creature to be in a mental state M. Indeed a recent trend has seen many naturalistic philosophers attempt to use their favoured account of intentionality to explain the nature of phenomenal consciousness. Some have argued that if a creature is in a phenomenal conscious state M then M must have an intentional content of a certain kind.2 Others have argued that if a subject S is in a phenomenal conscious state M then the fact that S is in M must be represented by means of a higher-order mental state of some kind.3 I will call those theorists who attempt to provide an exhaustive analysis or explanation of phenomenal properties by appeal to a conscious mental state’s intentional content, “reductive representationalists”. Reductive representationalists assume that the problem of accounting for intentionality and the problem that 1

I would like to offer many thanks to my commentator at Mind2005 David Levy for his helpful and thought-provoking response to my paper. I would also like to thank the conference organisers; Profs Tim Williamson, Christopher Peacocke and Hanjo Glock; Matthew Nudds and the other members of the audience for their probing questions which I hope have enabled me to strengthen my arguments. 2 See for instance Byrne (2001); Dretske (1995); Harman (1990/1997); and Tye (1995). 3 Lycan (1996) models the relevant kind of higher-order representation on perception while others take the higher-order representation to be either an actual thought (for instance Rosenthal (1986)) or a disposition to token a thought (Carruthers (2000); Dennett (1991)).


191

phenomenal consciousness presents for naturalism are separate problems. They believe that an adequate account of intentionality can be had independently of any account of phenomenal consciousness. One of my aims in this paper will be to challenge this assumption. I will argue that the type of intentionality that attaches to conscious mental states is essentially phenomenal. Consider the question of what the conditions are that fix a perceptual state S’s intentional content. The answer I shall defend says that a perceptual state S has the intentional content it does just in virtue of its phenomenal properties—those properties which determine what it is like for a creature to be in S. I will call intentionality which is essentially phenomenal in this way, “phenomenal intentionality”.4 In section 1 I introduce the idea of phenomenal intentionality in more detail. Section 2 argues against the conception of intentionality employed by reductive representationalists. In particular I will challenge the assumption that consciousness and intentionality are only accidentally related by drawing a distinction between intentionality conceived of as directedness and intentionality conceived of on the model of the reference relation. Directedness is an object-independent or nonrelational property while reference is an object-dependent or relational property. I will argue that intentionality is best conceived of as an object-independent or nonrelational property. This is something that is readily accommodated if we understand intentionality as phenomenal intentionality. Once we conceive of intentionality as an object-independent property one might worry that we must fail to account for the role that experience plays in grounding demonstrative judgment. Section 3 examines this worry and considers a couple of responses.5 Section 4 takes up the second objective of this paper which is to use the enactive theory of perception to sketch an account of phenomenal intentionality. Enactive accounts of sensory experience take an experience’s intentional content to derive from a certain kind of practical knowledge. An animal that is capable of perceiving an object, the enactive theory claims, will know how the appearance an object presents will change with movement. It is knowledge of this kind that fixes an experience’s intentional content. I shall argue that enactive theories of perceptual content are committed to the claim that an experience has its intentional content just in virtue of its phenomenal properties. The enactive theory of perception, I shall claim, shows us how it can be true that an experience has phenomenal intentionality. 4

Charles Siewert has argued for a similar position. See in particular chapter’s 7 & 8 of his (1998). Horgan and Tienson (2002); Loar (2003) and Zahavi (2003) also argue for the existence of something they call “phenomenal intentionality” though there are important differences in each of their accounts. 5 This difficulty is one I am working on in detail in a paper I plan to submit to a special issue of The Monist planned for 2007 on the topic of Phenomenal Intentionality.

192


1 What is phenomenal intentionality? Typically, discussions of the relation between intentionality and consciousness have focused on whether conscious experience can correctly be said to satisfy what Peacocke (1983) called “the adequacy thesis”. Peacocke formulates the adequacy thesis as follows: “The Adequacy Thesis states that a complete intrinsic characterisation of an experience can be given by embedding within an operator like “it visually appears to the subject that ….” some complex condition concerning physical objects. One component of the condition might be that there is black telephone in front of oneself and a bookshelf a certain distance and direction to one’s left, above and beyond which is a window.” (1983/1997, p. 343)

The question raised by the adequacy thesis is whether a complete and adequate description of perceptual experience can be given just by reference to what an experience represents. The adequacy thesis will be false if there is something about visual experience (and perceptual experience more generally) that a description of what an experience represents leaves out. Peacocke talks of ‘intrinsic features of experience’ as something the adequacy thesis may fail to account for. I shall talk instead of “phenomenal features”. The question the adequacy thesis raises is whether an experience’s phenomenal features can be fully and adequately described by reference to an experience’s representational content. What do I mean by “phenomenal features”? Sensory experiences present the world to a subject as seeming or appearing to be a certain way. Minimally, they present the world as containing things of a certain taste, sound, colour, shape and size occupying various spatial locations around the subject’s body. I understand phenomenal features to be those features of an experience in virtue of which the world appears to be these ways for a subject. Bodily sensations—aches and pains, twinges and tickles—also have phenomenal features; these states feel like something to their subjects. Bodily sensations have their characteristic feel in virtue of what I am calling a sensation’s phenomenal features. The question the adequacy thesis raises is whether phenomenal features can be given a complete and adequate description by making reference to what an experience represents. What a perceptual experience represents can be fully specified by making reference to the things and their properties in the world external to the subject. In the case of bodily sensations, the things and their properties in question might be parts or regions of the subject’s body.6 Can we fully account for phenomenal features by reference to the objects and their properties a sensory experience or sensation represents? I suspect that the answer to this 6

See Tye (1995) for a view along these lines.


193

question is ‘no’, though I will not attempt to defend this line here. My suspicion is that an adequate account of sensory experience must make appeal to what Crane (2003) calls “intentional modes” and Chalmers (2004) calls “manners of representation”.7 However rather than pursue this issue I want instead to focus on a question which is the converse of the one raised by the adequacy thesis. I shall consider whether a complete account can be given of an experience’s representational content just by making reference to an experience’s phenomenal features. I will answer this question in the affirmative by arguing for the existence of a kind of intentionality I will call “phenomenal intentionality”, a kind of intentional content that derives from a state’s phenomenal features. I will be defending the thesis that our conscious sensory experiences have their intentional features just in virtue of their phenomenal features. This thesis I will label “the Phenomenal Intentionality Thesis”. What are intentional features? Initially we can understand intentional features as those features in virtue of which a representational state is assessable for truth or accuracy, though we will see later in section 3 that this definition requires slight modification. Consider by way of illustration an experience as of a white rabbit. My visual experience will be accurate if it presents me with what appears to be a white rabbit, and there really is a white rabbit where my experience represents one as being. What I am assessable for in this case is analogous to what I would be assessable for if we were considering the truth or falsity of my assertion that there is a white rabbit before me. In the case of sensory experience what is being assessed for accuracy is the way my local environment appears to me at a particular time. The phenomenal intentionality thesis claims that a conscious sensory experience is assessable for accuracy solely in virtue of its phenomenal features. This is to say that there are no additional conditions that have to be met above and beyond an experience having certain phenomenal features in order for that experience to be assessable for accuracy. Reductive representationalists reject the phenomenal intentionality thesis. They argue that an experience has its phenomenal features in virtue of its intentional features, not the other way round as proponents of the phenomenal intentionality would claim. If reductive representationalists are to argue against the phenomenal intentionality thesis they must do so by identifying some extra condition in addition to the possession of phenomenal features by virtue of which a representational state is assessable for accuracy. In the next section I will look at the conditions that reductive representationalism identifies. I will argue that the reductive 7

Both these notions bear an interesting relation to what Husserl (1900) referred to as the “quality” of an act of consciousness, contrasting this with the “matter” of an act of consciousness or its intentional content.

194


representationalist fails to identify conditions that are necessary for a state to be assessable for accuracy. Thus they fail to establish that some extra condition is required for a state to be assessable for accuracy in addition to the state’s possession of phenomenal features.

