E t h ic a l N at u r a l ism
Ethical naturalism is narrowly construed as the doctrine that there are moral properties...
36 downloads
991 Views
2MB 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
E t h ic a l N at u r a l ism
Ethical naturalism is narrowly construed as the doctrine that there are moral properties and facts, at least some of which are natural properties and facts. Perhaps owing to its having faced, early on, intuitively forceful objections by eliminativists and non-naturalists, ethical naturalism has only recently become a central player in the debates about the status of moral properties and facts which have occupied philosophers over the last century. It has now become a driving force in those debates, one with sufficient resources to challenge not only eliminativism, especially in its various non-cognitivist forms, but also the most sophisticated versions of non-naturalism. This volume brings together twelve new essays which make it clear that, in light of recent developments in analytic philosophy and the social sciences, there are novel grounds for reassessing the doctrines at stake in these debates. s u s a n a n uc c e t e l l i is Professor of Philosophy at St. Cloud State University, Minnesota. She is editor of New Essays on Semantic Externalism and Self-Knowledge (2003) and, with Gary Seay, Philosophy of Language: The Central Topics (2007). She is the author of Latin American Thought: Philosophical Problems and Arguments (2002). g a r y s e a y is Professor of Philosophy at Medgar Evers College, City University New York. With Susana Nuccetelli, he is co-author of How to Think Logically (2007) and Latin American Philosophy (2004), and co-editor of Themes from G. E. Moore: New Essays in Epistemology and Ethics (2007).
E t h ic a l Nat u r a l ism Current Debates E di t e d b y Sus a n a N uc c e t e l l i and G a ry Se ay
c a mbr idge u ni v er sit y pr e ss Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo, Delhi, Tokyo, Mexico City Cambridge University Press The Edinburgh Building, Cambridge cb2 8ru, UK Published in the United States of America by Cambridge University Press, New York www.cambridge.org Information on this title: www.cambridge.org/9780521192422 © Cambridge University Press 2012 This publication is in copyright. Subject to statutory exception and to the provisions of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press. First published 2012 Printed in the United Kingdom at the University Press, Cambridge A catalogue record for this publication is available from the British Library Library of Congress Cataloging in Publication data Ethical naturalism : current debates / [edited by] Susana Nuccetelli, Gary Seay. p. cm. Includes bibliographical references and index. i s b n 978-0-521-19242-2 (hardback) 1. Ethics, Evolutionary. 2. Naturalism. I. Nuccetelli, Susana. II. Seay, Gary. bj 1311.e84 2011 171′.2–dc23 2011037727 isbn 978-0-521-19242-2 Hardback Cambridge University Press has no responsibility for the persistence or accuracy of URLs for external or third-party internet websites referred to in this publication, and does not guarantee that any content on such websites is, or will remain, accurate or appropriate.
Contents
List of contributors
page vii
Introduction
1
1 Naturalism in moral philosophy
8
Gilbert Harman
2 Normativity and reasons: five arguments from Parfit against normative naturalism
24
3 Naturalism: feel the width
58
4 On ethical naturalism and the philosophy of language
70
5 Metaethical pluralism: how both moral naturalism and moral skepticism may be permissible positions
89
6 Moral naturalism and categorical reasons
110
7 Does analytical moral naturalism rest on a mistake?
131
David Copp Roger Crisp
Frank Jackson
Richard Joyce
Terence Cuneo
Susana Nuccetelli and Gary Seay
8 Supervenience and the nature of normativity
144
9 Can normativity be naturalized?
169
10 Ethical non-naturalism and experimental philosophy
194
Michael Ridge Robert Audi
Robert Shaver
v
vi
Contents
11 Externalism, motivation, and moral knowledge
211
12 Naturalism, absolutism, relativism
226
Bibliography Index
245 259
Sergio Tenenbaum Michael Smith
Contributors
Robe r t Au di is O’Brien Professor of Philosophy at the University of Notre Dame. Dav i d C opp is Professor of Philosophy at the University of California at Davis. Ro g e r C r i sp is Uehiro Fellow and tutor in philosophy at St Anne’s College, Oxford, and Visiting Professor at Boston University. T e r e nc e C u n e o is Assistant Professor of Philosophy at the University of Vermont. G i l be r t H a r m a n is James S. McDonnell Distinguished University Professor of Philosophy at Princeton University. F r a n k J ac k s on is Visiting Professor of Philosophy at Princeton University and holds fractional research positions at the Australian National University and La Trobe University. R ic h a r d Joyc e is Professor of Philosophy at Victoria University of Wellington, New Zealand S us a n a N uc c e t e l l i is Professor of Philosophy at St. Cloud State University, Minnesota. M ic h a e l R i d g e is Professor of Moral Philosophy at the University of Edinburgh. G a r y S e a y is Professor of Philosophy at Medgar Evers College of the City University of New York.
vii
viii
List of contributors
Robe r t S h av e r is Professor of Philosophy at the University of Manitoba. M ic h a e l S m i t h is McCosh Professor of Philosophy at Princeton University. S e rg io T e n e n b au m is Professor of Philosophy at the University of Toronto.
Introduction
This collection offers new perspectives on ethical naturalism, narrowly construed as the conjunction of two core theses. One holds that there are moral properties and facts, the other that at least some such properties and facts are natural properties and facts. Thus understood, ethical naturalism is distinct from, though usually motivated by, philosophical naturalism, a more general metaphysical outlook according to which all there is is the world as conceived by science. Clearly, philosophical naturalism does not entail ethical naturalism, for it is compatible also with eliminativist accounts of morality that either reject the ethical naturalist’s core theses altogether (as in the error theory) or deflate them substantially (as in quasi-realism). But while eliminativism, especially in its various non-cognitivist forms, was a driving force through much of the twentieth century, ethical naturalism fell out of favor among philosophical naturalists until near the century’s end, perhaps as a result of having faced, early on, intuitively forceful objections such as G. E. Moore’s 1903 open question argument. In the last thirty years, however, increasing doubts about the cogency of those objections, together with some key developments in philosophy of mind and language, have contributed to a widespread renewal of interest in ethical naturalism. For many philosophical naturalists now, one appeal of ethical naturalism is its core thesis that there are moral properties and facts, especially when read as claiming that such properties and facts are mind- and language-independent. On this, ethical naturalists compete with non-naturalists, who also hold a thesis with a realist gloss, in conjunction with their defining claim that at least some such moral properties and facts are irreducible, non-natural properties and facts. But the latter claim appears to commit non-naturalists to a moral ontology and an epistemology that are at odds with philosophical naturalism. Thus non-naturalism, in spite of its initial influence, has appeared less attractive to naturalistically minded philosophers for whom the very notion of a non-natural property or fact seems 1
2
Introduction
metaphysically extravagant. Moreover, although it is now widely accepted that the moral supervenes on the natural, critics doubt that non-naturalists can explain how irreducible moral properties and facts could supervene on natural properties and facts (see Ridge, this volume). These and other apparently compelling objections to non-naturalism are among the factors that have contributed indirectly to the current attraction of ethical naturalism for philosophers inclined toward moral realism. But the appeal of ethical naturalism is undoubtedly also owing to its apparent ability to accommodate both a general philosophical-naturalist outlook and a representationalist account of moral language. On the one hand, ethical naturalism promises to deliver a non-eliminativist account of morality that might resolve the problem of locating moral value in the world as conceived by modern science. If ethical naturalism is correct, the philosophical naturalist’s puzzle of how to place morality in the natural order simply dissolves. For then, at least some moral properties and facts are supervenient on, and perhaps identical to, natural properties and facts. On the other hand, ethical naturalism promises to dissolve that puzzle without abandoning another attractive thesis in metaethics, representationalism about moral terms and sentences. For realist ethical naturalism can capture the common intuition that at least some moral terms denote legitimate natural properties, and some moral sentences represent how things are morally. This follows from the ethical naturalist’s view that at least some moral sentences have truth conditions of the sort countenanced by a robust moral realist theory. Beyond the two core theses mentioned above, however, ethical naturalists find much to disagree about. Some read those theses with a realist gloss. Others favor a relativist interpretation. Ethical naturalists are also divided on whether moral properties and facts are reducible without normative remainder to purely natural properties and facts. A further disagreement among them concerns whether moral terms and sentences are semantically equivalent to natural terms and sentences. What is sometimes called “analytical naturalism” holds that they are, while “metaphys ical naturalism” maintains that the relevant relationship between the moral and the natural involves properties and facts exclusively. Such controversies are the subject of extended treatment in the present collection. The first set of chapters focuses on epistemic and metaphysical problems thought to arise for a number of ethical naturalist doctrines. Among them is a well-known epistemic challenge to reductive ethical naturalism: namely, that no empirical methods can be invoked to decide among rival ethical theories. This challenge is one of Gilbert
Introduction
3
Harman’s concerns in his contribution. On Harman’s view, although the naturalistic reduction associated with normative functionalism cannot meet what he regards as the main epistemic challenge facing ethical naturalism, the response-dependent and social convention theories have the resources to avoid that challenge. Other concerns in his essay include the prospects of naturalistic approaches current in moral psychology that attempt evolutionary debunking accounts, a possible parallel between morality and language, and the roles (if any) of guilt and character in morality. David Copp’s contribution considers a recent objection to ethical naturalism by Derek Parfit (2011) that is now attracting considerable attention. According to this objection, ethical naturalism is unable to account for the normativity of moral properties and facts. But Copp sees no normativity problem for ethical naturalist doctrines that, like his, are reductionist, non-analytic, and realist. He sets out to substantiate this claim by looking closely at five attempts to raise the normativity problem for ethical naturalists, most of them by Parfit and some by Jonathan Dancy (2006) and by David McNaughton and Piers Rawling (2003). On Copp’s assessment, none of these attempts succeeds in showing that no natural property or fact could also be normative. Roger Crisp’s essay questions the common assumption that all versions of ethical naturalism are incompatible with non-naturalism. Given his argument, at least some forms of ethical naturalism might be consistent with non-naturalism of the sort recently defended by Parfit. This conflicts, of course, with a widely held view of ethical naturalism as being incompatible with non-naturalism. Crisp himself begins his essay by noting that there seems to be an irresolvable disagreement between realist, non-analytic naturalists and their non-naturalist opponents. Their disputes often lead to a dialectical standoff, which Crisp illustrates by considering how ethical naturalists could respond to Parfit’s recent attempt to raise a normativity problem for ethical naturalism. Contra Parfit, there seems to be logical space for naturalists to vindicate their central claim that normative facts and properties are nothing over and above natural facts and properties, a thesis roundly denied by non-naturalists. The main difference between the two parties, Crisp thinks, concerns their goals: naturalists seek to anchor normativity in the natural world, while non-naturalists aim at accounting for the distinctiveness of normative properties (by contrast with those of science). But there is room for a compromise, for if Crisp is right, ethical naturalists and non-naturalists could both embrace a non-reductive, supervenience account of normative properties couched
4
Introduction
in terms of emergentism – which he conceives as amounting to the metaethical analogue of emergentism in the philosophy of mind. Frank Jackson’s contribution addresses what he regards as an old challenge for ethical naturalists who are also cognitivists: can they accommodate both substantial agreement about how moral language represents things to be and also widespread dissension over any attempted identifications of ethical properties with natural properties? To do that, ethical naturalists must draw on a plausible semantics for moral terms, one that can account for their informative role among competent users of moral language. To Jackson, although a currently popular, externalist semantic theory fails to meet this condition, his own “network account” satisfies it. Given the network account, ethical terms/concepts form an interlocking system about whose informative role there is substantial agreement among competent users, even though the network itself is in part under negotiation. The possibility of such an agreement is consistent with there being widespread dissension about the identification of moral properties with natural properties. Richard Joyce’s essay addresses a different sort of issue that might be a problem not only for ethical naturalism but also for moral skepticism (i.e., the error theory and non-cognitivism): namely, that these apparently contrary accounts are based exclusively on conceptual reasons that might be equally indeterminate. For there might be no fact of the matter as to which of these apparently rival accounts is correct. That is, if Joyce is right, such apparently contrary accounts might both be affected by indeterminacy of the sort claimed by Quine in the case of theories of meaning. To support a radical claim along these lines, Joyce draws on early work by David Lewis, together with some evidence stemming from the ambiguity of notions, such as “assertion,” commonly invoked in the dispute between ethical naturalists and moral skeptics. To make matters worse, no pragmatic reasons seem available for any attempt to resolve the indeterminacy problem facing ethical naturalism and moral skepticism. To say that moral naturalism and moral skepticism might both be affected by Quinean indeterminacy commits Joyce to a kind of metaethical pluralism. But elsewhere Joyce (2001, 2006) has offered reasons for preferring the error theory over rival views, including ethical naturalism. Terence Cuneo’s essay takes issue with one of Joyce’s arguments for that conclusion, the so-called categoricity argument. On Cuneo’s view, this argument suffers from an “arbitrariness problem,” since it arbitrarily counts certain features of ordinary moral practices while discounting others. In addition, if Cuneo is right, Joyce’s defense of the error theory
Introduction
5
faces another problem: moral naturalism seems to square better than the error theory with Joyce’s own standards for the acceptability of a moral theory. Even if, as Cuneo contends, ethical naturalism can meet the challenge raised by Joyce’s categoricity argument, it may still need to respond to other objections before it can get its two core theses off the ground. Prominent among them is G. E. Moore’s “open question argument,” which he famously offered together with the “naturalistic fallacy” charge. Although there is consensus that this extended inference fails to undermine all varieties of moral naturalism, the open question argument is often vindicated as having some intuitive force against analytical moral naturalism. By contrast, the charge that analytical naturalism commits the naturalistic fallacy usually finds no takers at all. In their essay here, Nuccetelli and Seay revisit each of these Moorean arguments with an eye to showing that analytical naturalism of the sort recently proposed by Frank Jackson (1998, 2003) and Michael Smith (2000) does after all rest on a mistake – though perhaps not the one Moore had in mind when he made the naturalistic fallacy charge. The non-naturalist opponents of ethical naturalism, of course, face problems of their own, not least of which is their seeming inability to account for the supervenience of the moral on the natural, a widely accepted relation sometimes invoked by the slogan, “Necessarily, no normative difference without descriptive difference.” In his contribution to this volume, Michael Ridge reconstructs the supervenience objection against non-naturalism. Standardly construed, the objection points to the non-naturalist’s apparent inability to explain how irreducibly non-natural properties and facts could supervene on entirely natural properties and facts. To Ridge, the objection can be sharpened so that it covers also the non-naturalist’s apparent inability to explain why there should be any such irreducible non-natural properties at all. Although a recent non-naturalist account by Ralph Wedgwood (2007) might be beyond the reach of the supervenience objection standardly construed, on Ridge’s view it does not escape it when sharpened in the way proposed in his contribution to this volume. Another problem for non-naturalism that arose early on, at least for Mooreans, is that the doctrine appears incompatible with a plausible moral epistemology. But that wouldn’t be so if a perception-based epistemology for moral properties and facts, of the sort outlined by Robert Audi in his essay included here, could get off the ground. For Audi’s project amounts to a naturalistic epistemology for moral properties and
6
Introduction
facts that seems available to non-naturalists. One building block of Audi’s project is the claim that at least some judgments ascribing moral properties are epistemically grounded in a kind of perception, though not of a representational sort. If so, such perceptions afford a type of perceptual knowledge, and this is the “naturalistic anchor” which is available not only to ethical naturalism but also to “non-reductive realism.” Audi’s non-reductive realism is a “consequentiality” doctrine holding that there are irreducible moral properties that are consequential upon natural properties. Thus construed, the thesis is consistent with the nonreductive realist view of classical non-naturalists such as Moore (e.g., in his “Conception of Intrinsic Value” [1922a]). If Audi’s proposal is found compelling, then non-naturalism, cast as non-reductive realism, might after all avoid the epistemic version of the “queerness” objection often taken to undermine it. Yet recent work in experimental philosophy and some branches of empirical psychology might undermine the epistemology of non-naturalism by pointing to its extreme dependence on unreliable methods based on thought experiment and intuition. Robert Shaver explores some consequences of this work for non-naturalism. His paper looks closely at whether the argumentation strategy of non-naturalists could succeed in supporting their views, given that the strategy is often heavily dependent on thought experiments, as charged by experimentalists. He also considers the empirical strategies of experimental philosophers. Close examination of the strategies used by each of these parties appears to show that there is logical space for skepticism about any across-the-board advantage to be found in the experimentalist strategies over the a priori strategies of non-naturalists. But Shaver’s paper invokes some recent results of empirical tests that appear to undermine one of the two types of a priori argument preferred by non-naturalists, the so-called wrong-reasons argument. Sergio Tenenbaum’s contribution asks whether certain varieties of realist moral naturalism are compatible with an externalist, Humean theory of motivation. Given Michael Smith’s 1994 “fetishism objection,” argues Tenenbaum, they are not. For virtuous agents must have non-derivative motivations to pursue specific ends they believe to be morally right, and externalist theory ascribes to the virtuous agent only a direct de dicto desire to do what is morally right. After reconstructing Smith’s objection, Tenenbaum contends that there is an understanding of virtuous motivation, available to realist moral naturalists, that is immune to Smith’s objection.
Introduction
7
In his own essay for the volume, Michael Smith challenges Gilbert Harman’s (2000a) contention that moral relativism is favored by philosophical naturalism over its competitor, “moral absolutism.” On Smith’s view, not only is naturalism silent about whether moral relativism or absolutism is right, but Harman has failed to identify the real source of disagreement between these doctrines. As reconstructed by Smith, moral absolutism is a version of moral rationalism, a set of doctrines attractive to many current theorists inspired either by Kant or by Brentano and Ewing. To Smith, Harman’s argument appears sound only if we assume certain principles that supposedly govern the formation of an agent’s intentions. But there are rival assumptions equally compatible with naturalism that may be available to moral absolutists. Once those assumptions are taken into account, the disagreement between absolutists and relativists (and among the absolutists themselves) can be seen to turn not on naturalism but instead on whether it is the relativist characterization of the functional roles of beliefs and desires that is the correct one or that offered by the absolutist. Thus, if Smith’s response to Harman is on the right track, Harman’s argument for the claim that naturalism favors moral relativism would be unsound, for it would rest after all on a claim in need of support: namely, a certain disputed assumption about the connection between moral demands and sufficient reasons.
ch apter 1
Naturalism in moral philosophy Gilbert Harman
1.1 I n t roduc t ion 1.1.1 Narrow and wide conceptions of philosophy and philosophical method Naturalism in philosophy is a special case of a more general conception of philosophy. In this conception there is no special philosophical method and no special philosophical subject matter. Consider some of the ways in which philosophy interacts with and is continuous with other disciplines. Aesthetics is obviously pursued in philosophy departments and in departments of literature, music, and art. Monroe Beardsley, who wrote the most important survey of aesthetics in the twentieth century, was one of the authors of an important statement of a central aspect of the “New Criticism.” More recently, Richard Wollheim (who may have invented the expression “minimal art”) and Arthur Danto have had a significant influence on art theory and criticism. They themselves have been important critics. Alexander Nehamas is another important contemporary example. Anthropology. Anthropologists are often involved with philosophy and philosophers have sometimes acted as anthropologists to study the moralities of one or another culture. Richard Brandt lived with the Hopi in order to study their ethics. John Ladd lived with the Navaho in order to study their ethics. The anthropologist Dan Sperber is the same person as the philosopher Dan Sperber.1 Economics. Recent figures include Robert Nozick, Amartya Sen, maybe John Rawls, David Gauthier, Allan Gibbard, John Broome, Philip Pettit,
For example, Brandt (1954); Ladd (1957); Sperber (1973); and Sperber and Wilson (1986).
1
8
Naturalism in moral philosophy
9
and many more. Political theory is of course a related example with many of the same players. Linguistics is another very clear case. Philosophers were involved early in the development of generative grammar (e.g., Jerry Katz and Jerry Fodor). Many more wrote about Chomsky’s ideas and argued with them (e.g., Paul Ziff, Hilary Putnam). Famously, at the end of the first chapter of A Theory of Justice, John Rawls suggested that generative grammar might be a good model for moral theory.2 Earlier Robert Nozick tried to sketch how that might work.3 John Mikhail has been developing this idea in some detail. In recent years there has been philosophical interest in and interaction with developments in linguistics. And there has been much interdisciplinary research in semantics involving philosophers and linguists. Psychology is another clear case. In his Theory of Justice Rawls suggested that an adequate moral theory had to be sensitive to developmental psychology, especially in Piaget. Rawls’ early work on justice in turn influenced the psychologist Lawrence Kohlberg’s (1981, 1984) adaptation of Piaget. Donald Davidson more or less regularly discussed rationality with psychologists like Amos Tversky and Daniel Kahneman, trying to get them to accept that there were limits on how irrational people could be interpreted to be. J. L. Austin’s (1956–57) study of excuses was influential on psychological studies of children’s development by John Darley and his colleagues. In recent years there has been considerable back and forth between psychologists and philosophers on many issues. Relevant philosophers include Daniel Dennett, Stephen Stich, and many younger people working in the general area of (real) moral psychology.4 One important issue has concerned whether social psychology undermines ordinary conceptions of character traits and threatens certain forms of virtue ethics. But there are many other issues too. Computer science. Artificial intelligence, machine learning, and related topics have been considered highly relevant to philosophy of mind. For example, the philosopher John Pollock (1995) studied epistemology by designing computer programs to simulate reasoning in accord with one or another set of epistemic principles. Rawls (1971: section 9). 3 Nozick (1968). Doris (2010); Sinnott-Armstrong (2008a).
2
4
10
g i l be r t h a r m a n
Philosophy of science is another obvious example. Philosophers discussing the interpretation of quantum field theory may publish in physics journals (for example, my colleague Hans Halvorson). I went into philosophy because it allowed me to pursue interests in linguistics, artificial intelligence, and cognitive science. My earliest publication was in linguistics.5 Soon after that Donald Davidson and I organized workshops that brought linguists and philosophers together.6 Later the psychologist George Miller and I started the Princeton University Cognitive Science Laboratory and an undergraduate program in Cognitive Studies. More recently, I have co-taught courses with faculty in linguistics, psychology, computer science, and engineering. Most of my colleagues at Princeton take a wide view of philosophy in one or another respect. 1.1.2 Naturalism Philosophical naturalism is a special instance of the wider conception of philosophy, taking the subject matter and methods of philosophy to be continuous with the subject matters and methods of other disciplines, especially including the natural sciences. From a naturalistic perspective, productive philosophers are those who (among other things) produce fruitful more or less speculative theoretical ideas, with no sharp distinction between such theorizing by members of philosophy departments and such theorizing by members of other departments. (In my view, department boundaries are of interest only to administrators.) Naturalism also often has an ontological or metaphysical aspect in supposing that the world is the natural world, the world that is studied by the natural sciences, the world that is available to methodological naturalism. But the main naturalistic theme is methodological. In what follows, I discuss certain prospects for naturalism in moral philosophy. I begin with metaphysical issues of the sort just mentioned, having to do with naturalistic reduction in ethics. I then say something about a few recent naturalistic methodological approaches in moral psychology. 1.2 N at u r a l i s t ic r e duc t ion Naturalistic reduction in ethics attempts to locate the place of value in a world of (naturalistically conceived) facts. Harman (1963).
5
6
See Davidson and Harman (1972).
Naturalism in moral philosophy
11
In one view, goodness and evil, and rightness and wrongness are not features that have a place in the naturalistic world as described by science. Naturalists who take this view either abandon ethics altogether or try to provide a nonfactual account of it. Alternatively, naturalists might try to identify an act’s being morally right and wrong, good and bad, just and unjust, etc., with certain natural properties of the act. The most straightforward naturalistic reductive strategy appeals to the supervenience of the moral on the natural facts. Any change in what the agent ought morally to do requires a change in the (natural) facts of the case. This appears to imply that there is a more or less complex natural relation between an agent, a possible act, and the agent’s situation (conceived as a whole possible world) that holds when and only when the agent in that situation is morally permitted to do that act. The idea then is to identify the property of being what an agent is morally permitted to do in a given situation with the property of being a possible act for which this natural relation holds. For example, suppose that act utilitarianism provided the correct account of what an agent is morally permitted to do. Given that supposition, the supervenience strategy identifies a possible act’s being what an agent is morally permitted to do in a given situation with its being an act that maximizes utility in that situation. More generally, the strategy identifies a possible act’s being what an agent is morally permitted to do in a given situation with the holding of the relevant natural relation, whatever it is, which exists between agent, act, and situation if and only if the agent is morally permitted to do that act in that situation. It is not a good objection that such an identification fails to capture the meaning of “morally permitted.” To suppose that water can be identified with H2O is not to say what the word “water” means as used by ordinary people. It is true that the moral case raises a methodological issue for naturalism, since different moral theories disagree with each other. There are competing versions of utilitarianism, social contract theory, virtue theory, Kantianism, and many others. Is there a naturalistically acceptable way to resolve these disputes by testing them against the world as competing scientific theories can be tested? Instead of trying to answer this question directly, let us consider three kinds of naturalistic reduction, associated with theories of normative functionalism, response-dependent theories, and social convention theories.
12
g i l be r t h a r m a n 1.2.1 Normative functionalism and virtue ethics
One kind of virtue ethics7 appeals to a normative functionalism that seeks to derive normative results from assumptions about functions – about designed or natural functions, purposes, roles, etc. For example, the most important function of a clock is to keep time. Whether something is a clock depends on its function, not on what it is made of or what it looks like, as long as it can serve to indicate to an observer what the time is. Furthermore, a clock can be evaluated in terms of its function. So, a good clock is one that keeps time accurately. That’s what a clock ought to do. If it does not do so, something is wrong with it. It is defective. The features of a good clock that contribute to its accurate functioning are virtues of the clock. Bodily organs are also defined by their proper functions. A heart is an organ whose nature or function is to pump blood steadily. Lungs are organs that function in breathing. Whether something counts as a heart or lung is not a matter of its shape or what it is made of, but whether it has the relevant function. One that actually does so is to that extent a good heart or lung. A heart that fulfills its function poorly, by irregular pumping, or by leaking blood, is a bad heart and something is wrong with it. The virtues of a heart include steady pumping and not leaking. People who have social roles have associated functions or purposes. A good teacher is one who teaches well, who enables students to learn. Something is wrong with a teacher whose students do not learn. Virtues in a teacher are those characteristics that enable the teacher’s students to learn as well as they can. A teacher who cannot get students to learn is not a good example of a teacher, not a real teacher. It is in the nature of human beings and certain other animate beings (bees and chimpanzees, for example) that they are social beings. A good human being has various virtues, like courage and compassion. Lack of courage or compassion is a defect. A man lacking courage is not a good example of a man, not a real man. Various issues arise for views that attempt to derive moral assessments from functionalism. Do human beings have functions or purposes as part of their nature as human beings? Is the relevant function or purpose to lead a good life, or even the best life? Can this function or purpose be characterized naturalistically? Most important, from a naturalistic perspective, is Foot (2001); Hursthouse (1999); Thomson (2008a).
7
Naturalism in moral philosophy
13
there a way of testing competing views of the best life these views against the world in the way that scientific hypotheses can be tested? 1.2.2 Response-dependent theories and social convention theories Another rather different naturalistic approach identifies moral categories in terms of something about human responses to the consideration of possibilities, in the way in which colors are sometimes identified in terms of something about the responses of normal human perceivers. In this approach, an act’s being wrong might be identified with the dispositions of impartial unbiased sympathetic people to feel moral disapproval of the act on being made vividly aware of the facts of the situation. David Hume and Adam Smith defend different versions based on different interpretations of sympathy. Hume has a tuning-fork account of sympathy: Humean sympathy leads someone to vibrate in tune with others and feel similarly (if less intensely) what others are perceived to be feeling. This yields a utilitarian result. Since people would rather be happy than unhappy, they will favor situations in which there is more net happiness. Smith objects that Hume’s conception of sympathy cannot account for the fact that unhappy people crave sympathy and feel better when they receive it. Humean sympathetic vibrations would make an acquaintance of an unhappy person sympathetically unhappy and then the unhappy person would vibrate with the acquaintance’s unhappiness, making the originally unhappy person even more unhappy. Since the sympathy of an acquaintance makes an unhappy person less unhappy, Hume is wrong about what sympathy is. Smith observes that ordinary sympathy involves approval. If someone gets a minor bump and moans and complains, observers who are aware of the minor pains involved will not sympathize, because they will not approve of the complainer’s reactions. According to Smith, people want sympathy because they want approval. Furthermore, in Smith’s view, the relevant sort of approval tends to be an internalized reflection of community standards. My desire for the approval of others leads me to imagine how they will react to me. I imagine being one of them to consider how I would react, in this way internalizing their standards. This yields a different view of morality from Hume’s – one in which what counts as right or wrong is more heavily influenced by the conventional practices of one’s
14
g i l be r t h a r m a n
society. Smith’s theory, while response-dependent, sees morality as more of a matter of social convention than Hume’s does. It is true that Hume takes social convention to be important for those aspects of morality having to do with justice: people are disposed to approve of those conventions that promote the general welfare. But for Smith social conventions affect approval and disapproval more directly.8 1.2.3 Worries about relativism I think that the most promising naturalistic reductions have relativistic implications. Adam Smith’s is explicitly relativistic, because what captures one’s sympathy is directly affected by local customs. The point generalizes to other response-dependent theories to the extent that the relevant response, usually approval, is directly influenced by varying customs or personal values. And functionalist theories may have to suppose that moral conclusions are relative to one or another competing conception of the best life, the purpose of life, moral defects, etc. Any absolutist (non-relativist) reduction of morality faces the epistemological problem of showing how that conception of morality is better supported than its competitors. The problem is that there are competing moral frameworks and no obvious way to test them against the world. Compare the dispute between Cardinal Bellarmine and Galileo about whether the earth is at rest. Their dispute assumed that there is such a thing as being absolutely at rest. The correct resolution of their dispute is that motion is always relative. This conclusion is grounded in the fact that there is no empirical difference between competing views about what is at absolute rest. Similarly, there appears to be no empirical difference that would resolve fundamental moral differences, which suggests that, from a naturalistic perspective, there may be no reason to believe in absolute right and wrong (Harman 1996). Some might respond that to believe in moral relativism would be to accept moral nihilism, at least if one’s initial conception of moral values is absolutist and not relativistic. But that would be like saying that to believe in the relativity of motion would be to give up on the idea that things move or are at rest! Consider people who were brought up to believe that, when they said that something was wrong, they intended to be saying that it violated I say more about this difference between Hume and Smith in Harman (2000b).
8
Naturalism in moral philosophy
15
God’s commands, but who have now become atheists. They may have at one time firmly believed that, if God had not existed, nothing would have been morally prohibited. But now, having come to doubt that God exists, they continue to accept the same moral principles as before (at least as regards nonreligious matters) and instead have stopped believing that morality is the expression of God’s will. In the same way, someone who is initially committed to moral absolutism who later decides that moral relativism is true can still accept (most of) the same moral principles as before. 1.2.4 Naturalism as a response to evolutionary debunking Some recent attempts to debunk nonutilitarian moral intuitions appeal to possible explanations of the intuitions in terms of evolution by natural selection. The claim is that the relevant moral intuitions result from factors having nothing to do with their truth, namely tendencies to develop whatever intuitions might help to get one’s genes into following generations. While some theorists have argued that such evolutionary explanations debunk intuitions that conflict with utilitarianism, leaving utilitarianism unchallenged as the correct normative view (Greene 2008a; Singer 2005), others say that if the explanations debunk nonutilitarian intuitions, they also debunk intuitions that appear to support utilitarianism (Kahane, 2011). Naturalism offers a response to these debunking arguments. Compare a naturalistic response to a corresponding attempt at evolutionary debunking of color perception. The response is to identify colors as responsedependent properties, determined by how we perceive them. If that is right, an evolutionary account of how we happen to have the color experience is not a debunking account. Similarly if what is right or wrong is response-dependent in one or another of the ways already considered, an evolutionary account of moral intuitions is not a debunking account (Street 2006). This is all I am going to say about a possible naturalistic reduction of morality. 1.3 Mor a l p s yc hol o g y I now want to consider how certain issues in moral psychology look from a naturalistic point of view.
16
g i l be r t h a r m a n
Philosophers have at times treated philosophical moral psychology as an armchair subject based on a priori reflection on meanings and concepts.9 A naturalistic approach rejects that methodology and takes moral psychology to be an empirical subject that is continuous with scientific psychology, requiring a more intersubjective approach (Doris 2010; Doris and Stich 2008; Sinnott-Armstrong 2008a). I am going to say something (briefly) about three rather different topics in moral psychology: linguistics as a guide to moral theory, guilt feelings, and character traits. 1.3.1 Linguistic analogy Despite great superficial differences between languages, linguistics aims at a theory of universal grammar. Faced with apparently great differences in moralities, perhaps moral theory can similarly develop a universal moral grammar. Generative grammars The primary object of study of contemporary linguistics is not language in the ordinary sense in which English, German, Mohawk, and Chinese are languages. Any ordinary language in this sense has different dialects that blend into one another in ways that do not correspond to national or geographical boundaries. There is a well-known saying that a language (in the ordinary sense) is a dialect with an army and a navy. What counts as a dialect of French rather than Flemish is a social or political issue, not an issue in the science of language. It may be that any two speakers have at least somewhat different dialects, with at least somewhat different vocabularies. Chomsky and other linguists are concerned with language as a property of a particular person, assumed to be abstractly specifiable by an internal grammar, or I-grammar. Indeed, individuals are assumed to have more than one I-grammar if they are bilingual. The analogy between language and morality suggests that the primary object of study for moral theorists, then, might not be the morality of a group but rather a property of a particular person, abstractly specified by something like an internal “moral grammar” or I-morality. A “generative grammar” of a particular person’s language would specify the linguistic structure of expressions of that language, indicating the nouns, verbs, adjectives, prepositions, etc., as well as the phrases of which For example, Chan (2008).