2 Against reductive representationalism Reductive representationalists offer an explanation of an experience’s phenomenal features in terms of that experience’s intentional features. In order to avoid the charge of explanatory circularity a reductive representationalist must oppose the phenomenal intentionality thesis. She must argue that a sensory experience is assessable for accuracy in virtue of some conditions which do not make reference to an experience’s phenomenal features. Reductive representationalists disagree about the precise nature of this condition. Michael Tye (1995, 2000) works with a conception of intentionality as causal covariation under ideal conditions: “The causal connections that matter to phenomenal content, I suggest, are those that would obtain, were optimal or normal conditions operative….Experiences represent various features by causally correlating with, or tracking, those features under optimal conditions.” (Tye, 2000, p. 64)

Dretske (1995) makes an appeal to evolutionary function central to his account of intentionality.8 According to Dretske: “…a system, S, represents a property, F, iff S has the function of indicating (providing information about) the F of a certain domain of objects.” (p.6) Dretske goes on to add that the function of a system is ‘what it was designed to do’ and our senses have the function of providing information, a function which derives from our sensory system’s evolutionary history. Both these proposals give priority to cases of successful representation in specifying the extra condition required for the possession of intentional features. In both cases what I shall call a sensory experience’s “accuracy conditions”—the conditions under which it represents accurately—are fixed by appeal to the causal conditions the sensory experience would be causally connected to if it was carrying information—that is to say if it was representing correctly. Thus both Dretske and Tye offer explanations of intentionality which make the possession of intentional features contingent upon a causal connection to certain environmental conditions. I will call this the “causal connection condition”. The causal connection condition comprises two claims. 8

Tye (2000, ch.6) argues that evolutionary history is necessary for specifying content in some but not all cases.


195

(1) For any experience of type E there is some environmental condition C which experiences of type E are normally caused by.9 (2) It is the causal connection that holds between C and E that fixes the content of type E experiences. When this causal connection holds E can be said to carry information about C. What will proponents of the causal connection condition say about situations in which this condition fails to hold? Consider first Dretske’s version of the causal connection condition according to which an experience has its representational content fixed by ‘the biological functions of the sensory systems of which it is a state’ (Dretske, 1995: 15). What will Dretske say about a situation in which an experience is tokened by a sensory system which has no biological function as would be the case for Swampman, a molecule for molecule duplicate of me who has spontaneously sprung into existence? Dretske is committed to the view that when Swampman springs into existence his “mental states” will be entirely lacking in intentional features. For according to Dretske our sensory systems have evolved to carry information about our local environments—this is what they are supposed to do. Swampman’s sensory systems have not so evolved, so they will not have the function of carrying information. Dretske must say then that Swampman’s “experiences” will not represent anything. Dretske also holds the view that phenomenal features are identical with intentional features.10 Since Swampman’s “experiences” lack intentional features they will also lack phenomenal features—there will be nothing it is like for Swampman to sense his environment.11 An alternative, and it seems to me more plausible conclusion, one might draw about the case of Swampman is that it is not necessary for a state of a sensory system to have served a biological function in the past in order for this state to exhibit an intentional content. This is the line proponents of the phenomenal intentionality thesis will take.12 They will argue that Swampman enjoys experiences with phenomenal features—there is something it is like for Swampman to see colours, taste foods, smell roses etc. They will say that Swampman’s experiences are assessable for accuracy just in virtue of their possession of these 9

I am assuming here that experiences are typed (they are classified as being of the same kind or of different kinds) according to their intentional contents. 10 See Dretske (1995, ch.3) for a defence of this claim. 11 Dretske defends this implication (Dretske 1995, ch. 5). 12 See for instance Siewert (1998, p. 242-45). Siewert doesn’t discuss Swampman explicitly but he does look at theories which claim that environmental and behavioural links are necessary for the possession of accuracy conditions. I am applying an analogue of his objection to such accounts.

196


phenomenal features. The extra condition over and above the possession of phenomenal features which Dretske stipulates is unnecessary. A creature need not have ancestors whose sensory states have carried information in order for that creature to enjoy sensory states that are assessable for accuracy. Not all theories that endorse the causal connection condition have difficulties with Swampman. Tye’s version of the causal connection condition doesn’t face such difficulties for instance. He argues that a sensory state’s intentional features are fixed by the conditions it would track under optimal conditions. For creatures that do have an evolutionary history, Tye argues we should appeal to this history to determine what a creature’s representational states normally track. We have already seen how this won’t work in the case of Swampman so Tye appeals instead to what his states would track under “conditions of well-functioning” (Tye, 2000, p. 122). We might consider for instance under what conditions Swampman would form perceptual beliefs which enable him to satisfy his desires.13 We can then appeal to these conditions as fixing the contents of Swampman’s experiences. Moreover, we can say that Swampman misrepresents when these conditions of well-functioning do not obtain. Thus Tye can allow that Swampman’s sensory states have intentional features and continue to hold that the causal connection condition is necessary for the possession of intentional features. What will Tye say about creatures whose states have never tracked, and will never track, the environmental conditions they would track under conditions of well-functioning? Take the case of the brain-in-a-vat. We can imagine a distant planet whose inhabitants grow brains just like human brains which they hook up to a powerful computing machine. We can further imagine that the computing machine causes these brains to undergo “experiences” just like our own. Since these brains have been grown they will have never been in causal contact with the environmental conditions our brains normally track. We can further imagine that these brains will never come into causal contact with the kind of environment we inhabit. A compelling intuition says that a brain-in-a-vat would nevertheless share with us experiences which have the very same accuracy conditions as our own. It is just that a brain-in-a-vat’s sensory experiences are never accurate. Tye endorses this intuition. He says of the brain-in-a-vat that it: “has inaccurate perceptual experiences—things are not as they seem—because the brain is not in optimal perceptual conditions and the relevant brain-states are not tracking those features they would track, were optimal conditions to obtain.” (Tye, 2000, p. 64.) Now a proponent of the phenomenal intentionality thesis can make the same move against Tye as he made against Dretske. The brain-in-a-vat scenario is a case in which the causal connection condition fails to hold in a radical way—the 13

This is my own gloss on what Tye might mean by “conditions of well-functioning”. Tye himself doesn’t offer any details.


197

experiences of the brain-in-a-vat never happen under “optimal perceptual conditions”. Yet Tye is happy to concede that a brain-in-a-vat could enjoy experiences which share the very same intentional features as our own. If the causal connection condition is never satisfied for the brain-in-a-vat, if optimal conditions never hold, doesn’t this establish that the causal connection condition isn’t necessary for the possession of intentional features? If states can possess intentional features even though the creature to which they belong never tracks any environmental conditions surely this shows that tracking under optimal conditions isn’t a necessary condition for the possession of intentional features. Of course this is exactly what we proponents of the phenomenal intentionality thesis will say. We say that a state can have intentional features just in virtue of its phenomenal features. So long as the brain-in-the-vat shares states with the very same phenomenal features as our own, the brain-in-a-vat will have identical intentional features to us. Many philosophers find the brain-in-vat only dubiously conceivable, so it would be best not to base one’s case against Tye’s brand of reductive representationalism on such thought experiments. Setting to one-side the conceivability or otherwise of the brain-in-a-vat scenario then, it seems to me that in general theories which appeal to the causal connection condition operate with a false conception of intentionality. They make the mistake of modelling intentionality on cases of successful reference. They try to identify the conditions that obtain when we use words to successfully refer and they take these conditions to be definitive of intentionality. Brian Loar (2003), a proponent of something like the phenomenal intentionality thesis, has recently recommended a distinction between what he calls “reference” and “directedness”. Loar takes the property of “directedness” to be an objectindependent property of intentional states, by which he means that “directedness” is a property that doesn’t involve a relation to an object. “Reference” is however the paradigm of an object-dependent property: it makes no sense to talk of a mental state’s referring to something when the object of one’s mental state doesn’t exist. The property of reference depends on a relation holding between a representational item and its object while directedness does not involve any such relation to an object. Directedness is better thought of as the property of purporting to refer or apparently referring, a property a representational item can possess even when this item fails to stand in any relation to the object represented. Consider as an illustration of directedness, Husserl’s notion of bracketing, or what he sometimes called the “phenomenological epoché”. Husserl conceived of bracketing as a method for discovering the realm of the phenomenological. One is to suspend judgment on the existence of the objects of one’s intentional states, neither affirming nor denying the existence of those objects. Directedness or intentionality is a feature of our mental states that enables us to maintain this