9
Naturalism in moral philosophy
17
these are the “heads,” as when a noun is the head of a noun phrase and a preposition the head of a prepositional phrase. A generative grammar might specify important aspects of the meanings of expressions, depending on how the expressions are structured and possibly indicating the scope of quantifiers and other such operators. The grammar would also specify aspects of the pronunciation of expressions and would relate pronunciations of expressions to possible interpretations, in this way indicating certain sound–meaning (phonetic–semantic) relationships. An analogous generative moral grammar would attempt explicitly to characterize an individual’s moral standards. Just as an I-grammar specifies the structure of a well-formed linguistic sentence by using a specialized linguistic vocabulary, an I-morality might specify the structure of permissible and impermissible actions, using a specialized vocabulary. A generative moral grammar might be an action grammar (or a situation grammar, etc., depending on what sorts of things can be morally assessed). The rules of a generative grammar have a recursive character. These rules imply that a larger sentence can contain a smaller sentence within it, which can contain another even smaller sentence, and so on without limit. Analogously, an act may contain a smaller act within it. A generative I-morality might imply that it is wrong to encourage someone to do something that is wrong, a principle that is recursive in the sense that it implies that it is wrong to encourage someone to encourage someone to do something that is wrong, etc. Similarly, the I-morality might imply that it is wrong to promise to do something that it would be wrong to do. There is evidence that whether people accept certain seemingly descriptive claims about whether an agent does something intentionally can depend on whether they accept certain normative claims about side-effects of actions. And whether people judge that a particular person caused a certain result can depend on whether they judge that the person is morally at fault for doing what he or she did. For example, whether one judges that Tom’s omitting to water the plants caused their death can depend in part on whether one thinks Tom was obligated to water the plants. That suggests that relevant structural descriptions of the situations or actions may themselves sometimes involve normative assessments. In developing a theory of linguistic grammar, Chomsky (1972) famously distinguishes what he calls “competence” from what he calls “performance”. By “competence” Chomsky means an individual’s internalized grammar. He uses the term “performance” to refer to all other factors that
18
g i l be r t h a r m a n
may affect the use of language. In Chomsky’s framework an individual’s linguistic intuitions may or may not reflect the individual’s competence, in Chomsky’s sense. In moral theory as in linguistics, we can consider whether it makes sense to postulate a distinction between an individual’s internal morality or moral “competence” in contrast with various other factors that determine the person’s intuitions and actions. For linguistics, the corresponding assumption has been theoretically fruitful and illuminating. With respect to moral theory, the assumption amounts to assuming that there is a distinctive moral competence or I-morality. Universals A child picks up language from its interactions with others; there is no need for explicit teaching, even though basic linguistic principles have only been formulated recently and cannot have been explicitly taught to the child, who therefore seems innately predisposed to acquire an I-grammar satisfying such principles. Similarly, a number of moral philosophers have suggested that ordinary moral intuitions obey certain non-obvious rules. For instance, it may be that moral intuitions reflect some version of a Principle of Double Effect. Double Effect: it is worse knowingly to harm one person X in saving another Y if (a) the harm to X is intended as part of the means to saving Y than (b) if the harm to X is merely a foreseen unintended side effect of the attempt to save Y.
For another example, Thomson (1976)10 suggests that some ordinary moral intuitions might reflect a principle of the following sort: Deflection: it is better to save a person X by deflecting a harmful process onto another person Y than by originating a process that harms Y.
These principles are not generally recognized, and it is unlikely that they are explicitly taught to children. If children do acquire moralities containing such principles, it would seem that they must be somehow predisposed to do so. If so, we might expect such principles to be found in all natural moralities that children naturally acquire. On the other hand, moralities differ about abortion and infanticide, about euthanasia, about slavery, about the moral status of women, about the importance of chastity in women and in men, about caste systems, about cannibalism, about eating meat, about how many wives a man can have at the same time, about the relative importance of equality versus 10
But see her later article (Thomson 2008b).
Naturalism in moral philosophy
19
liberty, the individual versus the group, about the extent to which one is morally required to help others, about duties to those outside one or another protected group, about the importance of religion (and which religion is important), about the importance of etiquette (and which standards of etiquette), about the relative importance of personal virtues, and so on. Linguists have evidence that the child uses its limited experience to set a relatively small number of parameters and acquire the “core” syntax of a language. Other “non-core” aspects of syntax, perhaps involving stylistic variation, are learned as exceptions (Baker 2001). The linguistic analogy suggests considering whether it is possible to develop a theory of morality involving principles and parameters. For example, perhaps, all moralities accept principles of the form: avoid harm to members of group G, share your resources with members of group F, etc., where G and F are parameters that vary from one morality to another. However, we will need to have more explicit generative accounts of a variety of moralities before being able to think about moral principles and parameters. I have described various issues suggested by the analogy between morality and linguistics. I discussed what might be involved in a generative “moral grammar” and have discussed what might be involved in a universal moral grammar that parallels universal linguistic grammar. I believe that the issues suggested by such analogies are worth pursuing.11 1.3.2 Guilt I believe that moral theorists have made a number of related mistakes about guilt – defining morality in terms of nontrivial guilt, identifying having a moral conscience with susceptibility to nontrivial guilt, and holding that one cannot be a moral person if one lacks a susceptibility to nontrivial guilt. (I suspect that these ideas arise from theorists overgeneralizing from their own experience, failing to realize that many quite moral people do not experience the sort of nontrivial guilt required for these claims.) What is the required (nontrivial) emotion of guilt? It must at least include the thought or impression that one is guilty of having done something wrong. Feelings of guilt trivially involve this. But a cold-hearted For more discussion see Roedder and Harman (2010) and references given there.
11
20
g i l be r t h a r m a n
psychopath might have such thoughts without experiencing nontrivial feelings of guilt. More is needed. At least (agent) regret. But a feeling of agent regret is not enough either. One can regret doing things without feeling guilty in any nontrivial sense. In particular, a cold-hearted psychopath might regret having done something morally wrong without suffering from nontrivial feelings of guilt. Typically, the theorists I am talking about suppose that nontrivial guilt feelings involve more or less agonized self-punishment or deep shame. (Deep shame, of course, is more general than guilt in that one can feel shame for one’s appearance, accent, parents, or president.) Some nontrivial feelings of guilt and/or shame may derive from internalized embarrassment at being seen by others to have done something wrong or for something shameful. Nontrivial guilt feelings might also be or instead be internalizations of external resentment at others’ wrongdoing, internalized blame, and outrage. It seems to me that we should reject all of the following proposed connections between morality and such nontrivial guilt feelings. (1) Wrong acts warrant such nontrivial guilt feelings. (2) Morality can be defined in part in terms of such nontrivial guilt feelings. (3) It is abnormal not to be susceptible to such nontrivial guilt feelings for wrongdoing. (4) Susceptibility to such nontrivial guilt feelings for wrongdoing is necessary for moral motivation. Here are two reasons to doubt that such nontrivial guilt feelings are central to morality. First, and most obviously, it is easy to imagine an otherwise moral agent not susceptible to such nontrivial guilt feelings. Second, some more or less famous moral paragons are not obviously susceptible, for example, Nietzsche’s overman and Aristotle’s fully virtuous person. In addition, I believe that I am not susceptible to nontrivial guilt feelings. I think I am reasonably moral, and many others have told me that they also do not seem to be susceptible to nontrivial guilt feelings. It is sometimes thought that there are good empirical psychological reasons to think nontrivial guilt feelings are central to morality, based on studies of psychopaths, narcissists, and children with certain brain deficits. R. D. Hare’s (1993) standard criteria for psychopathology include lack of guilt for wrongdoing. But to conclude from these studies that nontrivial guilt feelings are central to morality is to confuse “All As are Bs” with “All Bs are As.” There are many moral people who are not
Naturalism in moral philosophy
21
susceptible to guilt and are not psychopathic, narcissistic, or brain-damaged. And psychopaths (for example) not only lack nontrivial guilt feelings but also (and more importantly) lack empathy and other sources of moral motivation. It is also sometimes thought that the anticipation of nontrivial guilt is needed as a way of indicating that a given act would be wrong. But it is enough to have the belief or impression that the act would be wrong with no need for any anticipation of nontrivial guilt feelings. It may also be that some people are motivated to do the right thing in order to avoid nontrivial guilt feelings. But not everyone needs this motivation. It can be enough that one cares about others and about doing the right thing. It is true that the expression of nontrivial guilt feelings can have the social function of appeasing people whom one has wronged. But a moral person not susceptible to nontrivial feelings of guilt who does something wrong can at least honestly express regret and sympathy. These points imply that it is a mistake to define morality in terms of nontrivial guilt feelings. It is a mistake to define moral standards as those standards violation of which warrants nontrivial guilt feelings or to define one’s individual moral standards as those for which one is susceptible to guilt for violating. My diagnosis of these common mistakes is that they arise from introspective moral psychology treated as a priori insight by people subject to guilt. To summarize my own views about guilt. First, morality does not require a susceptibility to nontrivial feelings of guilt. Second, I am inclined to think it is a defect in an otherwise moral person to be susceptible to nontrivial guilt feelings. Third, I think is possible to be brought up so as not to be susceptible to nontrivial guilt feelings. And fourth, I think if one is susceptible to such feelings, it is possible to get oneself not to be susceptible. Where does the philosophical emphasis on guilt come from? I suggest it comes from the method of introspection. Some philosophers find a connection with their moral views and guilt feelings, a connection they imagine holds for everyone. But they do not actually undertake an empirical study of the extent to which that speculation is true.12
12
For further discussion see Harman (2009) and C. Maley, “Against Guilt-Free Morality,” unpublished MS.
22
g i l be r t h a r m a n 1.3.3 Character
Finally, let me say something about character traits. People are very hasty in attributing broad and robust character traits to others, but it seems that such traits are very rare, if they exist at all. This may cast doubt on versions of virtue ethics that treat virtues as robust character traits. Social psychologists have discovered that ordinary people have a bias toward explaining behavior by appeal to robust character or personality traits while overlooking important situational factors. This bias seems to be associated with a perceptual tendency to pay more attention to a figure than to its ground, and there appear to be significant cultural differences in the extent to which people are subject to this tendency. Having once attributed a trait to a given person, an observer has a strong tendency to continue to attribute that trait to the person even in the face of considerable disconfirming evidence, a tendency psychologists sometimes call “confirmation bias,” a bias toward noting evidence that is in accord with one’s hypothesis and disregarding evidence against it (Ross and Nisbett 1991). Even in a world with no individual differences in robust character traits or personality traits, people would still strongly believe that there were such differences. This means that the apparent obviousness of the claim that people differ in such traits is less evidential than one might think. True, it may seem “obvious” that some people have different character and personality traits from others. But our finding this fact so obvious is predicted by our tendency to a “fundamental attribution error” whether or not there are such differences. Sometimes a person acts well or badly in a seemingly unusual way. Concerning any such case, there is an issue as to what makes the difference that leads to such seemingly unusual behavior. When you perceive or learn about someone you do not know doing such an unusual thing, you have a strong tendency to attribute the behavior to some good or bad trait of the person in question. For example, when you learn that a certain seminary student walked right past someone who seemed to be having a heart attack, actually stepping right over the person, you tend to think of the student as incredibly callous, but the student may have just been in a hurry. In the Milgram experiment, subjects were led by gradual steps to administer what they thought were very severe electrical shocks to another person. The gradualness of the process with no obvious place to stop seems an important part of the explanation of why they obeyed a command to shock the other person in that experiment although they would not have
Naturalism in moral philosophy
23
done so if directly ordered to give the severe shock at the very beginning (Milgram 1974). What these and many related experiments show is that aspects of a particular situation can be important to how a person acts in ways that ordinary people do not normally appreciate, leading them to attribute distinctive actions to an agent’s distinctive character rather than to subtle aspects of the situation. Observers are strongly inclined to blame those participants who did not stop to help or who provided intense shocks, thinking that the explanation of these agents’ immoral actions lies in their terrible character. But the observers are wrong: that cannot be the explanation. The evidence indicates that people may differ in certain relatively narrow traits but do not seem to have broad and stable dispositions corresponding to the sorts of character and personality traits we normally suppose that people have (Doris 2002; Kunda 1999). We must distinguish individual acts of honesty or dishonesty, courage or cowardice, compassion or coldness from the corresponding character traits. The ordinary conception of a character or personality trait is of a relatively broad-based disposition to respond in the relevant way with acts of the corresponding sort. Aristotelian-style virtue ethics shares with folk psychology a commitment to broad-based character traits of a sort that people do not seem to possess. Why have so many philosophers taken character-based virtue ethics seriously? Relying on the method of introspective intuition, it seems obvious that people have the relevant sorts of character traits. But social psychology casts doubt on the existence of such traits and explains why philosophers think they do. 1. 4 F i n a l r e m a r k s To sum up, I have tried to say something about contemporary naturalistic approaches to morality. I started by describing more traditional issues of naturalistic reduction. I then discussed three current issues in naturalistic moral psychology, having to do with the possibility of a linguistic analogy for moral theory, with the unimportance to moral theory of guilt feelings, and with the nonexistence of character traits.
ch apter 2
Normativity and reasons: five arguments from Parfit against normative naturalism David Copp
Moral naturalists and non-naturalists are moral realists. They agree that there are moral properties. They presumably would agree, for example, that actions can have the property of being morally right or of being wrong, that social institutions can have the property of being just or of being unjust, and so on. Naturalists and non-naturalists agree as well that there are moral facts. They should agree, for instance, that torture is morally wrong, and if so, they should agree that it is a fact that torture is wrong. This paper is about the debate between naturalists and non-naturalists, so I set aside anti-realist views, which naturalists and non-naturalists agree in rejecting.1 Naturalists and non-naturalists disagree about the nature of moral properties and facts. Naturalists think, roughly, that moral properties and facts are similar in all metaphysically and epistemologically important respects to (other) natural properties and facts, such as meteorological ones. Nonnaturalists deny this, primarily, I believe, because they think that naturalism cannot account for the normativity of moral properties and facts. Of course, a naturalist might deny that moral properties and facts are normative. To make good on such a denial, she might contend that there is no such thing as normativity. But I do not recommend this approach, I thank Adam True Bendorf, Jonathan Dancy, David McNaughton, Susana Nuccetelli, Derek Parfit, Gary Seay, and Michael Smith for helpful comments on earlier versions of this essay. I also thank the members of DaGERS, the Davis Group in Ethics and Related Subjects, with whom I discussed the ideas in this paper. I presented a version of this paper to the Philosophy Department at Cornell University, and I thank those who participated in the discussion for their helpful comments. 1 For my purposes here, I take moral realism to be the view that there are true moral propositions, or moral facts, that ascribe moral properties to things. For a more articulated account, designed to deal with various complications, see Copp (2006b: 7–8). For present purposes, I take it that a fact is a true proposition. I use the term “property” to refer to the contribution made by a predicate, whatever that contribution is, to the truth conditions of the propositions we assert in making assertions using that predicate (excluding uses in opaque contexts). We express a proposition in assertorically uttering the sentence “Torture is morally wrong,” and the predicate “morally wrong” contributes something to the truth conditions of that proposition. I say that it contributes the property of moral wrongness.
24
Normativity and reasons
25
and I will set it aside. It seems clear, for example, that the fact that torture is wrong is normative in at least some sense, whereas the fact that torture is widespread is not normative. It seems clear, moreover, that this difference between the facts is due to the nature of wrongness, which is at least in this respect unlike the nature of the property of being widespread. In any event, I am going to assume for present purposes that naturalists agree with non-naturalists that moral properties and facts are normative. They might disagree about what the normativity of these facts and properties consists in, but they agree, I assume, that they are normative in some sense. The fundamental issue, if I am correct, is whether naturalism can account for this. In this paper, I consider five arguments against naturalism that have recently been proposed by Derek Parfit (2011), and, following Parfit, by David McNaughton and Piers Rawling (2003) and Jonathan Dancy (2006). All five attempt, in one way or another, to show that a natural fact cannot be normative. To understand the arguments, and to understand what is at stake, we need to understand what naturalists and nonnaturalists mean by a “natural” property or fact and we need to understand what the “normativity” of a property or a fact might consist in. I consider the idea of a natural fact in section 2.1 and the idea of normativity in 2.2. In section 2.3, I provide an overview of the five arguments. In section 2.4, I present the essentials of my own account of normativity, so that I can use this account to illustrate problems with the arguments. I discuss and criticize the arguments in detail in sections 2.5–2.9. I conclude with a brief attempt to characterize where the arguments go wrong. 2 .1 N at u r a l fac t s a n d n at u r a l prope r t i e s In this section, I attempt to explain what is at stake in the debate between naturalists and non-naturalists. Why does the debate matter philosophically? I then attempt to characterize the idea of a natural property and fact. Finally, I explain how a naturalist might attempt to settle the debate. What must a naturalist achieve, and what must a non-naturalist show cannot be done? All parties to the debate agree that there are natural properties. They agree moreover that there are moral properties but disagree whether these are a kind of natural property.2 When characterized in this way, 2 Some philosophers who view themselves as naturalists deny that any moral or other normative property is (or is identical to) a natural property but argue nevertheless that there is a sufficiently close relation between normative properties and natural properties to underwrite a weaker form
26
dav i d c opp
the disagreement might appear to be merely a matter of taxonomy. But there is more involved than this. To get to the bottom of the debate, we need to understand what is at stake that might seem to be philosophically important. Naturalists claim that moral properties are natural properties, but it is not obvious what this means, and, more important, it is not obvious why it matters philosophically whether naturalists are correct. As I said, I think the fundamental issue is whether the normativity of moral properties and facts can be understood naturalistically. Nonnaturalists hold that the normativity of moral properties and facts sets them apart from natural properties in a philosophically important and deep way. Because they are normative, they are sui generis, and in a different metaphysical and epistemological category from natural properties and facts. This, in any case, is what I shall be assuming lies behind the non-naturalist’s view. Naturalists have to agree that moral properties and facts differ from non-normative properties and facts, but, despite this, they hold that the normativity of these properties and facts can be understood naturalistically. For the naturalist, even though moral properties are normative, this does not mean that these properties are metaphysically or epistemologically distinctive in any philosophically fundamental way. Consider a few analogies. Vitalists claimed that living organisms are “inherently different” or “fundamentally different” from nonliving entities.3 They claimed that their property of being alive sets living organisms aside from nonliving entities and precludes their being understood exclusively in physical terms. Their opponents denied this. Dualists claim that mental states, including especially conscious experiential states such as the state of being in pain, are different in such a fundamental way from other states of a person that they cannot be understood physically, in the way that these other states can be understood.4 Their opponents deny that a physical understanding of such states is ruled out. I believe the debate between naturalists and non-naturalists is of the same kind as these debates even though the issue is the adequacy of naturalism rather than the adequacy of physicalism. The issue is whether the normativity of moral properties is a natural phenomenon or one that can be understood of naturalism. For example, Ralph Wedgwood argues that “all contingent normative facts are realized in natural facts” and he characterizes this as a form of naturalism (2007: 6; see 145). I set aside theories of this kind. The arguments I consider in this paper are intended to attack the thesis that normative properties are natural properties. 3 Bechtel and Richardson (1998). Derek Parfit suggests this comparison (2011: v. II, 324). 4 The classic statement of this view is found in Nagel (1974).
Normativity and reasons
27
naturalistically. Call this the “naturalistic explanatory issue.”5 Lying in the background is the important task of explaining normativity, the task of explaining what it is for a property or fact to be normative. It is because this background issue is compelling that the naturalistic explanatory issue is compelling.6 To understand this issue, however, we need to understand the idea of a natural phenomenon. G. E. Moore thought that the characteristic thesis of moral naturalism is that moral properties are natural properties.7 The difficulty is to explain the idea of a natural property, and indeed Moore confessed, in the preface to the second edition of Principia Ethica, and also in his “Reply to My Critics,” that his attempts to explain the idea in Principia had been “hopelessly confused” (Moore 1993a [1903]: 13; also Moore 1968: 582). His initial characterization of moral naturalism in Principia has nevertheless been highly influential. According to “naturalistic ethics,” he said, “Ethics is an empirical or positive science: its conclusions could all be established by means of empirical observation and induction” (Moore 1993b [1903]: section 25, 91). He added that by “nature” he meant “that which is the subject matter of the natural sciences and also of psychology” (section 26, 92). Following Moore, many writers have attempted to explain the idea of a natural property or fact by reference to the subject matter of science. Parfit says, for example, “Some fact is natural, on one common definition, if facts of this kind are investigated or discussed by people working in any of the natural or social sciences.”8 Russ Shafer-Landau characterizes naturalists as holding that moral properties “would figure ineliminably in perfected versions of the natural and social sciences.”9 We can call such characterizations “disciplinary.” 5 This way of understanding the basic issue helps to explain why someone might be a moral naturalist without wanting to commit herself to a global metaphysical naturalism. She might think that normativity can be understood naturalistically but be uncertain whether mathematical facts are natural facts or whether there are supernatural facts. 6 I mean to leave open the possibility of a philosophically interesting non-naturalist account of what it is for a property or a fact to be normative. Ralph Wedgwood denies that normative properties and facts are natural properties and facts, but he views himself as a naturalist because his account of normativity supports the view that “all contingent normative facts are realized in natural facts” (2007: 6). The interesting issue is not whether Wedgwood’s theory qualifies as a kind of naturalism. It is whether his account of normativity is satisfactory. 7 Moore (1993b [1903]: section 26, 91). For reasons we need not explore, Moore was primarily concerned with whether goodness is a natural property. 8 Parfit (2011: v. II, 305). He admits this may be vague (306). 9 Shafer-Landau (2003: 59). In the end, Shafer-Landau interprets this characterization epistemologically, as I will explain in a note.
28
dav i d c opp
One trouble with disciplinary characterizations is that many kinds of facts and properties that are plausibly taken to be natural do not figure in, and are not investigated by, any science. For instance, there is the property of being heavier than a 1966 Cadillac. There is the property of being a steak knife and the property of having been born in September. There is the property of being a pebble. The list could be greatly extended. Engineering is usually distinguished from science, but properties of interest in engineering are plausibly taken to be natural. Consider, for example, the property of being a suspension bridge. Non-naturalists would reject the idea that normativity can be understood in terms of engineering properties, or in terms of properties relevantly similar to the other properties I listed. The fact that these properties are not of interest in science would not speak to the underlying issue. A second problem with disciplinary characterizations is that there is not a sharp boundary between science and other endeavors. Ethical issues are sometimes investigated in the social sciences. Economists study theories of distributive justice, for example.10 It is not clear that we are warranted in treating this area of economics as nonscientific. Historians also sometimes make moral judgments in the course of their work,11 and it is not clear why it should matter for our purposes whether history is a social science. It is dubious that ethics is a science, but it is not clear why anything of interest to us here turns on this. The issue, I want to say, is whether normativity can be understood naturalistically, not whether it can be explained scientifically. For these reasons, I think we should abandon disciplinary characterizations of the natural. Moore suggested a better approach in saying that, according to the naturalist, ethical conclusions could all “be established by means of empirical observation and induction.” We can take Moore here to be suggesting an epistemological characterization of moral naturalism, according to which ethical knowledge is empirical, or as he says, it is based in “empirical observation and induction.” It is no part of naturalism, of course, to deny that we have a priori knowledge of conceptual or analytic moral truths. Hence, the naturalist should be taken to hold that moral facts are empirical in that any knowledge or justified belief we have of synthetic, substantive moral truths is empirical; moreover, moral properties are empirical in that any knowledge or justified belief we have of synthetic, substantive propositions about their instantiation is empirical.
See, for instance, Roemer (1996). This has been noted by Nicholas Sturgeon (1985).
10 11
Normativity and reasons
29
Or, perhaps better, the naturalist should be taken to hold that we have or can have substantive moral knowledge or justified belief of synthetic moral truths but that no such knowledge or justified belief is a priori.12 Call this the “empirical conception” of the natural. I have claimed that the fundamental disagreement between naturalism and non-naturalism concerns the naturalistic explanatory issue. On the empirical conception of the natural, the issue is whether normativity is an empirical phenomenon. On this conception, then, naturalists need to defend the claim that the property of being normative, like the moral properties themselves, is natural or empirical. But to address the explanatory issue in a satisfying way, naturalists need to go beyond merely defending this claim. They need to explain what it is for a fact or property to be normative without invoking, referring to, or quantifying over any non-empirical facts or properties. And again without invoking, referring to, or quantifying over any non-empirical facts or properties, they need to explain what it is for something to have a moral property, such as to be wrong, and to show that moral properties are normative, on their account of normativity. Nicholas Sturgeon has pointed out that the thesis that the moral properties are natural properties commits naturalists, trivially, to the doctrine that, for each moral property, M, there is a natural property, N, such that M is identical to N. For, of course, if M is a natural property, then since M is identical to itself, M is identical to some natural property, namely itself (Sturgeon 2006: 98). We might add that naturalists are committed, trivially, to the doctrine that there is a natural property that is identical to the property that moral properties have of being normative. For, of course, if being normative is a natural property, then since it is identical to itself, the property of being normative is identical to some natural property, namely itself. I claim, however, that to adequately address the naturalistic explanatory issue, naturalists need to propose a substantive and philosophically satisfying account of what it is for a property and a fact to be normative. And, I believe, they also need to propose a substantive and philosophically satisfying account of what it is for something to have a moral property, such as to be wrong. To do this, I think, they ideally would defend a nontrivial and explanatorily and philosophically interesting statement of 12 For a more detailed and cautious development of an epistemological characterization, see Copp (2007: ch. 1 [“Why Naturalism?”]). Shafer-Landau opts in the end for an epistemological characterization (2003: 61). Parfit in effect proposes a similar account (2011: v. II, 306–307). Sturgeon objects to an epistemological characterization (2006: 109–10).
30
dav i d c opp
the form “To be normative is to be N,” where “N” is replaced by a term standing for a natural property. And I believe they also ideally would defend nontrivial and explanatorily and philosophically interesting statements of the form “To be M is to be N*,” where “M” is replaced by a moral predicate, such as “wrong,” and “N*” is replaced by a term standing for a natural property.13 If it aims to address the explanatory issue in a philosophically fully satisfying way, moral naturalism must, in this way, be “reductive.”14 Reductive naturalism of this kind is not committed to the further thesis that the crucial statements of the forms, “To be normative is to be N,” and “To be M is to be N*,” are analytic or conceptual truths. “Analytic naturalism” accepts this further thesis, but naturalism as such is not committed to it. For this reason, I will focus attention in what follows on “non-analytic naturalism,” which is a form of reductive naturalism that does not claim that the crucial statements are analytic or conceptual truths. The key point, then, is that, to address the explanatory issue, naturalists need to provide the needed explanations, ideally by proposing and defending nontrivial and explanatorily and philosophically interesting property-identity statements of the above forms. Non-naturalists need to argue that no such statements are true. They need to show that propertyidentity statements of the above forms must either be trivial or false, that there are no true, nontrivial, and philosophically explanatory propertyidentity statements of the kind at issue. Suppose, however, that the non-naturalist arguments are successful and that there is not a satisfactory naturalistic explanation of normativity. It does not follow that non-naturalism is correct, for it is compatible with naturalism to hold that normativity is fundamentally unanalyzable and primitive. Even if there is not a philosophically satisfying naturalistic account of normativity, it might still be true that the moral properties, as well as the property these properties have of being normative, are natural properties. A non-reductive naturalism of the kind proposed by Sturgeon (2006: 98–99) might be true. But the issue whether naturalism is true is 13 Notice that the naturalist will have to argue that the natural property N* is normative, assuming that the moral property M she identifies it with is normative. If naturalism is correct, and if moral properties are normative, some natural properties are also normative. 14 Sturgeon defends a view he calls “non-reductive naturalism,” according to which, although moral properties are natural properties, there may be no true and non-trivial statement of the form, “To be M is to be N” (2006: 98–99). Non-reductive naturalism of this kind does not aspire to provide a satisfactory response to the explanatory issue. In what follows, I use the term “nonreductive naturalism” to refer to Sturgeon’s view.
Normativity and reasons
31
much less compelling if we give up the goal of addressing the naturalist explanatory issue in a satisfying way and limit ourselves to defending non-reductive naturalism. In this section, I have proposed that the debate between naturalists and non-naturalists is of philosophical interest chiefly because of the important background issue of explaining what it is for a property or fact to be normative. The naturalistic explanatory issue is the issue whether normativity can be explained naturalistically. It raises the important question of whether a kind of reductive naturalism can be defended. I have distinguished between two kinds of reductive naturalism, analytic and non-analytic naturalism. Both aim to address the explanatory issue by defending reductive, philosophically substantive and interesting propertyidentity claims of the kinds at issue. But because non-analytic naturalism is not committed to the further thesis that these identity claims are analytic or conceptual truths, it is easier to defend. The key question, then, is whether a version of non-analytic naturalism is defensible. 2 .2 Nor m at i v e fac t s a n d nor m at i v e prope r t i e s Normative facts and properties are often said to be “prescriptive,” or “action-guiding,” or “authoritative,” but these terms are unhelpful, and there is, unfortunately, little agreement about the nature of normativity. We need to distinguish the property of being normative from other properties of moral facts and properties. The trouble is that there is substantive disagreement about what the normativity of moral properties consists in or involves. Because of this, the naturalistic explanatory issue is not well defined, and philosophers who address the issue often talk past one another. A non-naturalist might contend that, to be normative, a moral fact must have some specified feature, but a naturalist might deny this. Naturalists might devote energy to showing that naturalism can account for some feature of moral properties, and non-naturalists might agree that naturalism can do this but deny that, in doing so, naturalism has accounted for the normativity of moral properties. Lack of agreement about the concept of normativity means that the debate about the explanatory issue is frustrating and indecisive. It should be evident, moreover, that morality is not the only field in which we find normative properties and facts. There are many kinds of non-moral facts that are normative, including facts about justified belief, about prudential reasons to act one way or another, and so on. There is, for instance, a requirement of self-grounded practical reason that one pay
32
dav i d c opp
sufficient attention to one’s own future good. A naturalist would want to explicate normativity in general, and not merely the normativity of moral properties and facts. The explanatory issue, then, is whether normativity is a natural phenomenon, not merely whether the normativity of morality is a natural phenomenon. We are concerned with the plausibility of normative naturalism and not merely moral naturalism. There is disagreement about how “stringently normative” moral facts are, by comparison with other kinds of normative fact. Some view practical reason as the foundation of normativity. On a view of this kind, moral requirements are normative only if they are requirements of practical reason. Others view morality as the pinnacle of normativity. On a view of this kind, moral requirements override requirements of practical reason in cases of conflict, if any, between requirements of the two kinds. Others view morality and rationality as being on a par. On a view of this kind, moral requirements are normative even if they can conflict with what is required by practical reason, and even if they do not override what is required by practical reason. Disagreements about stringency further confuse the debate about the naturalistic explanatory issue. Naturalists might argue that moral facts have a certain feature, and non-naturalists might agree that to have this feature is to be normative in a way but claim that this is a trivial kind of normativity by comparison with the more stringent normativity of moral facts. I cannot explore the concept of normativity in any depth in this paper. I have looked at the issues elsewhere and do not want to repeat myself.15 Perhaps, however, we can agree on the following rough characterization, which I will call the relevance characterization. A normative fact has a certain kind of direct intrinsic relevance to our decisions about what to do, or to our choices; in virtue of the content of this fact, it constrains or guides our decisions or choices.
This characterization needs to be clarified, of course, and it is vague. There is room to disagree about exactly what kind of relevance to decisions or to choices a normative fact would have, or about exactly in what sense a normative fact would constrain or guide decisions or choices.16
15 See Copp (2007: ch. 8 [“Moral Naturalism and Three Grades of Normativity”]); Copp (2009, 2010a, 2010b). 16 Dancy suggested in personal communication that the concept of relevance might be normative. Even if it is, I intend here merely to suggest a characterization of normativity, not a reductive definition.
Normativity and reasons
33
We can use the relevance characterization to explain the notion of a normative property. Consider the fact that torture is morally wrong. This fact obviously would be relevant to decisions whether to torture anyone and to choices of whether to support proposals to torture anyone, so it would be normative according to the relevance characterization. The important point is that this fact consists in torture’s being wrong. The fact that torture is widespread is not normative, and the important difference between it and the fact that torture is wrong is obviously that the latter is about the wrongness of torture whereas the former is about its being widespread. Therefore, it seems, the fact that torture is wrong is normative because of the fact that it consists in torture’s having the property wrongness. In this sense, wrongness is a normative property.17 Naturalists and non-naturalists might interpret the relevance characterization in different ways. To use an example of Dancy’s, suppose it is a fact that some action A will make many people more comfortable without making anyone less comfortable (Dancy 2006: 134). This fact would seem to have a direct relevance to a decision about whether to do A. So it looks as if this fact would count as normative, on the relevance characterization, even though it is plainly a natural fact. Naturalists might take comfort in this point. Dancy thinks, however, that a fact of this kind is intuitively non-normative, and furthermore, he says, characterizations that count facts of this kind as normative will be of no help in showing “what the nonnaturalists are driving at when they say that no natural fact could be a normative fact” (Dancy 2006: 143, n. 9; 134). He would say, then, that the relevance characterization provides at best a necessary condition of a fact’s being normative, not a sufficient condition. Looked at differently, however, the example suggests why naturalists and non-naturalists would want to interpret the relevance characterization in different ways. A nonnaturalist might claim that even if a fact of the kind in question might be relevant to a decision as to what to do, it is not relevant in the direct and intrinsic way that is characteristic of normative facts, such as the fact that doing A would be right.18 17 Nothing here turns on metaphysical disputes about the nature of properties. See above, note 1. 18 Some moral facts might not count as normative, according to the relevance characterization. For example, the fact that [either torture is morally wrong or California is west of New York] is not normative, on the relevance characterization. Nevertheless, all “basic” moral facts count as normative – where a basic moral fact is a fact that consists in something’s having a moral property. That is, a “basic” moral proposition is one that ascribes a moral property such as wrongness or moral goodness to something. The fact that torture is wrong is “basic” in this sense, for, as moral realists understand matters, it consists in torture’s having the property of being wrong. Any basic moral fact would be normative under the relevance characterization, for any such fact has
34
dav i d c opp
For Dancy, the central normative facts are facts regarding what one ought to do, or what one has reason to do. They “bear their practical relevance on their face; they are explicit answers to the question what to do” (Dancy 2006: 136). Non-normative facts can be normatively relevant, so they can of course be cited in answers to the question what to do. Consider the fact that doing A will make many people more comfortable without making anyone less comfortable. This fact may make it the case that one ought to do A or it may be a reason to do A. It is therefore normatively relevant, and in citing it, I may be telling you what to do in an inexplicit way. Such facts are not themselves normative, however, because the answers they provide to the question what to do are not explicit and direct. Some facts about such facts are normative, however. These are metafacts, regarding a fact F, that it stands in a normative relation to an action A by either making it the case that someone ought to do A or being a reason to do A. Dancy proposes that metafacts of this kind are “the central normative facts, by reference to which the normativity of all others is to be understood” (Dancy 2006: 137). Each metafact of this kind, as well as the more basic and simple normative facts regarding what one ought to do or what there is a reason to do, provide direct and explicit answers to the question what to do.19 I will refer to Dancy’s account as the “explicitanswers” account. I believe that Dancy’s proposal is unsatisfactory. I will mention two problems. First, if Alice asks Bill what to do, the most direct and explicit answers would be imperatives of the form, “Do A” (see R. M. Hare 1952: section 1.1). If Bill were to answer, “You morally ought to do A,” or “You have good reason to do A,” he would have been rather inexplicit by comparison, and Alice might imagine him to be thinking, “But do B instead.” Dancy would reply that the most explicit and direct factual response to the question what to do is always a normative claim.20 But if Bill said, “Doing A would make many people more comfortable without making anyone less comfortable,” it is not obvious to me that his answer would be any less explicit and direct than if he gave an overtly normative the characteristic direct relevance to choices and decisions that distinguishes the normative from the non-normative. 19 I set aside Dancy’s account of evaluative facts, which he says are normative because they make a difference to how to act even though they do not specify what is to be done (Dancy 2006: 137–38). The trouble is that they do not provide a direct and explicit answer to the question what to do. They therefore would seem to be on a par with non-normative facts, such as the fact that doing A would make many people more comfortable without making anyone less comfortable, which also make a difference to how to act. 20 He gave this reply in a personal communication.