198


attitude of neutrality. The contents of our intentional states are such that they do not require us to affirm, but nor do they require us to deny, the existence of the objects of those intentional states. We can remain quite neutral on question of the existence of the objects of our mental states. This is because it makes no phenomenal difference to a mental state whether that mental state genuinely presents its subject with an object or whether it only seems to. The directedness a mental state exhibits doesn’t depend on there being a relation a subject stands in to the object represented. To put this more precisely, it is not the case that there exists a relation R and an object y such that for all experiences x if x represents y then x bears relation R to y. Once we model intentionality on reference, we operate with a conception of intentionality that is incompatible with this attitude of neutrality. If we model intentionality on reference we will suppose that there is always some relation to an object or situation that fixes an intentional state’s content. We will then have to treat cases in which this relation fails to hold—cases of misrepresentation or reference failure—as aberrations, as deviations from a norm in which this relation holds. We will have to say that in cases of misrepresentation and reference failure there is some relation between the bearer of an intentional state and the world which accounts for the directedness of these states. Proponents of the causal connection condition have famously struggled, and so-far failed, to make this work. I suggest that this is because directedness and reference are distinct properties. Directedness is a non-relational property: it is not the case that there is some relation between the bearer of an intentional state and the world which accounts for an intentional state’s content. The mistake that proponents of the causal connection condition make is that they try to account for a non-relational property— directedness—by reference to a relational property—reference. The prospects for such a project are undoubtedly dim indeed. We would do better to recognise that an experience’s intentional content isn’t fixed by any relation to an object, be it causal or otherwise. Rather intentional content is something that derives from an experience’s phenomenal features. There is however a difficulty that arises for any attempt to treat the intentionality attaching to perceptual experiences as an object-independent or nonrelational property. The difficulty concerns the role that experience plays in grounding singular or demonstrative judgments. In the next section I will briefly sketch the difficulty and propose a way out of it.


199

3 Perceptual content and demonstrative thought14 Suppose someone points to a white rabbit and tells me ‘That’s Topsy’ and then points to another numerically distinct but qualitatively identical white rabbit and tells me “That’s Flopsy”. Now suppose that later I encounter Topsy and Flopsy and I think to myself “That’s Flopsy”, looking at one of the rabbits. Whether my demonstrative judgment is true or not will depend on whether it is Flopsy rather than Topsy I am seeing when I make my judgment. The truth of my demonstrative judgment will depend upon which of the rabbits I am experiencing. More specifically it will be conditional on whether the experience is of the particular rabbit—Flopsy—which my judgment concerns. It seems to follow that the contents of perception must be individuated by reference to particular objects—in this case particular rabbits. What the content of my experience is depends on which particular rabbit I am seeing. If the contents of experiences can be individuated by reference to particular things, the contents of experience cannot be object-independent. For recall that object-independence is a property a representational state has when it doesn’t involve a relation to a particular object. Thus the role that experience plays in acting as a reason for demonstrative judgments seems to threaten the account I have endorsed of perceptual content as object-independent. This threat extends to the phenomenal intentionality thesis. For the phenomenal intentionality thesis seems to entail that the contents of experience are general and that we do not need to make reference to particular things in specifying how an experience represents the world.15 Consider again my experience of Topsy and Flopsy. My experience is of two rabbits that are qualitatively indistinguishable. If my experience has its intentional content by virtue of its phenomenal properties we will have to say that my experiences of Topsy and Flopsy share the same intentional content. If these two experiences are qualitatively indistinguishable they will share all the same phenomenal properties. The phenomenal intentionality thesis entails that they will also share the same intentional content. Yet we have just argued that my experience of Topsy must have a different intentional content from my experience of Flopsy if we are to account for the role of perception in the making of demonstrative judgments. There are, so far as I can tell, two ways out of these difficulties for anyone who wants to treat sensory intentionality as a non-relational or object-independent property. The first concedes that perceptual content is object-dependent but argues that an experience’s phenomenal properties also vary with the particular object that 14

This section was prompted by questioning from Prof. Tim Williamson and Matt Nudds. For a defence of the claim that perceptual content is in this sense general see Davies (1992) and McGinn (1982).

15

200


is represented. This move doesn’t seem that plausible however. Consider the case of Topsy and Flopsy. We have said that an experience of Topsy is qualitatively indistinguishable from an experience of Flopsy. Yet if an experience’s phenomenal properties do vary with the particular object represented we will have to say that what it is like for me to experience Topsy is different from what it is like for me to experience Flopsy. Since Topsy and Flopsy are stipulated to be qualitatively indistinguishable this will be a difference that can’t be discovered through introspection. For introspection alone won’t tell me whether my experience is of Topsy or Flopsy. Thus a view that takes phenomenal properties to depend on the particular represented is committed to the possibility of differences in phenomenal properties that aren’t introspectably discoverable. Fortunately there is another way out for the defender of the phenomenal intentionality thesis.16 He can continue to hold that perceptual content is objectindependent and thus context-independent but combine this with the claim that the accuracy conditions for perceptual experiences are however object-dependent and thus context-dependent. Consider how this would work for the case of Topsy and Flopsy. We can continue to say that the intentional content for my experience is fixed by my experience’s phenomenal features. My experience’s phenomenal features do not tell me whether I am experiencing Topsy or Flopsy. However we must deny that my experience has its accuracy conditions solely in virtue of its phenomenal features. In addition we must consider the particular context in which the experience is tokened to determine which particular rabbit is the object of my experience and thus which rabbit we are to examine in assessing my experience for accuracy. Particulars get involved in an experience’s accuracy conditions, not through an experience’s intentional content, but through the context in which the experience is tokened. Consider by way of analogy indexical expressions such as “I”, “here” and “now”, each of which has a meaning which is context-invariant. “I” for instance means something like “the speaker or thinker of this very token”. It has this meaning across users—what “I” means in my mouth is just the same as what it means in your mouth.17 Now consider the contribution “I” makes to an utterance’s truth conditions. The contribution it makes will vary from person to person depending on which person it is that has made the utterance. My suggestion is that we think of experiences as also having a content that is independent of context just as “I” has a meaning that is independent of context. This is quite compatible with allowing that two experiences which share the same intentional content may have 16

M.G.F. Martin (2003) makes such a suggestion in a discussion of intentionalism more generally. 17 I am following Perry’s (1979/1993) treatment of indexicals.


201

accuracy conditions which vary with the context in which the experience has been produced. Does this make an experience’s intentional features depend on something other than its phenomenal features? I don’t think so. I have defined an experience’s intentional features as those features in virtue of which a subject of experience is assessable for accuracy. We can continue to work with this definition of intentional features. All that is necessary is that we modify how we think about an experience’s accuracy conditions. For I have just argued that an experience’s accuracy conditions are fixed by a combination of an experience’s phenomenal features and the context in which the experience is produced. I can continue to say that it is by virtue of an experience’s intentional content that an experience is assessable for accuracy, while insisting that we cannot read off an experience’s accuracy conditions from its intentional content alone. We also need to make reference to the context in which the experience has been tokened. Thus it is context plus an experience’s intentional features that fix an experience’s accuracy conditions. This is to concede nothing to the proponent of the causal connection condition. The thesis stands that an experience can have the intentional content it does just in virtue of its phenomenal features. The remainder of this paper will argue that the enactive theory of perception has the resources to explain how the phenomenal intentionality thesis can be true. First I should note something about the restrictedness of this claim. Most proponents of the phenomenal intentionality thesis have taken it to apply across the board to all states that are assessable for truth or accuracy.18 They take phenomenal features to attach across the board, not just to sensory experiences and sensations as is the norm, but also to thoughts, desires and acts of willing. Thus they take there to be a pervasive kind of intentionality that is object-independent. In what remains of this paper I will be taking the phenomenal intentionality thesis to be restricted in its application to sensory states. I will have nothing to say here about belief, desire or the will.

4 Towards an enactive theory of phenomenal intentionality Reductive representationalism held the promise of solving two problems for a naturalistic theory of mind. First it was going to tell us what it is for a physical state to exhibit intentionality. Next it was going to use this solution to solve the problem of consciousness. The phenomenal intentionality thesis casts doubt on this project. If conscious experiences have their intentional content just in virtue of their phenomenal features we can no longer use our favoured theory of intentional content to account for an experience’s phenomenal features. Now what of the 18

See for instance Horgan and Tienson (2002) and Siewert (1998, ch. 7 and 8).