Normativity and reasons
35
answer, such as “You morally ought to do A.” In each case, Alice could imagine him to be thinking, “But do B instead.” Second, whether a statement directly answers a question depends on how the question and answer are formulated, and not merely on what fact the answer cites. Suppose you ask me where Muhammad Ali was born. If I reply that Cassius Clay was born in the USA, I have not answered your question directly and explicitly even though the fact that Ali was born in the USA is identical to the fact that Clay was born in the USA. I return to this second problem below. For these reasons, it seems to me that Dancy’s account is unsatisfactory. Parfit holds that “normativity is best understood as involving reasons or apparent reasons” (Parfit 2011: v. II, 269), and he indicates that he mostly uses “normative” in what he calls the “reason-implying sense” (v. II, 310). He says, “When I call some claim normative in the reasonimplying sense, I mean roughly that this claim asserts or implies that we or others do or might have some reason or apparent reason” (v. II, 268). There are, he suggests, reason-implying uses of the words “good” and “bad” (v. I, 38) as well as of the words “ought,” “should,” and “must” (v. I, 33). A fact gives us a reason for something, Parfit says, when it “counts in favor” of that thing (v. I, 31). He thinks that “all reasons have normative force” (v. I, 35). In cases in which we have most reason to act in one way rather than any other, Parfit stipulates that our reasons for acting that way count as “decisive,” and he stipulates as well that when our reasons for doing something are decisive, this is what we ought to do in the “decisive-reason-implying sense” (v. I, 33). He thinks there are other senses of “ought,” however, including “several moral senses” (v. I, 37). Non-normative facts can give us reason to do something. For instance, the fact that there is an angry snake on the path in front of us gives us reason to get off the path. This fact is, in this sense, reason-involving and reason-implying. But normative facts, such as the fact that we ought to get off the path, are reason-implying in a stricter sense. I think Parfit means to claim that normative facts are those that entail the existence of a reason. The fact that there is a snake on the path does not entail that we have a reason to get off the path. Parfit recognizes, however, that there are views on which “we might have no reason to do what we ought morally to do” (v. I, 38). Hence, we cannot assume that the fact that we morally ought to do something entails that we have a reason to do it. This means that on Parfit’s proposal, if it is understood the way I suggest, we cannot assume that moral facts are normative in the reason-implying sense.
36
dav i d c opp
Parfit’s view therefore faces a problem. To avoid counting some non-normative facts as normative, it needs to be understood in a way that might lead to the conclusion that moral facts are not normative in the reasonimplying sense. To avoid this conclusion, Parfit would need an argument to show that moral facts entail the existence of reasons. In the absence of such an argument, moreover, Parfit could not fault moral naturalism for an inability to account for the normativity of moral facts in the reasonimplying sense. Dancy criticizes accounts that, like Parfit’s, seek to explain normativity in terms of reasons on the ground that they merely point to a central normative concept, the concept of a reason, without saying what normativity is. He contends that non-naturalists need to have an account of what normativity is in order to show that nothing natural is normative (Dancy 2006: 136). It will be useful to keep this criticism in mind. Dancy’s account is open to a similar criticism, however, for it seems merely to point to a feature of normative facts, the fact that normative facts can be cited in answering the question what to do. This seems to point to something all normative facts have in common without telling us what normativity consists in. If I am correct, naturalists and non-naturalists alike are in a difficult position. For the naturalist explanatory issue is not well defined if we lack a shared account or concept of normativity. I do not mean to say that we need a theory of what normativity consists in before we can begin. But we need to be sure that we are not talking at cross-purposes. If we do not have a shared account or concept of normativity, we might be quite unclear about what the naturalist is attempting to explain, and we might not be able to tell whether a proposed explanation is successful. We might also be unable to tell whether a proposed argument against naturalism is successful. Despite this, however, I now proceed to look at the five arguments. 2 .3 F i v e a rgu m e n t s The best-known and most-discussed argument against moral naturalism is G. E. Moore’s “open question argument.” I shall ignore it in what follows both because I want to focus on the five new arguments and because, in my view, Moore’s argument has been shown to be unsound (see Dancy 2006: 130–31; Sturgeon 2006: 93–99). Moreover, Moore’s argument presupposes that naturalism must take the form of analytic naturalism and is ineffective against non-analytic naturalism, which is the form of
Normativity and reasons
37
naturalism that interests me. Some of the new arguments can perhaps be viewed as attempts to show that non-analytic naturalism faces problems similar to those that Moore thought to afflict analytic naturalism. But I will not explore the relation between Moore’s argument and the new arguments. The new arguments overlap in a way that makes it difficult to separate them cleanly. I will nevertheless distinguish five, or five families of, arguments. Each aims to show that moral facts and properties cannot be natural, and each turns, in one way or another, on the idea of normativity. If successful, each shows that no natural fact or property can be normative. The first and perhaps the most fundamental turns on the idea that normativity is to be understood in Parfit’s reason-implying sense. The objection is that naturalism cannot provide a defensible account of practical reasons. Call this the Argument about Reasons. The second argument rests on the idea that although a natural fact can be the reason why some normative fact obtains, there is a distinction between a fact’s being normatively relevant in this way and a fact’s being normative. Any normative fact holds because some natural fact obtains. We could not make sense of this if this normative fact were that natural fact. Call this the Argument from Because. Any reductive naturalistic theory must identify each normative property M with some natural property N. According to the third argument, any such property-identity thesis will drain the normativity from claims to the effect that things that are N are M. Intuitively, such claims are substantive and normative, but this is not so, if the naturalist is correct. Call this the Normativity Objection. According to the fourth argument, the Triviality Objection, if M is identical to N, then the claim that things that are N are M is equivalent to the trivial claim that things that are N are N, and the claim that M is N is equivalent to the trivial claim that N is N. But then it is unclear what important information might be conveyed by the property-identity thesis. If naturalism is true, it is trivial. Finally, if the naturalist is correct, there is no point to making moral claims. Naturalism is akin to nihilism. Attempts to deny this face the Soft Naturalist’s Dilemma, which is the final argument I will consider. In previous work, I myself have proposed a kind of normative naturalism (see Copp 1995 and 2007). My responses to the five arguments reflect my positive views, but I will avoid bringing my own version of naturalism into the discussion except to illustrate how a naturalist can resist the arguments. In order to lay the foundation for doing this, I use the next section to introduce briefly the basic idea of my position.
38
dav i d c opp 2 . 4 Nor m at i v i t y a n d probl e m s of nor m at i v e g ov e r n a nc e
In my view, normative facts are facts that concern solutions to, or ways to ameliorate, certain generic problems faced by human beings in the circumstances they face in their ordinary lives.21 I call them problems of normative governance because they are problems that we can better cope with when we subscribe to appropriate systems of norms. Two of the most important and most familiar problems of this kind are the problem of sociality and the problem of autonomy. The problem of sociality is the problem humans face because, although they need to live in societies, there are familiar causes of discord and conflict that risk undermining societies or at least making societies less successful than they otherwise could be at enabling people to have lives of the kinds they want to have. The currency of a moral code in a society can help to ameliorate this problem provided its content is such that people who subscribe to it are thereby motivated to cooperate, and generally to avoid discord and conflict and to act in “pro-social” ways. Of course, some moral codes would do more than others to ameliorate the problem. My “society-centered” proposal is that, roughly, the moral truth is a function of the content of the moral code the currency of which in society would do most to ameliorate the problem of sociality. Call this the “ideal code.” The idea is that we are morally required to do something if and only if, and because, the ideal code requires us to do this. The fact that we are morally required to do something is the fact that the ideal code requires us to do it. We have decisive moral reason to do something if and only if, and because, the ideal code requires us to do it. On this view, morality is the solution to a problem in social engineering, the problem of equipping people to live comfortably and successfully together in societies.22 The problem of autonomy is a familiar problem each person faces because of the complexity of our psychologies and because we live through extended periods of time. We have things that we value, in that we aim to achieve them and we attach great psychological significance to whether we are successful. We tend to feel enhanced or sustained when we are successful and we tend to feel shame or guilt or to feel diminished when we fail. In this sense, our values are aspects of our “identities.” An 21 For details, see Copp (2009). See also the essays and books I mention in other footnotes in this section. 22 For details, see Copp (2007, esp. the introduction), and, for an early version of the view, Copp (1995).
Normativity and reasons
39
“autonomous” person does well at governing her life in accord with what she values. The trouble is that we are easily distracted by temptations and we often find our commitment to our values wavering, especially in cases in which costs to our values are uncertain or temporally remote. Our subscription to a norm that calls on us to promote the conditions of our autonomy can ameliorate this problem by enhancing and reinforcing our motivation to live in accord with our values. Call this the “autonomy norm.” My proposal is that the truth as to what we have most practical reason to do is a function of what is required by the autonomy norm. We are required to do something as a matter of practical rationality if and only if, and because, the norm of autonomy requires us to do this. We have decisive practical reason to do something if and only if, and because, the norm of autonomy requires us to do it. The fact that we have decisive practical reason to do something is the fact that the norm of autonomy requires us to do it.23 My account of the nature of normative moral facts and facts about practical reasons can be generalized to provide an account of all kinds of normative fact. I call the generalized view pluralist-teleology. It is pluralist, for it says that there are different kinds of normative fact. There are, for instance, different kinds of reasons, including moral reasons, self-grounded practical reasons, epistemic reasons, and so on. The view is teleological in that it seeks to explain normativity as “grounded in” solutions to problems of normative governance. Pluralist-teleology aims to provide a unified account of reasons of all kinds. Pluralist-teleology is an example of non-analytic normative naturalism, for of course I do not claim that the theory or its central claims are conceptual or analytic truths. It therefore illustrates the kind of view that I want to test against the five arguments. Obviously this is not the place to present the theory in detail. What I have said raises many questions, but I will have to set them aside since my main task in this paper is to look at the arguments. 2 .5 T h e A rgu m e n t a b ou t R e a s ons The naturalist explanatory issue is central to the debate between naturalism and non-naturalism. So if we accept Parfit’s proposal that normativity 23 For details, see Copp (2007: ch. 10 [“The Normativity of Self-Grounded Reason”]). In that chapter, I spoke of the “values standard” rather than the “norm of autonomy.” This is because the view developed there simplified matters by setting aside reasons given us by our basic needs. I address this matter in Copp 2001. See also Copp (1995: 172–88).
40
dav i d c opp
is to be understood as reason-implying, the key question is whether naturalism can provide an adequate account of reasons. In the debate between naturalism and non-naturalism, Parfit writes, “reasons provide the decisive battlefield.” He begins with “non-moral practical reasons.” If naturalism fails here, he believes it will also fail “when applied to morality” (Parfit 2011: v. II, 269). Naturalists and non-naturalists agree that a natural fact F can give someone a practical reason. For instance, the fact that your food is poisoned might give you a reason not to eat it. In such a case, the fact F has the normative property of being a reason. The metafact that the fact F gives the person a reason to do that thing is also normative. And the claim that this metafact obtains is normative. “It is easy to overlook” such metafacts and claims, says Parfit, as well as the property of being a reason. The question, however, he says, is whether this property, and facts and claims of these kinds, are “irreducibly normative” (Parfit 2011: v. II, 280, 266–67). That is, he explains, the question is whether the property of being a reason, and normative facts and claims involving this property, can be “defined or restated in non-normative terms.”24 Parfit appears to be discussing analytic naturalism in this passage, where, as I explained, analytic naturalism adds to the basic doctrine of reductive naturalism the thesis that the characteristic naturalistic identity claims are analytic or conceptual truths. I want to focus attention on non-analytic naturalism. Non-analytic naturalism aims to defend “reductive” naturalistic identity claims, yet because it does not hold that these claims are analytic or conceptual truths, it apparently would count, in Parfit’s terms, as holding that normative claims are “irreducibly normative.” This is needlessly confusing. I will say instead that non-analytic naturalists hold that the property of being a reason and facts and claims involving this property are “naturalistically reducible” but not “analytically reducible.” The key question is whether the property that a fact can have of being a reason for someone to do something can be given an informative and philosophically satisfying naturalistic explication or reduction. This property is actually a relation that relates a fact F to a person P and a kind of action A when F is a reason for P to do A. A naturalistic account of this relation would identify a natural relation N that holds between facts, persons, and kinds of action, and it would propose that for F to be a reason 24 He explains that he calls a normative concept, claim, or fact “irreducibly normative” if it “cannot be defined or restated in non-normative terms” (Parfit 2011: section 24 [82], 572).
Normativity and reasons
41
for P to do A is for F to stand in relation N to P and A. The question is whether Parfit has an argument to show that no such proposal can be true. Parfit does not address the issue in such general terms. Instead he considers Subjectivism about Practical Reason according to which For F to be a decisive reason for a person to do A is for F to be such that doing A is what “would best fulfil this person’s present fully informed telic desires, or is what, after fully informed and procedurally rational deliberation, this person would be most strongly motivated to do, or would choose to do” (2011: v. II, 270).
Some of Parfit’s arguments seem intended to attack the thesis that this proposal is a conceptual or analytic truth. He suggests, for instance, that if this proposal were a conceptual or analytic truth, Subjectivism about Reason would not be a “substantive normative claim,” but it would be a “concealed tautology” (2011: v. II, 277), and, more important, claims about reasons would not be normative and could not be used to give advice (2011: v. II, 278–85). I will not discuss this objection. The important question is whether Parfit has good reason to deny the proposal, not whether he has good reason to deny that it is a conceptual or analytic truth. Parfit might have in mind the similar objection that if Subjectivism is true, then the claim that a person has decisive reason to do what would best fulfill her present fully informed telic desires is trivial. For it states the same fact as the claim that a person’s doing what would best fulfill her present fully informed telic desires would best fulfill her present fully informed telic desires. I will discuss this kind of argument later, where I address the Triviality Objection. Although Parfit appears to be arguing against analytic Subjectivism about Reason, his reasoning suggests the following pair of arguments against the non-analytic version of the theory. First, facts about what a person ought all things considered to do are normative in the reason-implying sense. But it is possible for an action to be what a person ought all things considered to do even if, according to Subjectivism, the person does not have decisive reason to do it because the action would not “best fulfil this person’s present fully informed telic desires” and the person would not be motivated to do it or choose to do it “after fully informed and procedurally rational deliberation.” Parfit gives two examples. In “Burning Hotel,” Allan is at the window of a burning hotel and can only save himself by jumping into the canal under his window. In this situation, all
42
dav i d c opp
things considered, Allan ought to jump even if jumping would not best fulfill his present desires or be what he would choose to do after ideal deliberation (2011: v. II, 292). In “Revenge,” Brenda wants to kill her enemy. This is what would do most to fulfill her present desires, and is what she has chosen to do after ideal deliberation, yet, all things considered, she ought not to do this (2011: v. II, 281, 284). Hence, Subjectivism implies, implausibly, that facts about what a person ought all things considered to do are not necessarily normative in the reason-implying sense. According to the second argument, Subjectivism is committed to a kind of semantic error theory. For we might know all the relevant psychological facts about Allan in Burning Hotel and about Brenda in Revenge, yet, despite what is implied by subjectivism, we might think that Allan ought to jump into the canal and that Brenda ought not to kill her enemy. If Subjectivism is true, however, then if we have these thoughts we must be mistaken about the truth conditions of our beliefs about reasons (2011: v. II, 292–93). Subjectivism about Reason therefore has implications that are difficult to accept. There are at least three major problems with these arguments. First, it is not obvious that normativity must be understood in the reason-implying sense. I have already mentioned worries about this idea. Second, in many cases a semantic-error theory seems highly plausible. For example, even though, as Parfit says, heat is molecular kinetic energy (2011: v. II, 299), many people might not realize this. They might not understand that their beliefs about heat refer to facts about molecular kinetic energy. A naturalist would say, similarly, that many people do not understand what facts are referred to by their normative beliefs. Third, even if Parfit’s argument undermines non-analytic Subjectivism about Reason, it does not thereby undermine naturalism unless naturalism is committed to Subjectivism. Parfit thinks that naturalism cannot avoid Subjectivism. He quotes Stephen Darwall’s claim that “For the philosophical naturalist, concerned to place normativity within the natural order, there is nothing plausible for normative force to be other than motivational force.”25 In Parfit’s view, this is because “subjective theories offer unified accounts of how a great variety of facts can give us reasons.” On a subjectivist view, a great variety of facts can give me reasons to act for they can ensure that “these acts might all fulfil one of my present desires, or be acts that, after some process of deliberation, I would be motivated to do, or would choose to do.” Parfit concludes, “If Naturalists are not Subjectivists, there is no similar 25 Darwall (1992: 168). Cited at Parfit (2011: v. II, 363). Darwall made a similar claim (1983: 80). Cited at Parfit (2011: v. II, 294).