202


problems of intentionality and consciousness? Without reductive representationalism the naturalistically inclined philosopher is left completely in the dark about the place of both intentionality and consciousness in the natural world. The hard problem of consciousness was the problem of explaining how the brain or some other physical organ could generate the kind of rich inner-life each of us enjoys. If there is a species of intentionality constitutively determined by an experience’s phenomenology, the hard problem just got even harder. We will have to explain not only the qualities of experience, we also have to explain how our brains are able to reach out and make contact with the multifarious objects of which we are conscious. In the remainder of this paper I will argue that the enactive theory of perception may hold a solution to this difficulty. The hard problem of consciousness as it arises for a proponent of the phenomenal intentionality thesis is the problem of explaining how a brain state could generate phenomenal intentionality. The enactive theory of perception begins by rejecting this formulation of the question. It rejects the claim that the vehicles of phenomenal intentionality are brain states.19 For according to the enactive theory, perceiving is an activity of the whole organism. It is a mistake to view perceptual experiences as something that happens within an animal’s brain. Instead we should think of perception as a skilled activity which may loop back and forth across brain, body and world.20 The appropriate question to ask, on this conception of perception, is how the whole animal in its interactions with the world could generate states with phenomenal intentionality. The answer the enactive theory returns appeals to a certain kind of practical knowledge or understanding the creature draws upon in perception. It is this practical knowledge which fixes an experience’s intentional content. However it does so only by making reference to an experience’s phenomenal features. Hence the account the enactive theory offers of perceptual content is also a theory of phenomenal intentionality. As a way into this approach to understanding perception consider a visual experience of a cube. It is impossible to see all of the sides of a cube in a single take as it where, what one perceives of the cube at any given moment are the sides facing one. There will however always be other sides of the cube which are hidden from view: the cube will always have profiles that go beyond those which are 19

See for instance Noë (2005, ch. 7) and Noë and Thompson (2004). For a more detailed discussion of the view of experience just sketched see Haugeland (1998, ch. 9); Hurley (1998, ch. 8-10); Noë (2004); O’Regan and Noë (2001); Rowlands (2002); and Thompson and Varela (2001). 20


203

presented to one at any particular moment.21 Now according to the enactive view when one sees a cube one has an understanding of the other available views the cube affords which one could take up by changing one’s relation to the cube. One has an understanding of how one’s movements or the movements of the cube will effect how one sees the cube. One understands for instance, how if one were to turn the cube in one’s hand, the sides which are currently hidden will come into view and one understands how the sides which are currently presented to one will become hidden. One understands how if one were to increase one’s distance from the cube it will appear to decrease in size, and if one were to move closer to the cube it will appear to increase in size. One understands how if one were to move one’s head to the left the position the cube is currently occupying in relation to one will shift to the right and how the angle from which the cube is seen will change accordingly. I will call understanding or knowledge of this kind “sensorimotor knowledge”. I will be arguing that it is sensorimotor knowledge of this kind that fixes, in a way to be explained, a sensory experience’s intentional content. Sensorimotor knowledge is knowledge of how the appearances an object presents to a subject vary with movement. According to the enactive theory which I will defend, an experience has its intentional content in virtue of a subject’s knowledge of how an experience’s phenomenal features would vary with movement, either on the part of the perceiver or of the object being perceived.22 I will begin by explaining the claim I have just made that sensorimotor knowledge is knowledge of how an experience’s phenomenal features vary with movement. Having done so I will explain how sensorimotor knowledge can fix an experience’s intentional content. Let us begin by considering the following observation that Hume makes in his Enquiries: “The table which we see, seems to diminish, as we remove farther from it: but the real table, which exists independent of us, suffers no alteration: it as, therefore, nothing but its image, which was present to the mind. These are the obvious dictates of reason, and no man, who reflects, ever doubted, that the existences, which we consider, when we say, this house or that tree, are nothing but perceptions in the mind, and fleeting copies or representations of other existences, which remain uniform and independent.” (Hume, 1975, p. 152)

21

This is a familiar phenomenological observation of a kind which Husserl made much use of in developing his account of perceptual intentionality. See for instance Husserl (1973, §19) 22 Here I am assuming that there is something that is being perceived. I will have a little to say about non-veridical cases later in this paper.

204


Hume takes the observation that the table appears to diminish in size as we move away from it, and that it appears to increase in size as we move closer to it, to show that the objects of perception cannot be mind-independent objects. The table doesn’t change in size with our movements, its size remains the same, and yet something appears to change. So what appears to change he thinks cannot be the table but must instead be something like an image of the table that the perceiver has in mind. We see the table by seeing an image of the table in our minds. The enactive theory of perception agrees with Hume that perception is a twostage process: we see the size of the table by seeing what I shall call the table’s “apparent properties”. However the enactive view disagrees with Hume that what we are directly and immediately acquainted with in perception is something akin to an image. Indeed the enactive view downplays the role of what Hume called “images” and what we now call “internal representations” in perception. The enactive view claims instead that what we are aware of when we see the table’s apparent size is the table and not an image or internal representation of the table. Now this claim would seem to land the enactive theorist with precisely the contradiction which Hume sought to avoid. If what we are aware of in perception is the table we seem to be committed to saying that we are aware of something that gets smaller or larger with movements and something that stays the same size with our movements. But of course we cannot be aware of something that has both these properties simultaneously. The enactive theorist’s solution to this puzzle is to distinguish between the real size of the table understood as an intrinsic property of the table and the apparent size of the table as a relational property—the property of being seen by a perceiver from a certain distance.23 According to the enactive theory, we come to represent the table’s real size in perceiving the table’s apparent size and how the apparent size which the table presents, changes with our movements. We come to represent the table’s actual size in our sensory exploration of the table. For in the variation of appearances the table presents through our exploration of it there is something that is invariant, and this is the table’s real size. It is this invariance that we pick up on as we actively explore the table through our movements in relation to it. Sensorimotor knowledge is knowledge of how the apparent properties that a thing presents change with one’s movement. It is an experience’s phenomenal features that fix how a thing appears to a subject—the properties the thing appears to have. Thus we can say that sensorimotor knowledge is knowledge of how an experience’s phenomenal features will change with one’s movement. I have assumed here that talk of an experience’s phenomenal features and talk of a thing’s apparent properties are equivalent in meaning. That is, I have assumed that both expressions pick out one and the same property. Phenomenal features are 23

See for instance Noë (2002) and (2005, ch.3.3).


205

commonly understood as qualia, where an experience’s qualia are taken to be the determinants of what it is like to undergo a conscious experience, and are conceived of as introspectively discoverable, intrinsic properties of experience. Yet by suggesting that we understand an experience’s phenomenal features as coextensive with an object’s apparent properties, I am committed to an understanding of phenomenal features as relational properties.24 The purpose of introducing apparent properties is to capture those features of experience which are not accounted for just by talking about what an experience represents—the objects and their properties represented. This is precisely the role that was played by appeal to qualia or what Peacocke (1983) called “sensational properties” of experience. Consider again the example from Hume of the table which appears to diminish in size as the viewer increases her distance from it. We cannot account for a change of this kind by reference to what the viewer’s experience represents. For what she represents in this case is a table that retains a constant size. If we are to account for what it is like to undergo this experience we must make reference to how the table is being represented. The fact that the table appears to get smaller is a part of how the table is being represented. Apparent properties are introduced, just as qualia were, to account for how an object of perception is represented. Apparent properties are environmental properties, properties an object has only in relation to other objects including a perceiver’s body under certain illumination conditions. However apparent properties are also properties that exist only in relation to subjects of experience. There is something it is like to perceive a thing’s apparent properties. Consider the property that a penny has of appearing to be elliptical when viewed from a certain angle. There is something that it is like to perceive the penny as being shaped elliptically. What it is like to perceive a penny that appears to have an elliptical shape is qualitatively different from what it is like to perceive a penny that appears to be circular. We cannot characterise apparent properties purely in environmental terms—that is purely in terms of relations between objects, and the conditions under which those objects are viewed. We must also make reference to subjects of experience—subjects for whom there is something it is like to perceiver a thing’s apparent properties. It seems to me that Noë, one of the leading exponents of the enactive theory of perception, somewhat errs on this point. He suggests that we can think of a property like apparent shape as the ‘shape of the patch needed to occlude the object on a plane perpendicular to the line of sight.’ (Noë 2005, p. 83) Noë’s treatment of apparent properties places them firmly on the world side of the perceiver-world 24

It might be thought that the claim that phenomenal features are relational properties is incompatible with my earlier characterisation of intentional features as object-independent properties. I will deal with this objection in the final part of my paper.

206


relation. In doing so he downplays the fact that apparent properties have a subjective or qualitative dimension to them. It is this qualitative dimension that I mean to bring out by talking of an experience’s phenomenal features. Phenomenal features as I understand them fall on the perceiver side of the perceiver-world relation in terms of which we must characterise apparent properties. Conceding that phenomenal features have a subjective dimension to them doesn’t entail thinking of these features as belonging to experiences intrinsically; we can continue to think of phenomenal features as relational properties. It’s just that we must think of them as properties that belong to objects only in relation to subjects of experience. I want to leave to one side the question of whether the enactive theory of perception gives us an adequate explanation of the nature of phenomenal features, thus construed. My excuse is that I am interested in giving an account of the contents of experience according to which an experience has its content by virtue of its phenomenal features. The question of whether phenomenal features can be given a reductive explanation along the lines that the enactive theory proposes is one that is deserving of separate discussion. It suffices for our current purposes to have a grasp of what is meant by phenomenal features, and I believe I have said enough about this issue for us to see that phenomenal features might reasonably be understood on the model of apparent properties. It is time then that we explained how sensorimotor knowledge could fix an experience’s intentional content. It is on this claim that my second thesis in this paper rests—namely that the enactive theory of perception shows us how the phenomenal intentionality thesis can be true. What does it mean to claim that an experience’s intentional content is fixed by a subject’s sensorimotor knowledge? Consider a representation of a shape property first, for instance circularity. According to the enactive theory one represents a thing’s circularity in representing this thing’s apparent shape and the ways in which the thing’s apparent shape changes either with one’s own movements or with the movement of the thing one is representing. By representing a thing’s apparent shape and the changes that take place in its apparent shape properties with movement, one comes to represent a thing that is circular. Now of course one needn’t actually explore all the possible vantage points from which a thing can be viewed in order to visually represent a thing’s shape. According to the enactive theory, one has a practical grasp of how the apparent properties a thing presents to one would change were one to alter one’s spatial relation to a thing. One has expectations about how a thing would look, or feel to the touch, or sound, were one to alter one’s relation to that thing. These expectations are assessable for accuracy—they will either be confirmed or disconfirmed through further exploration of the object.


207

Suppose for instance, that I am presented with an object that appears to have an elliptical shape. My sensorimotor knowledge is such that I have a practical grasp of the fact that things which appear elliptical when viewed from one angle can appear circular when viewed from another angle. Thus I will form the expectation that if I were to change the angle from which this object is currently being viewed, I would experience something that seems to be circular. This expectation will be confirmed if I am seeing something that is circular and disconfirmed if I am not seeing something circular but am seeing something that is in fact elliptical. Sensorimotor knowledge fixes an experience’s intentional content by fixing expectations about the apparent properties a thing presents, and how these apparent properties will change with movement. I mentioned in passing above how an animal represents a thing’s invariant properties—its real size, shape, colour or whatever—by representing the variations in the apparent properties a thing presents. There is a lawful dependence between the apparent properties a thing presents and the changes that take place in those apparent properties with an animal’s movement or with the object’s movement. The exact character of this lawful dependence will be determined by a thing’s invariant properties. It is for this reason that an animal can represent a thing’s invariant properties by representing the changes that take place in its apparent properties with movement. An experience has as its intentional content that a particular thing has certain invariant properties. Thus sensorimotor knowledge fixes what an experience represents—the invariant properties it represents a thing as having—by fixing expectations about how the apparent properties a thing presents will change with the animal’s movement. Since we come to represent a thing’s real or invariant properties by representing changes in its apparent properties and sensorimotor knowledge fixes how we represent the latter, sensorimotor knowledge can also be said to fix how we represent a thing’s invariant properties. How does this argument bear on the phenomenal intentionality thesis? I have argued that an experience represents a thing’s invariant properties by representing how its apparent properties will change with movement. The phenomenal intentionality thesis says that an experience has its intentional content just in virtue of its phenomenal features. Now we can see that the enactive theory is also committed to a thesis along these lines. I have argued above that to represent the changes in a thing’s apparent properties brought about through movement just is to represent changes in an experience’s phenomenal features. I have also argued that we represent a thing’s invariant properties by representing changes in a thing’s apparent properties brought about through movement. It follows that to represent changes in an experience’s phenomenal features just is to represent a thing’s invariant properties. This is what the phenomenal intentionality thesis claims. It says that an experience has its intentional content—it represents a thing’s invariant properties—just in virtue of its phenomenal features.

208


I want to finish up by briefly considering an objection. The objection concerns the characterisation I have given of an experience’s phenomenal features as relational properties. If we combine this conception of phenomenal features with the phenomenal intentionality thesis we seem to be required to conceive of an experience’s intentional content as a relational property. Yet in section 2 and 3 I characterised intentional content in terms of directedness which I suggested we think of as an object-independent or non-relational property of experience. Thus it looks like an enactive account of phenomenal intentionality is inconsistent with my earlier characterisation of intentional content as directedness. However once we recognise the role that expectations play in fixing an experience’s intentional content I think we will see that this objection is unfounded. Recall that sensorimotor knowledge fixes an experience’s intentional content by fixing a set of expectations about how an experience’s phenomenal features would change with movement. We make a mistake when we think of perception on the model of a series of static snapshots of the world, each of which represents a scene in full detail. In fact perception has a dynamic structure in which expectations are constantly being formed about what further active exploration would reveal of an object. There is an important sense in which what an experience represents of an object is fixed in advance of actually perceiving an object. It is fixed by one’s sensorimotor knowledge. Recall the example of the cube which I used above to introduce the enactive theory. At any given moment what is given to us in experience are the cube’s facing sides. Yet what we represent in experience isn’t just the facing sides of the cube but the cube in its entirety including those sides which are not currently visible to us. We represent the absent sides of the cube by drawing on our practical understanding of how the cube would appear where we to change our relation to it. A subject comes to visually represent a cube by forming expectations about what further visual exploration would reveal of it. These expectations are either confirmed or disconfirmed by the subject’s active exploration of the object of its experience. It is in this sense then that the contents of perception are object-independent. The contents of experience are fixed prior, in a temporal sense, to the perceiver entering into a relation with an object. For the contents of experience take the form of expectations the perceiver forms about what he would in the future experience of an object. These expectations concern changes in an experience’s phenomenal features. Certainly it is true that according to the enactive theory these phenomenal features take the form of relations between the perceiver and its environment. However the process by which an experience’s intentional content is fixed doesn’t involve relations of this kind. It is instead a matter of the perceiving animal drawing on her sensorimotor knowledge to form expectations. It is these expectations which fix what an experience represents.


209

Now it might be objected that the enactive theory departs from the phenomenal intentionality thesis in claiming that it is expectations about changes in phenomenal features which fix intentional content, not phenomenal features themselves.25 Aren’t the expectations an animal forms based on its sensorimotor knowledge now doing all the work in fixing intentional content? The work expectation does in my account is accomplished only because of an experience’s phenomenal features. The objects of such acts of expectation are precisely an experience’s phenomenal features—what we anticipate is how an experience’s phenomenal features will change with movement. So while I do indeed want to say that it is ultimately expectations that fix intentional content, since these expectations concern changes in phenomenal features my account remains a version of the phenomenal intentionality thesis. I am committed to the claim that it is an experience’s phenomenal features—via our expectations about how phenomenal features change with movement—that fix an experience’s intentional content. Finally I want to raise a more general worry one might have with the enactive theory. I have claimed that sensorimotor knowledge fixes an experience’s intentional content. Yet animals of different species will differ in their sensorimotor knowledge. Consider for instance two species of animal that have eyes located at different positions on their heads. We can suppose that one species of animal has eyes located at the front of their head while the other has eyes situated on the side. The same movement of the head performed by animals of these different species will give rise to very different changes in the apparent properties a thing presents. Equally different movements of the head will give rise to the same change in apparent properties. Yet we can suppose that despite this difference in sensorimotor knowledge the two species of animal will nevertheless represent a particular object in the same way. They will both represent the roundness of round objects for instance. It seems to follow then that differences in sensorimotor knowledge do not translate into differences in intentional content. Thus sensorimotor knowledge cannot be what fixes an experience’s intentional content.26 I don’t believe this objection is as devastating as it might at first seem. Recall that sensorimotor knowledge fixes intentional content by fixing a set of expectations about how the apparent properties a thing presents vary with movement. Now different species of animals will undoubtedly form different expectations as a result of differences in gross bodily form. However what matters for the fixing of the contents of experience is that there is a lawful dependence between variation in the apparent properties a thing presents and the animal’s movement. So long as there is a lawful dependence of this kind which underwrites a perceiving animal’s sensorimotor knowledge, an animal will be able to represent 25 26

My thanks to the editors for pressing this objection. My thanks to Prof. Christopher Peacocke for raising this difficult challenge.