Normativity and reasons
43
way in which they could explain how such a great variety of facts could give us reasons” (2011: v. II, 364). Subjectivism about Reason is of course a specific theory, and some philosophers have pursued other ways of explaining normativity in terms of motivational force.26 Even if we think that normativity must be explained in terms of motivational force, we do not need to accept Subjectivism. Moreover, there are naturalistic theories that do not explain normativity in terms of motivational force.27 One example is the pluralist and teleological theory of normativity that I sketched in 2.4. This theory holds that there are many different kinds of reasons, and yet it offers a unified account of reasons. Parfit therefore seems incorrect to suppose that naturalism is committed to Subjectivism about Reason. One need not favor my theory in order to avoid Subjectivism. Some philosophers take it to be a necessary truth that any agent who is morally required to do something has a good practical reason to do that thing, such that if she is fully rational, she will take this reason into account in deciding how to act. I think that this thesis is false, but I have argued elsewhere that it is compatible with moral naturalism (see Copp 2010a). If this is right, then a naturalist could take this thesis to constrain the theory of practical reason so that, in effect, her moral theory constrains her theory of practical reason. On this approach, a naturalist would first provide a naturalistic theory of morality and then develop a theory of practical reason that takes it as given that there is good reason to do what one is morally required to do. I have explained the idea in more detail elsewhere (Copp 2010a). For present purposes, the important point is that an approach of this kind is available. This means that a naturalist is not committed to Subjectivism about Practical Reason unless her moral theory is subjectivist in a relevant way. What a naturalist says about practical reasons may depend on what she says about morality. 2 .6 T h e A rgu m e n t f rom Be c aus e This argument rests on a distinction between normative facts and the non-normative facts in virtue of which, or because of which, these normative facts obtain. This is the distinction between facts that are themselves 26 I have in mind philosophers who deploy some version of internalism about the relation between moral belief and motivation in order to explicate normativity. See, for example, Smith (1994, esp. ch. 3). 27 I have argued that it is a mistake to think that normativity can be explained in terms of motivation. See Copp (2007: ch. 10 [“Moral Naturalism and Three Grades of Normativity”]).
44
dav i d c opp
normative and facts that are normatively relevant (Parfit 2011: v. II, 298– 303). Any normative fact obtains because some normatively relevant fact obtains.28 For example, if capital punishment is wrong, there are certain facts about capital punishment that make it wrong. More generally, if some action was wrong, it must have had some non-moral features in virtue of which it was wrong. Perhaps it was the telling of a lie and was wrong because of this. Nothing could be right or wrong or just or unjust or good or bad without having some other relevant features in virtue of which this is so. McNaughton and Rawling see a problem here for normative naturalism. A reductive naturalist must propose a theory that identifies each normative property M with some particular natural property N. But, McNaughton and Rawling contend, if such a proposal is to be at all plausible, it must at least be the case that when something is M, it is M because it is N. But if so, then a thing’s being M is not identical to its being N. The “because relation” is asymmetrical; it cannot hold between a fact and itself (McNaughton and Rawling 2003: 33). If an action is wrong, it is wrong because of some of its natural properties, and this means that wrongness is distinct from these natural properties. According to Utilitarian Naturalism, for example, the property of moral rightness is the property of maximizing happiness; given this, we are right to do what would maximize happiness. McNaughton and Rawling would say that if this is at all plausible, then if an act would be the right thing to do, it is right because it would maximize happiness (McNaughton and Rawling 2003: 42). But if this is so, the theory is false. For if this is so, the property of being the right thing to do and the property of being the act we can do that would maximize happiness are distinct. Yet Utilitarian Naturalism is true only if the properties are one and the same. Either a naturalist theory is not plausible or it is false. The trouble is that there is a “reductive use” of “because.” If Alice is asked whether Bill knows that Flin Flon is in Manitoba, she might reply that Bill does know this because it is true and Bill justifiably believes it. 28 J. L. Mackie famously asked, “what in the world is signified by this ‘because’?” (1977: 41). He claimed that there is a kind of metaphysical “queerness” here. In general, if something has a normative property, its having this property must be “consequential” or “supervenient” on its having some non-normative natural property. Suppose that some action is wrong “because it is a piece of deliberate cruelty.” Mackie asks, what is signified by “because?” The connection is not causal, and it “cannot be an entailment, a semantic or logical necessity.” One might claim that the “because relation” is not a natural relation and conclude that moral naturalism must be false. But non-naturalists also must explain what grounds this relation. For this point, see Blackburn (1985). For discussion of the point, see McNaughton and Rawling (2003).
Normativity and reasons
45
Alice says this, presumably, because she thinks that knowledge is justified true belief. Her use of “because” is an example of the “reductive because.” For another example, we might say that a certain figure is a circle because it is a line enclosing an area in a plane and there is a point within the area that is equidistant from every point on the line. This is obviously compatible with the thesis that to be a circle is to be a line enclosing an area in a plane such that a point within the area is equidistant from all the points on the line.29 Parfit also gives an example. Someone might say that an object is hotter because its molecules are moving more energetically. “Having such greater energy,” Parfit says, “does not cause this thing to be hotter, but is the same as being hotter, or is what it is to be hotter. Heat is molecular kinetic energy” (Parfit 2011: v. II, 299). There is, Parfit concedes, a “trivial” sense in which rightness is the property that makes acts right just as redness is the property that makes red things red.30 But there is also a different sense in which “when some act has some other property – such as that of saving someone’s life – this fact can make this act right.” Call this the substantive normative sense in which a property can “make” an act be right. Parfit suggests that when, in the substantive normative sense, “some property of an act makes this act right, this relation holds between two quite different properties” (2011: v. II, 301). For example, the fact that some action saved someone’s life might be what made the action right, and given this, then obviously, the action’s property of being right is distinct from its property of being a saving of someone’s life. McNaughton and Rawling are arguing that, “at best,” naturalism will identify normative properties with non-normative properties that are merely normatively relevant. Parfit also finds this mistake in some forms of naturalism.31 But I see no reason to think that naturalism must involve this mistake. According to Utilitarian Naturalism, for example, for an action to be right is for the action to maximize happiness. A Utilitarian Naturalist might say that maximizing happiness is the property that makes acts right, or that right acts are right because they maximize happiness. 29 The example is from Zimmerman (2001: 115). I owe this reference to Brian Coffey. 30 Parfit (2011: v. II, 300). Parfit’s use of the word “trivial” here anticipates the Objection from Triviality, which I discuss below. 31 He points out that even if there is some natural property that is the property that makes acts right, it does not follow that this is the property of being right (Parfit 2011: v. II, 300). We cannot establish the truth of naturalism by showing that some natural property is the unique property that “makes” acts right in the substantive normative sense. We would have to show that this natural property is the property that makes acts right in the reductive sense. This means that we would have to show that this natural property is identical to rightness.
46
dav i d c opp
But she would intend these claims in the reductive sense. McNaughton and Rawling would object that they are true at best in the substantive normative sense, not in the reductive sense. But it would be questionbegging to say this without some independent argument to support it. Utilitarian Naturalism is only an example. There are other forms of non-analytic moral naturalism, some of which lend themselves much less plausibly to the objection that the property they identify with moral rightness or wrongness is merely a property that makes acts right or wrong in the substantive normative sense. For example, on the society-centered view I described in 2.4, above, the property wrongness is (roughly) the property of being precluded by the moral code the currency of which in society would do most to ameliorate the problem of sociality. In my view, this is the property that makes acts wrong; actions that are wrong are wrong because they have this property. Obviously I intend these claims to be understood in the reductive sense. 2 .7 T h e Nor m at i v i t y Obj e c t ion The Normativity Objection seems to be Parfit’s chief objection. The goal is to show that no natural fact can be normative in the reason-implying sense. As Parfit says, “There is a deep distinction … between all natural facts and such reason-involving normative facts” (Parfit 2011: v. II, 310). Normative and natural facts “are in two quite different, non-overlapping categories” (v. II, 324). Even those who aim to defend some non-analytic form of naturalism must recognize, Parfit contends, that the normative concepts constrain what a normative property could possibly be. The concepts leave open various possibilities, and we must decide among them on non-conceptual grounds, but “[m]any other possibilities are, however, conceptually excluded.” Similarly, the concept of heat constrains what heat could possibly be. “[H]eat could not have turned out to be a cabbage, or a king … given the meaning of these claims, they could not possibly be true.” Similarly, moral naturalism and normative naturalism could not possibly be true, Parfit thinks (2011: v. II, 325). Suppose that you are in the top storey of your hotel, and you are terrified of heights. You know that, unless you jump, you will soon be overcome by smoke. You might then believe … that you have decisive reasons to jump, that you should, ought to, and must jump, and that if you don’t jump you would be making a terrible mistake. If these normative beliefs were true, these truths could not possibly be the same as, or consist in, some merely natural fact, such as [the fact that jumping would do most to fulfil your present fully informed desires].
Normativity and reasons
47
This, Parfit says, is the Normativity Objection (2011: v. II, 326–27). The objection expresses what is surely a driving idea behind nonnaturalism, the idea that naturalism is conceptually impossible because a natural fact could not be normative. The argument is, however, extremely thin. It has one premise, the claim that the normative concepts exclude the possibility that a normative property be natural. The argument is valid and it is on target, but Parfit gives us no reason to accept its premise. What he does, in the passages where he presents the argument, is to state and explain the key premise, to give an analogy, the example of heat, and then to apply the argument to the burning hotel example. Let us think about the analogy and the example. It is plausible that the concept of heat rules out the possibility that heat is a cabbage or a king. Similarly, it is plausible that the concept of rightness rules out the possibility that rightness is a rocket or a mountain lion or that it is the property of being a yellow rose. But the fact that the concept rules out these possibilities gives us no reason to think that it rules out the possibility that rightness is a natural property. There is no reason I can see to think that naturalists are making a mistake comparable to the mistake of thinking that heat is a turnip. Suppose someone claims it is conceptually excluded that heat is a physical property and concludes on this basis that heat is not molecular kinetic energy. This person agrees that changes in molecular kinetic energy are necessarily correlated with changes in heat, but denies that heat could be molecular kinetic energy on conceptual grounds. We should not be moved by the undefended claim that this is so. We should ask what reason there is to think that the concept of heat rules out the molecular theory. Similarly, we need a reason to think that the normative concepts rule out the possibility that a normative property is a natural property. Parfit presents the example of the burning hotel to illustrate his view. But in presenting the example, he merely claims, without argument, that the assumed fact that the person in the hotel has decisive reasons to jump “could not possibly be the same as … some merely natural fact.” He has in mind Subjectivism about Practical Reason, but the issue is completely general. Parfit wants to claim that the fact that a person has decisive reason to jump “could not possibly be the same” as any natural fact. We need to ask why this is so. It is not enough to respond that the concept of a reason ensures this, for this response merely pushes the problem back one step. We need to be given a reason to think that the concept of a reason ensures that the fact that a person has decisive reason to jump “could not possibly be the same” as any natural fact.
48
dav i d c opp
Dancy has presented an argument against naturalism that he also calls the normativity objection. But it is a different argument. Like Parfit, Dancy aims to show that a normative property or fact could not possibly be a natural property or fact. His basic idea is that “if we try to identify moral facts, or facts about what we have most reason to do, with natural facts, their normativity is lost” (Dancy 2006: 132). To show this, Dancy invokes his explicit-answers account of normativity. He suggests that natural facts never provide explicit and direct answers to the question what to do. A natural fact F might be normatively relevant. It might make it the case that there is a reason for doing something, or make it the case that some action would be right, but if so, these further normative facts about F are not identical to the fact F. Dancy claims, moreover, that metafacts of this kind are not natural facts. For, he says, “there is a difference in subject matter … between the fact that another fact is of practical relevance and a fact that is of practical relevance” (Dancy 2006: 139). The key question is why metafacts about the practical relevance of other facts cannot be natural facts. Dancy suggests that the normativity of a fact depends on its subject matter and he contends that no natural fact has the same subject matter as a fact about that fact’s practical relevance. For example, he says, “it just isn’t true … that the fact that this action maximizes welfare (say) has the same subject matter as the fact that that fact would make the action right” (Dancy 2006: 140). This is correct, of course, but it is not to the point. Utilitarian Naturalism identifies rightness with the property of maximizing happiness. In this case, the question to ask is why it cannot be a natural fact that the fact that this action maximizes happiness would make the action right. Dancy remarks that it does not “look as if” we could have empirical knowledge or justified belief of such facts (Dancy 2006: 139). But he concedes that this is not an argument. He also concedes that some metafacts are natural. It is, for example, a natural fact that the fact that putting my hand on a hot stove would cause me pain deters me from putting my hand on a hot stove.32 If there can be natural metafacts, why cannot normative metafacts be natural? Dancy says, “What is driving the nonnaturalists here is the thought that to say that a feature is practically relevant is to make a different sort of claim from any claim that does not explicitly mention practical relevance.” He adds, “There is a vital distinction between direct and indirect ways of answering the practical question what to do” (Dancy 2006:
Dancy (2006: 142). I have changed the example.
32
Normativity and reasons
49
141). The trouble is, however, that whether a statement directly answers a question depends on how the question and answer are formulated, and not merely on what fact the answer cites. For example, if the question is whether this glass contains water, I have not directly and explicitly answered the question if I say that it contains “eau.” Yet this does not show that the fact that it contains eau is distinct from the fact that it contains water. Similarly, if I ask whether doing A would be right, you may not have directly and explicitly answered the question if you say that doing A would maximize happiness. Yet it does not follow that the fact that doing A would maximize happiness is distinct from the fact that doing A would be right. 2 .8 T h e T r i v i a l i t y Obj e c t ion Parfit offers three closely related arguments that turn on the idea of triviality, so this is really a family of arguments. First, he contends that if normative naturalism were true, then certain claims that we know quite well to be substantive and normative would instead be trivial. Second, he contends that if normative naturalism were true, it would be trivial. It can be substantive and interesting only if it is false. Given this, third, normative claims could not state facts that are both normative and natural. Parfit calls this the Fact-Stating Argument. This third argument resembles Dancy’s argument from normativity, which I discussed in the preceding section. Dancy also offers versions of the first two arguments. The main idea lying behind the first argument is stated by Dancy. He suggests that all forms of naturalism begin with some property that seems obviously to be right-making and then identify rightness with that property. This, he claims, trivializes the claim that actions with this property are right. Or they take a reason-giving property and identify it with rightness, thereby trivializing the claim that this property gives us a reason to do what is right (Dancy 2006: 140). Or in some similar way they trivialize some claim that is obviously non-trivial. According to Utilitarian Naturalism, for instance, rightness is identical to the property of maximizing happiness, so the proposition that actions that maximize happiness are right is equivalent to the proposition that actions that maximize happiness maximize happiness, which is obviously trivial. As Dancy says, however, “we all know perfectly well” that the original proposition is not trivial but that it is, instead, morally substantive and controversial (Dancy 2006: 131; see 131–32). The argument can be generalized. Every version of normative naturalism will hold that each normative predicate “M”
50
dav i d c opp
ascribes the same property as is ascribed by some expression “N” that evidently ascribes a natural property. It will therefore imply that statements expressed by sentences of the form “Anything that is N is M” are normatively trivial. And one might claim, again, that in each case we can see that these statements are not trivial. Parfit works with the example of Utilitarian Naturalism because he thinks that there is nothing special about this theory and that whatever problems it faces are faced by every other version of moral naturalism (Parfit 2011: v. II, 341). He formulates the argument as follows.33 Take Utilitarian Naturalism to be the doctrine that the property of being what one ought to do is identical to the property of being the act one could do that would maximize happiness. Now consider the claim that the act that would maximize happiness is what P ought to do. Parfit calls this claim “(A).” Intuitively, as Parfit says, “(A) is a substantive normative claim, which might state a positive substantive normative fact.” The trouble is that if Utilitarian Naturalism were true, “(A) would be only another way of stating the trivial fact that, when some act would maximize happiness, this act would maximize happiness.” Yet (A) obviously is not trivial in this way. So if (A) is a substantive moral claim, Utilitarian Naturalism is not true.34 A Utilitarian Naturalist presumably would reply that (A) is not a substantive normative claim. She holds that we ought to do whatever would maximize happiness because these things would maximize happiness, but in saying this she would intend “because” to be understood in the reductive sense rather than the substantive normative sense. This reply is not satisfying, however, for, as Parfit says, (A) is intuitively a substantive normative claim. I agree with Parfit that Utilitarian Naturalism is implausible. But we cannot safely generalize from the implausibility of Utilitarian Naturalism to the implausibility of normative naturalism in general. 33 See Parfit (2011: v. II, 343–44). The material I discuss in this paragraph as well as the quotations in the text are all found on pp. 343–44. 34 Unfortunately, Parfit formulates the argument in a question-begging way. For he stipulates that normative claims are “positive when they state or imply that, when something has certain natural properties, this thing has some other, different, normative property.” He then claims that (A) states a positive normative fact and objects that if Utilitarian Naturalism were true, (A) could not state such a fact. See Parfit 2011: v. II, 343. But given Parfit’s stipulation, it is question-begging in the context to claim that (A) states a positive normative fact. For Utilitarian Naturalists deny that the property of maximizing happiness is different from the property of being what we ought to do. Parfit cannot claim that (A) states a positive normative fact unless he has already shown that Utilitarian Naturalism is false.