210


a thing’s invariant properties by representing a variation in the apparent properties a thing presents. So long as sensorimotor knowledge is a means of picking up on such lawful dependencies both animals will be able to represent a thing’s invariant properties by representing variation in the appearances a thing presents. What of the hard problem of consciousness then as it arises for proponents of the phenomenal intentionality thesis? Has the enactive theory found a solution to the hard problem? I think the enactive theory has shown us how a version of the phenomenal intentionality thesis can be true. Whether it has succeeded in solving the hard problem will depend on whether the enactive theory gives us a satisfactory account of an experience’s phenomenal features. I think there is room for scepticism on this count. The enactive theory, as I mentioned in passing above, tends to place a good deal of emphasis on the world side of the perceiver-world relation in its account of phenomenal features. The result is that it tells us very little about what makes an experience’s phenomenal features subjective. This however is a subject for another paper. The enactive theory does at least have a promising story to tell about how the contents of experience can be fixed. It is a story that is at least consistent with a non-reductive representationalism. So even if the enactive theory turned out not to have a satisfying reductive story to tell about in virtue of what experiences have there phenomenal features it would still have made a contribution to a naturalistic solution to the hard problem of consciousness.

Bibliography Byrne, A. (2001). “Intentionalism Defended”, in Philosophical Quarterly 110: 199240. Carruthers, P. (2000). Phenomenal Consciousness: a Naturalistic Theory. Cambridge: Cambridge University Press. Chalmers, D. (2004). “The Representational Character of Experience”, in B. Leiter (ed.), The Future for Philosophy. Oxford: Oxford University Press. Crane, T. (2003). “The Intentional Structure of Consciousness”, in Q. Smith and A. Jokic (eds.), Consciousness: New Philosophical Essays. Oxford: Oxford University Press. Davies, M. (1992). “Perceptual Content and Local Supervenience”, in Proceedings of the Aristotelian Society XCII: 21-45. Dretske, F. (1995). Naturalising the Mind. Cambridge, Mass.: MIT Press. McGinn, C. (1982). “The Structure of Content”, in A. Woodfield (ed.), Thought and Object. Oxford: Clarendon. Noë. A. (2005). Action in Perception. Cambridge, Mass.: MIT Press. ——— (2002). “On What we See”, in Pacific Philosophical Quarterly 83: 57-80. Noë, A. and Thompson, E. (2004). “Are there Neural Correlates of Consciousness?”, in Journal of Consciousness Studies 11: 3-28.


211

Haugeland, J. (1998). “Mind Embodied and Embedded”, reprinted in his Having Thought: Essays in the Metaphysics of Mind. Harvard, Mass.: Harvard University Press. Harman, G. (1990/1997). “The Intrinsic Quality of Experience”, reprinted in N. Block, O. Flanagan, and G. Gűzeldere (eds.), The Nature of Consciousness. Cambridge, MA: MIT Press. Hume, D. (1975). Enquiries Concerning Human Knowledge (ed. L.A. SelbyBigge). Oxford: Clarendon. Hurley, S. (1998). Consciousness in Action. Harvard, Mass.: Harvard University Press. Husserl, E. (1973). Cartesian Meditations (trans. D. Cairns). The Hague: Martinus Hijhoff (now Kluwer Academic Publishers). ——— (1913/1970). Logical Investigations (trans. J.N. Findlay). London: Routledge. Loar, B. (2003). “Phenomenal Intentionality as the Basis of Mental Content”, in M. Hahn and B. Ramberg (eds.), Reflections and Replies: Essays in Honour of Tyler Burge. Cambridge, Mass.: MIT Press. Lycan, W. (1996). Consciousness and Experience. Cambridge, Mass.: MIT Press. Martin, M.G.F. (2003). “Particular Thoughts and Singular Thoughts”, in A. O’Hear (ed.), Thought and Language. Cambridge: Cambridge University Press. O’Regan, J.K., and Noë, A. (2001). “A sensorimotor approach to vision and visual consciousness”, in Behavioural and Brain Sciences 24: 939-73. Peacocke, C. (1983/1997). “Sensation and the Content of Experience: a Distinction”, from his Sense and Content (Oxford: Clarendon Press), reprinted in N. Block, O. Flanagan, and G. Gűzeldere (eds.), The Nature of Consciousness. Cambridge, MA: MIT Press. Perry, J. (1979). “The Problem of the Essential Indexical”, in Noûs 13: 3-21. Rosenthal, D. (1986). “Two Concepts of Consciousness”, in Philosophical Studies 49: 329-359. Rowlands, M. (2002). “Two Dogmas of Consciousness”, in Journal of Consciousness Studies 9: 158-80. Siewert, C. (1998). The Significance of Consciousness. Princeton, NJ: Princeton University Press. Thompson, E. and Varela, F. (2001). “Radical Embodiment: Neural Dynamics and Consciousness”, in Trends in Cognitive Science 5: 418-425. Tye, M. (2000). Consciousness, Colour, and Content. Cambridge, Mass.: MIT Press. Zahavi, D. (2003). “Intentionality and Phenomenality: A Phenomenological Take on the Hard Problem”, in E. Thompson (ed.), The Problem of Consciousness: New Essays in Phenomenological Philosophy of Mind. Calgary, Alberta: University of Calgary Press.

CONTRIBUTORS Asunción Álvarez received her BA in linguistics and her MA in philosophy from the Universidad Complutense de Madrid. After some early work in the philosophy of biology, she is currently writing a PhD dissertation on Fregean concepts at King's College, London. Gloria Ayob (BA Manchester, MA Warwick), is completing a PhD in the philosophy of psychology. Gloria is working on the problem of how empirical findings in the brain sciences bear on the philosophical attempt to understand mindedness and intentionality, using Wittgenstein's critique of explanatory projects as a framework for her research. Treasa Campbell (BA, MA Mary Immaculate College, Grad. Dip. Phil. University College Cork, MA University of Limerick) completed a research MA in philosophy in 2004 which looked at Hume’s account of human belief-formation. She is currently working on a PhD proposal which deals with naturalism in epistemology, drawing on contemporary work in philosophy of science and philosophy of mind. Ezio Di Nucci (BA Roma, MSc Edinburgh) is completing a PhD in the philosophy of action at Edinburgh University. His work investigates the role of automaticity within agency. His research interests also include animal ethics and political philosophy. Hans-Johann Glock (MA Berlin, D.Phil Oxon) has recently taken up a Chair at the University of Zurich. He was previously Professor of Philosophy at the University of Reading. He has published widely on Wittgenstein and in the philosophy of mind and language. He is the author of A Wittgenstein Dictionary and Quine and Davidson on Language, Thought and Reality. Richard Hamilton holds a BA in philosophy from Trinity College, Dublin, and an MSc in philosophy from Edinburgh University. He is currently working on a PhD on Heidegger's philosophy of technology at Trinity College, Dublin, where he holds an Ussher Fellowship. His other areas of interest are Heidegger's philosophy, in particular his readings of the Greeks, Aristotle, Plato and Ancient philosophy, and the philosophy of science and technology. Nick Jones has recently submitted his PhD thesis at the University of Nottingham. His research concerns the relationship between imagination and perceptual belief,


213

bridging epistemology, philosophy of mind and philosophy of psychopathology. Nick is currently teaching at Nottingham, and for the Open University. Julian Kiverstein received his PhD from Edinburgh University in 2006 and is currently working on papers on phenomenology and philosophy of mind. Conor McHugh (BA Dublin, MSc Edinburgh) is writing a PhD dissertation on self-knowledge at Edinburgh University. His research interests also include philosophy of mind and the epistemology of perception. Francisco Pereira (Lic. Phil. Universidad Católica de Chile, MA King’s College, London) is completing a PhD in philosophy of perception at King’s College, London. Pereira’s research interests are philosophy of mind, epistemology and the history of early modern philosophy. Markus E. Schlosser (Mag. Phil., University of Vienna) is finishing a PhD at the University of St. Andrews. His thesis is on the metaphysics of rational and autonomous agency. Other research interests include philosophy of mind, moral theory and the nature of normativity. Dave Ward (MA, MLitt St. Andrews) is a PhD student in philosophy at the University of Edinburgh. He works on the philosophy of consciousness, mental representation, and cognitive science.

INDEX

a priori, 25-26, 28 abilities, 37-38, 48-49, 54-59 act of pointing, 74, 83-84 Alston, W.P., 24-26, 31 Álvarez, A., 31 anomalous monism, 6 apparent properties, 204 argument from illusion, 21-22 Aristotle, 54 Artificial Intelligence, 9-11, 125-137 Good Old Fashioned AI, 125-127, 132 Strong AI, 126 Ayer, A.J., 22, 31 Ayob, G., 29 Being-in-the-World, 126, 134-136 beliefs, 38, 42-44, 52, 63-72 belief formation, 62 belief-feeling, 64 irrational beliefs, 62-65 natural beliefs, 62-72 non-rational beliefs, 64 Block, N., 7, 15, 31, 34, 120, 123 bodily sensations, 191 Bolter, D., 133 brain phenomena, 2 brain states, 201 Brandom, R., 97 Brentano, F., 6, 11-17, 32 Brooks, R., 133 Burge, T., 176 Butler, R.J., 62-63, 68, 70 Byrne, A., 17, 32 Campbell, J., 26-32, 74-87 Campbell, T., 16, 25

causal closure of the physical, 142 causal exclusion argument, 6, 139155 causal relations, 42, 46 causal theory of perception, 178 Chabris, C.S., 180-181, 188 Chalmers, D., 2-3, 7, 32, 105-108, 120-123, 192, 209 change blindness, 28, 180-188 Clark, A., 2, 28, 31-33, 104, 113, 117, 120, 123 close resemblance, 178, 182, 184187 cognition, 98 Commonality Thesis, 160-163, 171, 175-176 concepts, 36, 38, 40, 42, 45, 49, 57, 60, 61, 74-76, 80-83, 89-100 concept-possession, 30, 38, 49, 50-55, 58 conceptual capacities, 75-80, 83 conceptual scheme, 80, 85 demonstrative concepts, 74-76, 79-81, 86-87 perceptual concept, 91 pseudo-concept, 92, 94 sortal concept, 81-82 spurious concept, 90, 92-94, 96 theory of concepts, 89-90, 92, 9697, 99, 100 connectionism, 125, 132 conscious attention, 74-75, 79-87 conscious experience, 3, 7, 74, 104, 106 visual experience, 191-192, 201


consciousness, 2-3, 7, 9, 10, 189192, 200-201, 209-210 content, 36, 41, 43-44, 46 intentional content, 4, 12-23, 29, 158, 189-208 representational content, 39 problem of, 13-14, 29, 31 counterfactual dependence, 28, 153155, 180-187 counterfactuals, 180, 182-186 Crane, T., 37-39, 53-56, 59, 192, 209 cybernetics, 132 Dasein, 135-136 Davidson, D., 4-6, 32, 51, 59-60 Dawkins, R., 133 deflationism, 104-123 demonstrative terms, 75-79, 82-87 Dennett, D.C., 1, 16, 32, 106-109, 114-119, 123 Der Spiegel, 132 Descartes, R., 1, 39, 129 desire, 38-39, 43-44 determinables, 148-149 Determination Theory, 92-97 deterministic laws, 3-5 direct availability, 109, 120 disjunctivism, 160-161, 166, 176 dispositions, 48, 53, 66-70 Dretske, F., 14, 32, 189, 193-195, 209 Dreyfus, H., 126, 132-137 dualism, 4-5 Dummett, M., 89 Edge, D., 128, 133 Eilan, N., 1, 32 eliminativism, 106, 108, 122 epiphenomenalism, 5 epistemic norms, 63, 71 Evans, G., 25, 32 explanatory gap, 104-105 externalist theories of mind, 37

215

Fodor, J., 11, 30-32, 37-46, 50-60 Frege, G., 39-40, 45, 51, 58, 60, 8990, 94, 98, 101 Gaskin, J.C.A., 62, 64-65, 68 Geach, P., 48, 58, 60, 90, 101 Glock, H-J., 30, 36, 38, 43, 45, 49, 52, 55, 58, 60 grammar, 75, 78, 83-87 grand illusion, 181 Grice, P., 178, 188 Hacker, P.M.S., 47, 51, 59-60 Haller, R., 36, 60 hallucination, 158, 161-169, 172180, 183, 187 veridical hallucination, 178-179, 183 Hamilton, R., 11 hard problem, 105 Harman, G., 17, 32 Haugeland, j., 13, 32 Heidegger, M., 125-138 Being and Time, 127-128, 135137 Heidegger's philosophy of technology, 126 "The Question Concerning Technology", 128-134, 137 Highest Common Factor, 161-171, 175-176 Hinton, J.M., 160, 169, 176 Hume, D., 10, 16, 25-28, 32, 62-72, 202-204, 210 Husserl, E., 56, 60, 129, 192, 196, 202, 210 Hyman, J., 46, 54, 60 Ihde, D., 128 illusion, 158 implicit conceptions, 31, 90, 96-100 ineffability, 108-109, 114 information-processing mechanisms, 75, 79-87

216


instincts, 63, 67-72 intentional theory of perception, 157161, 164-165, 171 intentionalism, 17, 18, 19, 20, 22, 23 intentionality, 1-2, 11-29, 75-79, 86, 189-209 internal inconsistency, 93-94 intersubjectivity, 36 intrinsicality, 107-111, 114, 116, 120, 122 Jackson, F., 5-9, 32 James, W., 27-28 Kant, I., 1, 4, 24-28, 33 Kemp Smith, N., 62-63, 66, 70 Kenny, A., 36, 54, 60 Kim, J., 2, 33, 140-156 Kiverstein, J., 18 knowing how, 37, 53 knowledge, 91, 94-99 Künne, W., 49, 60 Langsam, H., 168-169, 177 language of thought, 41, 45 language practice, 75, 78-79, 83-84, 86-87 Laurence, S., 40-45, 60-61 Levin, D.T., 180-181, 188 Lewis, D., 8, 26-27, 33, 178-188 Lewontin, R., 128, 130-131, 137 Libet, B., 39, 60 linguistic community, 91 linguistic expressions, 36 Loar, B., 190, 196, 210 logic, 94-95 intuitionistic logic, 95 Lukacs, G., 131 Margolis, E., 40-45, 60-61 Martin, M.G.F., 19, 22-23, 33, 160161, 167, 172-177 Marx, K., 130 materialism, 3 McCormick, M., 62-67

McCulloch, G., 165, 177 McDowell, J., 20, 22, 25, 28, 33 McGinn, C., 106, 124 meaning, 74-86 mental anomalism, 4-5 mental causation, 38, 142, 155-156 mental contents, 96 mental episodes, 36 mental events, 4-6, 139-155 mental phenomena, 2, 13-15 mental processes, 1, 96 mental properties, 4-6, 140-147 mental representation, 38-44, 89, 157, 174 mental states, 38, 42, 45, 143, 149, 152-157, 169, 171, 189, 196-197 conscious mental states, 190 phenomenal conscious state, 189 Mentalese, 41-45 Merricks, T., 141, 152-156 metaphorical conception of the mind, 126 metaphysical disjunctivism, 157, 160-161, 169-176 Millikan, R.G., 14, 33 Minsky, M., 127 modes of representation, 36 Mumford, S., 129 natural instincts, 63, 68-71 neo-Fregeans, 37 Neo-Fregeans, 90 neurophysiological processes, 36, 39 Noë, A., 2, 18, 28-33, 109, 119, 124, 180-181, 188, 201-204, 209-210 non-reductive physicalism, 6, 139155 nonsensical, 47 non-truth-preservation, 93 Okrent, M., 126-138 ontic, 127 ontological, 127


ontological relativism, 79, 81 overdetermination, 139-141, 146155 Pavlov, I., 66-67 Peacocke, C., 17, 20, 26, 29, 31, 33, 44, 51, 57, 61, 89-101, 189-191, 204, 208, 210 Peacocke's Principle of Dependence, 90 Peirce, C.S., 41, 46-47, 61 perceptual experience, 179, 181, 191 visual experience, 179, 181-185 perceptual states, 157-166 Pereira, F., 21, 23 personal level, 100 personal-level, 113-121 Pettit, P., 111-119, 124 phenomenal consciousness, 104-113, 120, 122, 189, 190 phenomenal features, 191-209 phenomenal indiscriminability, 160, 167 phenomenal intentionality, 190, 192, 198, 200-201, 206-209 Phenomenal Intentionality Thesis, 192 phenomenal properties, 17-18, 158, 189-190, 198 physical events, 4-6, 139-155 physical properties, 140-150 physicalism, 3, 7, 9, 139-155 Pitcher, J., 171-172, 177 Planck, M., 131 Plato, 89 Popkin, R., 71 possession conditions, 90-100 primitively compelling, 91-97 pragmatism, 37-38, 42, 49-58 pre-reflective attention, 74-87 non-intentional, 76-78 pre-reflective acquaintance, 74-83