Normativity and reasons
51
Dancy concedes that the argument only works against certain forms of naturalism.35 It works against Utilitarian Naturalism because the proposition that actions that maximize happiness are right seems intuitively to be an implicit account of what can make an action be right (in the substantive normative sense). In general, Dancy thinks, the argument works against theories that propose to identify moral properties with “the sort of property that we would happily invoke in saying that an action is right because it has such a property,” where “because” is used in the substantive normative sense. The argument fails to get purchase, however, against other sorts of proposal. For example, Dancy thinks the argument does not get purchase against Frank Jackson’s “outlandish” proposal, according to which wrongness is identical to a “vast disjunctive descriptive property.” For Jackson, very roughly and intuitively, wrongness consists in the disjunction all of the “ways” in which an action could be wrong, such as, crudely, being a lie or being a breaking of a promise or being a case of child abuse.36 Consider then the claim that actions that have this vast disjunctive property are wrong. This is not an ordinary substantive normative claim, and Jackson can maintain that it is an implicit property-identity claim. So the argument does not get purchase against his proposal. On my society-centered proposal, to be wrong is (roughly) to be precluded by the moral code the currency of which in society would do most to ameliorate the problem of sociality. Again, the claim that actions that have this property are wrong is not an ordinary substantive normative claim. I maintain that it is an implicit property-identity claim. The argument does not get purchase against my proposal. The second argument from triviality is no more successful. Reductive forms of naturalism propose to identify each normative property with some specific natural property. They propose, for instance, that wrongness is identical to some natural property N. The trouble is that if this claim is true, it is trivial, for it is equivalent to the claim that N is identical to N. It appears, then, that if normative naturalism were true, its central property-identity claims would be trivial and not worth debating (see Parfit 2011: v. II, 338–39). True property-identity claims are trivial, says Parfit, because every property is identical to itself. But Parfit points out that some true property-identity claims are expressed in ways that enable them also to 35 Dancy (2006: 132). This paragraph follows Dancy’s discussion and the quotations all come from p. 132. 36 Jackson (1998: 129–51). See also Jackson and Pettit (1995). I discuss Jackson’s proposal in Copp (2007: ch. 4 [“Moral Necessities in a Contingent World”], 135–37).
52
dav i d c opp
imply significant information. For example, the thesis that the property of being water is identical to the property of being composed of H2O molecules gives us significant information about the nature of the property of being water and about the chemical composition of water.37 A normative naturalist can respond to Parfit’s argument by saying something similar about her key property-identity claims. She can say that her key claims supply information about the nature of the normative properties and not merely the trivial information about each of them that it is identical to itself. Call this the information response. The Utilitarian Naturalist can respond to Parfit’s argument, for instance, by saying that her key property-identity claim tells us that the property of being what we ought to do has, as constituents, the property of being happiness and the relation of being a maximization of something and that it relates these constituents in a specified way. Jackson can respond that his identification of wrongness with a vast disjunctive property tells us something surprising about the nature of wrongness. And I can respond that the central propertyidentification claims made by the society-centered theory tell us something significant about the nature of the moral properties. We can use Jeffrey King’s theory of “structured propositions” to help clarify the information response. On King’s approach, there is room to claim that, even though water is H2O, the proposition that water is H20 is distinct from the trivial proposition that water is water. Of course, it would be beyond the scope of this paper to explain the idea in detail.38 The point, however, is that, on King’s approach, the naturalist can say that her thesis that wrongness is identical to the natural property N is distinct from the trivial claim that N is identical to N.39 Unfortunately, however, the theory of structured propositions raises the awkward technical issue whether the naturalist is committed to something I call the strong fact-identification thesis. The theory of structured propositions provides what we might call a fine-grained criterion for individuating propositions. If we assume that a fact is a true proposition, as I do for present purposes, this criterion for individuating propositions 37 See Parfit (2011: v. II, 334–36). At pp. 336–37, Parfit says there are “two senses in which different claims may state the same fact,” the “referential sense” and the “informational sense.” I find this puzzling, so I avoid this formulation. 38 According to King’s theory, the proposition that water is H2O is distinct from the proposition that H2O is H2O because they have different structures. See King (1998). I have suggested before that a normative naturalist could exploit King’s theory. See Copp (2007: ch. 4 [“Moral Necessities in a Contingent World”], 129, n. 37). 39 The theory of structured propositions also undermines the first argument from triviality in both Dancy’s formulation and Parfit’s.
Normativity and reasons
53
commits us to a similarly fine-grained criterion for individuating facts.40 It commits us to a fine-grained account of facts. On such an account, there is room to distinguish the fact that wrongness is identical to N from the trivial fact that N is identical to N. The trouble is that, on such an account, if we do distinguish these facts, then we are committed to distinguishing the fact that, say, capital punishment is wrong from the natural fact that capital punishment is N. It may seem to follow that the fine-gained account of facts is incompatible with normative naturalism. If so, the theory of structured propositions is problematic for a normative naturalist. This worry turns on a mistaken idea about what a naturalist must claim. A naturalist must identify normative properties with natural properties. She must claim that, for each normative property M, there is a natural property N such that to be M is to be N. But it is a tricky matter whether this means that naturalism must accept the strong fact-identification thesis, the thesis that, if M is identical to N, then each fact to the effect that something is M is identical to the fact that that thing is N. Suppose the naturalist says that to be wrong is to be N. She must then say that capital punishment is wrong if and only if it is N. Furthermore, she must say that the proposition that capital punishment is wrong ascribes to capital punishment the natural property N and no other property. The fact that capital punishment is wrong is in this sense a natural fact. Nevertheless, on the fine-grained account of facts as I understand it, even if the naturalist’s theory is true, the fact that capital punishment is wrong may be distinct from the fact that capital punishment is N, and the fact that wrongness is N may be distinct from the fact that N is N. Similarly, even if water is H2O, this fact is distinct from the fact that H2O is H2O; and the fact that this glass contains water is distinct from the fact that it contains H2O. In short, the naturalist must claim that normative properties are natural properties and that normative facts are natural facts, but on a fine-grained account of the nature of facts, the naturalist is not committed to the strong fact-identification thesis. This is a subtlety that I have ignored to this point in the paper. Whether we think the naturalist is committed to the strong fact-identification thesis clearly should depend on what we think facts are. According to King’s account of structured propositions, the proposition that water is H2O is distinct from the proposition that H2O is H2O. So if a fact is a true proposition (or if it has a relevantly similar structure to the proposition it would make true), then the fact that water is H2O is distinct from the fact that H2O is H2O. See King (1998). 40
54
dav i d c opp
Parfit’s final argument from triviality, the Fact-Stating Argument, seems to depend on taking the naturalist to be committed to the strong fact-identification thesis. “According to Non-Analytical Naturalists,” Parfit says, any true normative claim states a fact that “could also be stated by some other non-normative, naturalistic claim.” The trouble is that this non-normative claim “could not state a normative fact.” It follows, then, that the fact stated by the normative claim is not a normative fact. This is absurd. Therefore, unless one of the argument’s premises can be denied, Non-Analytic Naturalism is false (Parfit 2011: v. II, 338–39). For now, let us assume that the naturalist is committed to the strong fact-identification thesis. Then, as Parfit recognizes, the entire weight of the argument falls on the premise that a “non-normative claim could not state a normative fact,” which is premise (6) in the argument as Parfit presents it. A naturalist would certainly deny this premise. To support it, Parfit turns to Subjectivism about Reason, both to illustrate the argument and because, on the reason-implying conception of normativity, the issue whether (6) is true is the issue whether a non-normative claim can state a fact that is reason-implying. Consider a simplified version of Subjectivism about Reason according to which the property of being a decisive reason for a person P to do A is identical to the property of being such that doing A is what would do most to fulfill P’s desires. Now suppose it is true that (S), P has decisive reasons to do A. According to Subjectivism about Reason, Parfit suggests, the fact stated by (S) could also be stated by the naturalistic claim (U), doing A is what would most fulfill P’s desires. Parfit argues that (U) could not state a normative fact, which is of course implied by premise (6) of the original argument. For if (U) states a normative fact, it presumably states the same fact as (S), and then there must be no “distinct” normative property “of being what we have decisive reasons to do.” And if there is no such distinct property, Parfit contends, there is no sense in which the fact stated by (S) is normative, so there is no sense in which the fact stated by (U) is normative. Subjectivism about Reason is therefore false, for it holds that (U) states a fact that is both normative and natural. Parfit concludes, “Similar arguments apply to all other forms of Non-Analytical Naturalism” (2011: v. II, 341). At root, the argument rests on the claim that the fact that we have decisive reason to do something A is normative only if the property of being a decisive reason for us to do A is distinct from the property of being such that doing A is what would do most to fulfill our desires. This
Normativity and reasons
55
claim is question-begging, however. It will seem plausible if we think that no natural property can be normative in the reason-implying sense. For if we think this, then it may seem that no property at all can be normative in the reason-implying sense unless the property of being a reason is distinct from any natural property. No naturalist could accept this reasoning, however. According to Subjectivism about Reason, moreover, the fact that we have decisive reason to do something is obviously normative in the reason-implying sense, for it entails that we have decisive reason to do something. Moreover, according to the theory, the fact that doing A is what would most fulfill our desires is normative in the reason-implying sense because it implies that we have decisive reason to do A. It implies this because, on the theory, the property of being what would do most to fulfill our desires is the property of being what we have decisive reason to do. It seems to me, then, that the Fact-Stating Argument is unsuccessful. It rests on a premise that will seem persuasive only to those who have already rejected normative naturalism. 2 .9 T h e S of t N at u r a l i s t ’s Di l e m m a A “Hard Naturalist” would claim that we have no need to use normative language in making normative claims for we could express all normative facts in purely naturalistic language. The moral terms could be removed from our language and the moral concepts could be removed from our thinking without any loss in expressive power. A “Soft Naturalist” would resist this conclusion and claim that we still might have good reason to use normative language in making normative claims. Parfit objects, however, that all naturalists hold that normative claims are intended to state facts. Moreover, on this view, there is reason to use normative language only if there are normative facts that cannot be expressed in naturalistic terms. But, of course, naturalists deny this. Parfit concludes that Soft Naturalism is incoherent. A naturalist is committed to the Hard Naturalist view (Parfit 2011: v. II, 364–66). Given this, and given that, he claims, Hard Naturalism implies that normative claims are of no importance, Parfit contends that “Naturalism is close to Nihilism” (v. II, 368). I think this is a mistake. Even if normative facts are natural facts, some natural facts are more important than others, so normative facts may be important. It is more important that my food has been poisoned than that my shoes are brown. A naturalist would say that this itself is a natural fact. A naturalist can claim that normative facts are in general highly
56
dav i d c opp
important. This can be so even if there is no need to use normative language in expressing such facts. Moreover, Soft Naturalism seems to me to be defensible even if Parfit is correct that naturalism implies that any information that can be given in making a normative claim can also be given in making some purely naturalistic claim. A normative naturalist can claim that when we make moral claims by using normative moral language we do more than merely communicate information. On the kind of realist-expressivism that I have proposed, a normative moral claim both expresses a normative moral proposition and expresses a relevant conative attitude.41 For example, I claim, a person who says that torture is wrong states a normative moral proposition, but she also expresses disapproval of wrong acts, including torture. This view can explain why we have reason to use normative language even if normative naturalism is true. We might also have cognitive reasons to continue to use normative moral language, since, I believe, the naturalistic truth conditions of moral claims are enormously complex. We might have good reason to continue to make moral claims and no reason to fret over stating them in naturalistic terms, given the difficulty of getting their naturalistic truth conditions exactly right. Hence, I think, the combination of the society-centered view with realist-expressivism is friendly to soft naturalism. 2 .10 C onc l us ion Parfit’s arguments are subtle, complex, and sophisticated. If I am correct, however, none of them is successful, and one might wonder where they go wrong. There are perhaps three central mistakes. First, Parfit unfortunately thinks that naturalists have no option but to attempt to explain normativity in terms of motivational force. I think this is a mistake.42 There are naturalistic theories, such as my own pluralist-teleology, that do not aim to explain normativity in terms of motivational force. Second, Parfit thinks that normative facts are reason-implying, but the only naturalistic account of reasons that he takes seriously is Subjectivism about Practical Reasons. He does not see that there can be other naturalistic accounts of reasons, and so other naturalistic accounts of normativity. Third, Parfit does not seem to see what information would be conveyed 41 See Copp (2007: ch. 5 [“Realist Expressivism: A Neglected Option for Moral Realism”], 153–202); and Copp (2008). 42 I argue that it is a mistake in Copp (2007: ch. 10 [“Moral Naturalism and Three Grades of Normativity”]).
Normativity and reasons
57
by a naturalist’s property-identity claims. He suggested analogies with scientific property-identity claims but did not, I think, see how the normative naturalist’s property-identity claims could be similarly informative and interesting. There is much that is of great value in Parfit’s examination of normative naturalism, and much that I have not discussed. I have criticized his arguments against normative naturalism, but I have not looked at his rebuttals of the arguments that naturalists have used in attempting to support naturalism. These rebuttals are important. Moreover, there are other arguments against naturalism that I have not discussed here. They would need to be addressed in a full defense of normative naturalism.43 Parfit thinks that if normative naturalism is true, then many philosophers have wasted much of their professional lives. He worries that Sidgwick, Ross, and he himself may have wasted much of their professional lives (Parfit 2011: v. II, 303–305 and 367). He is completely wrong about this, it seems to me. In debates about which things are right and which are wrong, in debates about what we have reason to do, and in debates about normative moral theory, it is irrelevant whether normative properties are natural properties. These debates can be fruitful and important even if normative naturalism is true. Much of philosophical importance is at stake in the dispute between normative naturalism and non-naturalism. Naturalists aim to provide a substantive and philosophically satisfying account of what it is for a property and a fact to be normative.44 They aim as well to provide a substantive and philosophically satisfying account of the normative properties, including the moral properties and the property of being a reason. Naturalists hope that such accounts can be given, and they hold that the desired accounts will be naturalistic. Non-naturalists are in the pessimistic business of arguing that this project cannot succeed. The project is, however, of undeniable philosophical importance. 43 I discuss several such arguments in Copp (2007). For instance, I discuss the “moral twin earth argument” in Copp (2007: ch. 6 [“Milk, Honey, and the Good Life on Moral Twin Earth”], 203–29; and ch. 7 [“Referring to Moral Properties: Moral Twin Earth, Again”], 230–45). 44 Here I set aside non-reductive naturalism.
ch apter 3
Naturalism: feel the width Roger Crisp
1. Certain claims we make, such as that someone has a reason to act in some way, that some action was wrong or cruel, or that a painting is great or poignant, may be taken to imply normative or evaluative facts or properties. A question then arises concerning the metaphysical status of these facts or properties. Many philosophers have been inclined towards socalled “naturalistic” answers to this question. There are many kinds of naturalism and hence of non-naturalism. Often the gulf between the two positions is clearly unbridgeable. Consider for example a naturalist who advocates an extreme form of normative and evaluative nihilism: there are no normative evaluative facts or properties, and we should stop talking as if there were, since such talk can be debunked by, say, some version of evolutionary expressivism. None of those who have described themselves as non-naturalists could be brought to accept such a view. But there are more capacious forms of naturalism, and my aim in this paper is to suggest that, if we uncover what motivates naturalism and its denial, we can find a version of naturalism broad enough to be plausible to those initially attracted to non-naturalism. 2. Before discussing those motivations, let me first narrow the focus a little. First, for the moment, I shall restrict my discussion to normative claims understood as those concerning reasons for action. This is merely to avoid the issue of the relation between the normative and the evaluative. (In fact, I believe that the broadly naturalistic view I shall propose can deal with evaluative as well as normative claims.) I owe a special debt of gratitude to Derek Parfit for extremely helpful comments on and discussion of earlier drafts of this paper. I am very grateful also to the editors of this volume, to Robert Audi, Jonathan Dancy, Daniel Robinson, Daniel Star, and Nick Zangwill, and to an audience at the Oxford/Scandinavian Conference on Moral Philosophy in Oxford, June 2010. The paper was completed during my tenure of the Findlay Visiting Professorship in the Department of Philosophy, Boston University. I would like to thank the department for its intellectual, social, and practical support.