217

Price, H.H., 20, 33, 39-40, 48, 52, 61 privacy, 108, 118 properties, 36, 45-48, 56 propositional attitudes, 3, 15-18, 3743, 90, 99 propositions, 36, 41, 43, 49, 56 psychophysical laws, 3-6 Putnam, H., 1, 33, 170, 176-177 qualia, 3, 7-9, 15, 17, 104-123 Quine, W.V., 37, 60, 72, 79, 85 rationality, 96-101 irrational judgement, 97 rational judgement, 97-98 reductionism, 3, 9 reductive representationalism, 193, 196, 201, 209 reductive representationalists, 189-192 reference, 74-98 common sense picture, 74-76, 85 referring terms, 74, 77 stationing, 77-84 reflective self understanding, 126136 Reich, S., 133 reliabilism, 86-87 Rensink, R.A., 28, 33 representational content, 160, 167, 191-194 representational theory of mind, 3756 representationalism, 38-56 representations, 36-55 Rey, G., 36, 44, 60-61 rival causes, 141-154 Robinson, H., 1, 33, 163, 177 Rorty, R., 1, 14, 33 rule-following, 75, 84-85 rule-following paradox, 75, 84-85 Russell, B., 20-22, 33, 74-87 Principle of Acquaintance, 74

218


Ryle, G., 37, 39 Sainsbury, M., 159, 174, 177 scepticism, 25, 62-65 Schlanger, J., 134 Schlosser, M., 4, 6 Searle, J., 1, 2, 6, 10, 14, 16, 32-33 Sellars, W., 13, 25, 34 semantic value, 92-94 semantics, 75-76, 85-86 sense, Fregean, 89 sense-data theory, 22 shareability, 40-48 Simon, H., 127 Simons, D.J., 180-181, 188 simplicity, 109, 118 Smart, J.J.C., 2, 34 Snowdon, P., 159-160, 165, 177 Socrates, 36, 42-43, 49, 50 sortal classification, 79, 81 Soteriou, M., 159, 165, 177 standing-reserve, 130 state of affairs, 93, 98 Strawson, P.F., 25, 34, 38, 49, 58, 61 subpersonal level, 9, 25, 82, 96, 99100, 112-114

subpersonal processes, 96 Swampman, 194-195 technological metaphor, 133 Technology, The Question Concerning, 130 Thau, M., 165, 177 token-identity, 139, 144, 147 tonk, 93, 94 truth tables, 99 truth value, 96 Tugendhat, E., 56, 61 Tye, M., 15-17, 34, 189-196, 210 uncritical realism, 74-75 universals, 36, 49 veridical perceptions, 167-176 visual experience, 157-176 Ward, D., 1, 8-9, 15 Wiggins, D., 81 Williamson, T., 37, 61, 169, 177 Wittgenstein, L., 1- 2, 34, 37, 39, 51, 59, 74-89, 101, 126, 134 Wolff, R.P., 66-70 Yablo, S., 141-156 zombie, 105-106, 120, 123

Content, Consciousness, and Perception: Essays in Contemporary Philosophy of Mind

Perception, Hallucination, and Illusion (Philosophy of Mind)

Content and Consciousness

Consciousness, Color, and Content

Content and Consciousness

Content and Consciousness

The Character of Consciousness (Philosophy of Mind)

Contemporary Debates in Philosophy of Mind

Vision and Mind: Selected Readings in the Philosophy of Perception

Vision and Mind: Selected Readings in the Philosophy of Perception

Vision and Mind: Selected Readings in the Philosophy of Perception

Mind, Reason and Imagination: Selected Essays in Philosophy of Mind and Language (Cambridge Studies in Philosophy)

Mind, Reason and Imagination: Selected Essays in Philosophy of Mind and Language (Cambridge Studies in Philosophy)

Mind, Reason and Imagination: Selected Essays in Philosophy of Mind and Language (Cambridge Studies in Philosophy)

Consciousness: From Perception to Reflection in the History of Philosophy (Studies in the History of Philosophy of Mind, 4)

Consciousness and Mind

Phenomenal Concepts and Phenomenal Knowledge: New Essays on Consciousness and Physicalism (Philosophy of Mind Series)

Phenomenal Concepts and Phenomenal Knowledge: New Essays on Consciousness and Physicalism (Philosophy of Mind Series)

Contemporary Debates in Philosophy of Mind (Contemporary Debates in Philosophy 8)

Sensation and Perception: A History of the Philosophy of Perception

Contemporary Debates in Philosophy of Biology (Contemporary Debates in Philosophy)

Action in Perception (Representation and Mind)

Essays on Nonconceptual Content

Matter and consciousness: a contemporary introduction to the philosophy of mind

Matter and Consciousness, Revised Edition: A Contemporary Introduction to the Philosophy of Mind

Purple Haze: The Puzzle of Consciousness (Philosophy of Mind Series)

Matter and Consciousness: A Contemporary Introduction to the Philosophy of Mind

Purple Haze: The Puzzle of Consciousness (Philosophy of Mind Series)

Purple Haze: The Puzzle of Consciousness (Philosophy of Mind Series)

Neurochemistry of Consciousness: Neurotransmitters in Mind

Content, Consciousness, and Perception: Essays in Contemporary Philosophy of Mind

Perception, Hallucination, and Illusion (Philosophy of Mind)

Content and Consciousness

Consciousness, Color, and Content

Content and Consciousness

Content and Consciousness

The Character of Consciousness (Philosophy of Mind)

Contemporary Debates in Philosophy of Mind

Vision and Mind: Selected Readings in the Philosophy of Perception

Vision and Mind: Selected Readings in the Philosophy of Perception

Vision and Mind: Selected Readings in the Philosophy of Perception

Mind, Reason and Imagination: Selected Essays in Philosophy of Mind and Language (Cambridge Studies in Philosophy)

Mind, Reason and Imagination: Selected Essays in Philosophy of Mind and Language (Cambridge Studies in Philosophy)

Mind, Reason and Imagination: Selected Essays in Philosophy of Mind and Language (Cambridge Studies in Philosophy)

Consciousness: From Perception to Reflection in the History of Philosophy (Studies in the History of Philosophy of Mind, 4)

Consciousness and Mind

Phenomenal Concepts and Phenomenal Knowledge: New Essays on Consciousness and Physicalism (Philosophy of Mind Series)

Phenomenal Concepts and Phenomenal Knowledge: New Essays on Consciousness and Physicalism (Philosophy of Mind Series)

Contemporary Debates in Philosophy of Mind (Contemporary Debates in Philosophy 8)

Sensation and Perception: A History of the Philosophy of Perception

Contemporary Debates in Philosophy of Biology (Contemporary Debates in Philosophy)

Action in Perception (Representation and Mind)

Essays on Nonconceptual Content

Matter and consciousness: a contemporary introduction to the philosophy of mind

Matter and Consciousness, Revised Edition: A Contemporary Introduction to the Philosophy of Mind

Purple Haze: The Puzzle of Consciousness (Philosophy of Mind Series)

Matter and Consciousness: A Contemporary Introduction to the Philosophy of Mind

Purple Haze: The Puzzle of Consciousness (Philosophy of Mind Series)

Purple Haze: The Puzzle of Consciousness (Philosophy of Mind Series)

Neurochemistry of Consciousness: Neurotransmitters in Mind

Recommend Documents