58
Naturalism: feel the width
59
Second, note that above I spoke of both normative facts and normative properties. If there were consensus on these in philosophy, that might not be a problem. But of course there is a great deal of disagreement about both. For that reason, it would be preferable if possible to restrict the debate to one or other of these notions. I suggest that we construe the debate – initially at least – as concerning properties rather than facts.1 This is because facts are in a sense posterior to properties (see Fine 1982: 53). Facts concern properties (and of course their objects, relations, and so on), so any conception of facts requires an account of properties. This is not true of properties, though of course there are properties of facts. Third, let me leave on the table only those forms of naturalism that allow that there are normative properties – that is, those that are consistent with some broadly realistic conception of these properties. Finally, let me confine the main focus of the discussion to some particular action in the real world, so as not to be sidetracked into broader metaphysical questions about possible worlds, abstract objects, and so on.2 Since I shall often assume, just to simplify the argument, the correctness of act utilitarianism, let me take that action to be one of maximizing the balance of overall happiness. 3. How, then, are we to understand naturalism, and the motivation for it? As one might expect, given the variety of forms of naturalism, there is no consensus on this in the literature (see e.g., Ridge 2008: Introduction). So let me begin with an outline of what I take the central issue to be. According to naturalists as I understand them, properties are metaphysically respectable to the extent that they are, in some sense to be specified, part of the scientific conception of the world (see Moore 1903: 40). This raises the further question of which sciences are involved in constructing that conception. I suggest we take a fairly ecumenical approach here, allowing in psychology, anthropology, economics, and sociology, for example. Non-naturalism, then, is the view that normative properties cannot be seen as part of the scientific conception of the world, even in an ecumenical form. Note that this conception of the debate between naturalists and non-naturalists is, in a sense, epistemic, referring as it does to scientific enquiry. But underlying the epistemology is a metaphysical view of the objects of study of normative ethics, on the one hand, and of the sciences, 1 It may of course be possible to restate arguments concerning facts in terms of properties. 2 Possible worlds and other metaphysical notions will come into view, but only in so far as we are seeking to elucidate the properties of our paradigmatic actual action.
60
ro g e r c r i sp
on the other. As Sidgwick puts it on the first two pages of The Methods of Ethics, “The student of Ethics seeks to attain systematic and precise general knowledge of what ought to be … [while] it is widely thought that a Science must necessarily have some department of actual existence for its subject-matter.”3 According to naturalists, “what ought to be” – the normative – bears a relation to the non-normative – what Sidgwick calls “departments of actual existence” – such that it can be seen as in some important sense “anchored” in the non-normative.4 One obvious such relation would be identity, but there are of course others, and it is by spelling out such relations that one can clarify further the nature of naturalism and non-naturalism. 4. One version of naturalism – the kind G. E. Moore had primarily in his sights in Principia Ethica – is analytic. On this view, consistency with the scientific worldview is guaranteed semantically, since terms for normative properties are claimed to mean the same as terms for certain natural properties. Recently, Frank Jackson has revived this position, arguing that it is analytic that an action is right if it has whatever descriptive properties would be identified with rightness by “folk morality … after it has been exposed to debate and critical reflection” (Jackson 1998: 133). It would be hard to persuade a non-naturalist that any such view doesn’t fall prey to a Moorean ‘open question’ argument. 5. More common in recent decades have been forms of non-analytic (NA-) naturalism which seek to respond to the apparent failure of Moore’s socalled “semantic test of properties” to show that goodness is not the same as some natural property (Brink 1989: 162). One of the most commonly used analogies here is due to Putnam (1967). “Heat” and “molecular kinetic energy” clearly have different meanings. But this does not entitle us to argue that, since it is an open question whether heat is molecular kinetic energy, heat is not molecular kinetic energy.5 The question is closed, since we have discovered that these concepts refer to the same property. So if we assume the truth of act utilitarianism, the NA-naturalist may claim that, though the concept of maximizing happiness is different from that of rightness (just as the concept of heat is different from that of molecular kinetic 3 Sidgwick (1907: 1–2). On “departments of actual existence,” note Sidgwick’s claim elsewhere that “the difference between ‘what is’ and ‘what ought to be’ may be reasonably held to be relatively reduced, when we contemplate, along with both, various forms of ‘what might be’ or ‘might have been’, as objects of more or less coherent thought for scientific or artistic purposes” (2000: 62). 4 The term is Audi’s; see Audi (this volume, Chapter 4). 5 Another standard analogy is the identity of water and H20. See Putnam (1975).
Naturalism: feel the width
61
energy), the property of “maximizing happiness” (which we may take to be natural) is the same property as “being right” (which is normative). NA-naturalism is likely to be more congenial to non-naturalists than analytic naturalism, since it avoids standard versions of the open question argument. But non-naturalists will want to argue that the difference between normative and natural properties is too great for an identity claim to be plausible. Consider, for example, Parfit’s normativity objection (Parfit 2011: v. II, 324–27). In the case of heat, a debate between those who held it to be some kind of basic substance and those who held it to be molecular kinetic energy would have made sense. But the view that heat is, say, a cabbage could not have been taken seriously. Parfit’s argument relies on the example of the “Burning Hotel,” in which, unless you jump into the canal, you will die.6 An NA-naturalist and a non-naturalist may agree that your jumping has the property of being what you ought to do. But the naturalist may suggest that this property is the same as some natural property, such as that of being what will do most to fulfill your fully informed desires. Parfit’s response is that we have two entirely different properties here. Claiming that the property of being what you ought to do is the same as that of being what will do most to fulfill your fully informed desires is like claiming that heat is a cabbage. And there is a good deal of plausibility in his claim. On the face of it, claims about what is right or what we ought or have reason to do are in a different category from claims that appear merely to describe certain natural states of affairs; and the most straightforward explanation of that is that these claims are indeed about different kinds of property. That is not the case with heat and molecular kinetic energy, where there is an accepted body of physical theory and evidence which explains their identity. But, Parfit can claim, there is no such body of theory and evidence in the case of normative and natural properties, and so we should accept the obvious explanation of the deep conceptual difference here – that it mirrors a real difference in reality. But there is much to be said for the opposing view. There is first an internal objection to Parfit’s analogy. Perhaps the claim that the property of being what one ought to do is the same as the property of being what will do most to fulfill one’s fully informed desires is remarkably counter-intuitive. But some remarkably counter-intuitive claims, such as that space is curved, turn out to be widely accepted as true. Further, many will anyway not find the naturalist identity claim here anywhere Parfit’s argument is about facts, but can be restated in terms of properties
6
62
ro g e r c r i sp
near as counter-intuitive as the claim that heat is a cabbage (or that space is curved). Indeed the “in virtue of” relation which any plausible theory must allow for in some sense or other might be taken to count in favor of there being some close relation between what ought to be done and certain natural properties. Even the staunchest non-naturalist will have to agree that, to use Gilbert Harman’s example (1977: ch. 1), setting fire to a cat is wrong at least partly in virtue of the pain caused to the cat.7 We appear to have a standoff. Parfit claims that normative and natural properties are too different from one another to be identified, while the naturalists deny that. Nor – and here is another difference between the normative case and debates about the nature of water – is it clear where the resolution to this debate is going to come from. All the facts are, as it were, in. There is nothing especially puzzling about the property of maximizing happiness, such that some new theory of it could demonstrate its identity (or lack of identity) with rightness. At this point, then, it is worth asking about the prospects of some kind of metaphysical rapprochement between naturalism and non-naturalism. 6. Two different pressures lead naturalism and non-naturalism away from one another. On the one hand, the naturalist wishes to anchor normative properties within the scientific conception of the world, and to avoid the attribution of properties which, by definition as non-natural, lie (mysteriously) beyond that conception. On the other hand, the non-naturalist, noting the apparently highly significant difference between natural or “descriptive” properties and normative properties, marks that difference by isolating normativity from science. According to NA-naturalism in the metaphysically reductive form discussed above, the property of rightness is identical to the property of, say, maximizing happiness in the same way that the property of water is the same as the property of H2O. In the case of both water and rightness, there is really only one property. I now want to suggest that the use of the “water” and “heat” analogies by NA-naturalists, though they well demonstrate the problems with the semantic test for properties underlying the open-question argument as it is usually construed, may have led naturalists into defending a stronger claim about the relation between natural and normative properties than they need.8 What the naturalist wishes is a For an interesting argument that reductionist claims in ethics, especially those concerning reasons, avoid the “propagating implications” of a claim such as “God is a nuclear force,” see Schroeder (2005). 8 Sturgeon, for example, appears to believe that physicalism about the mind must be committed to the reductive claim that mental properties are physical (Sturgeon 2003: 537). 7
Naturalism: feel the width
63
model in which normative properties are not (mysteriously) isolated from the scientific conception of the world, and there are ways of accomplishing that without reductive identity claims.9 In recent decades in philosophy of mind, the notion of supervenience has supplanted that of reduction.10 The relation between mental properties and physical properties so conceived may well provide a more fruitful model for the relation between normative and natural properties than the “water” and “heat” analogies.11 This is for two reasons. First, the properties of being water and of being H2O are both, on the face of it, physical, whereas normative properties and natural properties are by most parties in this debate placed initially at least in two rather different categories.12 So the analogy with mental properties raises the possibility that there may be a naturalistically acceptable form of distinctness which also satisfies the non-naturalist. Second, it may be that normative properties such as rightness are multiply realizable, at least at some levels. Physicalism is usually understood in contemporary philosophy as the view that mental states are identical with or supervenient on physical states (see e.g., Horgan 1993: 556). Like physicalism, NA-naturalism and nonnaturalism as I have been construing them are primarily metaphysical positions. NA-naturalism’s reductionism is analogous to type physicalism, according to which: T P : For every mental property F, there is some physical property G13 such that F = G.
According to NA-naturalism: N A N: For every normative property P, there is some natural property Q such that P = Q.
9 For a brief survey, see Finlay (2010: 335). 10 This is not to imply that reductionists have eschewed the notion of supervenience (indeed some philosophers have seen supervenience itself as a reductive notion – e.g., Dupré [1993: 96–97]). 11 See e.g., Shafer-Landau (2003: 73–87), though Shafer-Landau unnecessarily combines supervenience with the claim that normative properties are constituted by natural (or “descriptive”) ones. If constitution here is equivalent to identity, then Shafer-Landau would be better described as an NA-naturalist than as a non-naturalist. And if it is not, then the notion of constitution is perhaps best avoided. For a more skeptical discussion of supervenience in the metaethical context, focused on Shafer-Landau, see Ridge (2007). 12 See Dancy (2006: 139). In the case of water, we have explanatory reduction. It is not clear that we have this with e.g., pain and C-fibers (see Levine 1983) or indeed with normativity and the natural. 13 Or set of physical properties G1, G2 … Gn. I assume this disjunction, mutatis mutandis, for each characterization below.
64
ro g e r c r i sp
One theory of mind entirely at odds with physicalism of any kind is strong dualism:14 S D: Mental properties and physical properties are distinct, in that there is no relation of supervenience between them. There can be differences in mental properties with no difference in underlying physical properties.
An analogous position in metaethics we might call Strong Dualist NonNaturalism: S DN N: Normative properties and natural properties are distinct, in that there is no relation of supervenience between them. There can be differences in normative properties with no difference in underlying natural properties.
This is not Parfit’s position (or indeed that of anyone else I can think of). Parfit accepts that normative properties supervene on natural properties. Indeed he almost certainly accepts what Kim (1984) calls “strong supervenience” – that is, that any supervenience relation in this world carries across to all possible worlds.15 If act utilitarianism is true, then in this world the property of being what we ought to do supervenes on the natural property of maximizing happiness overall. And this relation is a necessary one, in the sense that there is no possible world in which some action has the property of maximizing happiness but lacks the property of being what the agent ought to do, or in which some action has the property of being what agents ought to do but lacks the property of maximizing happiness. That is part of what is involved in an action’s being right solely in virtue of its maximizing happiness. In philosophy of mind, as I have said, accepting supervenience (along with the denial of dualism) is usually said to be sufficient to make one a physicalist. Physicalism need not be reductionist, nor of course need it be analytic.16 According to Non-Analytic Non-Reductive Physicalism: N A N R P : Terms for mental properties and physical properties have different senses. Mental properties and physical properties are distinct; but for every mental property F, there is some physical property G such that F strongly supervenes on G. 14 Of course traditional dualism has concerned substances rather than properties. I am calling this version of dualism “strong” because versions of physicalism which allow for supervenience without reduction are often called versions of “property dualism,” and because any form of dualism which denies supervenience is clearly in that respect significantly stronger than dualisms that do not. 15 He can therefore avoid Blackburn’s request for an explanation of the “ban on mixed worlds” (Blackburn 1985). 16 Analytic positions in philosophy of mind have almost become defunct, since the seminal paper by Smart (1959).
Naturalism: feel the width
65
The analogous position in metaethics will be a Non-Reductive form of Non-Analytic Naturalism: N A N R N: Terms for normative properties and natural properties have different senses. Normative properties and natural properties are distinct; but for every normative property P, there is some natural property Q such that P strongly supervenes on Q.
This view is consistent with what Parfit calls Wide Naturalism: W N: Normative properties would be natural properties, even if such properties were irreducibly normative because these properties could not possibly be fully described in non-normative naturalistic terms.17
Parfit is not inclined strongly to object to WN, though he does say that offering a criterion for naturalness through reference to natural science is “too vague”. But what I am suggesting is that a plausible conception of the natural is available if we spell out the relation between the normative and the natural (understood in terms of natural science) using the notion of strong supervenience. Indeed, since Parfit denies SDNN and accepts strong supervenience, might we not characterize his view as NANRN rather than non-naturalism, thus closing the gap between naturalism and non-naturalism? This would be too swift. In recent years, an option in philosophy of mind has been discussed which is – on the basis of the distinctions above – neither straightforwardly physicalist nor straightforwardly dualist. This is emergentism, which in effect adds a “significant distinctness” claim to NANRP: E : Terms for mental properties and physical properties have different senses. Mental properties and physical properties are significantly distinct; but for every mental property F, there is some physical property G such that F strongly supervenes on G.
Because of its significant distinctness claim, E is often seen as a form of non-physicalism. Given Parfit’s stress on the distinctness of normative and natural properties, his position might be seen as an analogous form of non-naturalism, which we might call Normative Emergentism: N E : Terms for normative properties and natural properties have different senses. Normative properties and natural properties are significantly distinct; but
17 Parfit’s own version of WN (2011: v. II, 306) concerns facts rather than properties. I have moved the comma forward from its position after “irreducibly normative” in Parfit’s statement.
66
ro g e r c r i sp for every normative property P, there is some natural property Q such that P strongly supervenes on Q.
NE is essentially steering a course between any form of naturalism, on the one hand, and dualistic non-naturalism, on the other, just as in philosophy of mind emergentism aims to combine some of the attractions of both dualism (the mental does seem very different from the physical) and physicalism (it is very hard to deny supervenience). But it is important to remember that in the case of the mind, non-reductive physicalists already accept that mental properties and physical properties are distinct. They are, that is to say, property dualists. Emergentism rests on further ideas, such as the notion that properties are emergent if their “behavior” cannot be systematically predicted on the basis of a complete knowledge of the lower-level properties and relevant laws at that “level.” Exactly how to characterize emergence is a matter of lively debate in contemporary philosophy of mind. One view, indeed, is that we should think of emergence as strong synchronic supervenience (e.g., McLaughlin [1997]; cf. O’Connor and Wong [2005: 664]). Macdonald and Macdonald (2010: 12–13) identify three different lines of argument: (i) emergence requires not only numerical but also mereological distinctness; (ii) emergent properties are only nomologically and not metaphysically dependent on physical properties; (iii) “strong” emergence requires direct “downward” causation, that is, causation irreducible to causation at the base level. These debates are interesting and important, but at present it is not clear whether they are especially relevant to the metaethical debate between naturalism and non-naturalism. For the present, it would be wisest for both naturalists and those now identifying themselves as non-naturalists to take NANRN as a working hypothesis, and then seek to explain further the in virtue of relation between natural and normative properties.18 I suspect that this debate itself will be sustainable within a broadly scientific conception of the world with which both naturalists and (current) non-naturalists can be content. For what more can a non-naturalist want, metaphysically speaking, than that normative and natural properties be clearly metaphysically distinct, along with a form of strong supervenience that can be captured in ethical principles that state necessary truths? And what has a naturalist to fear from normative properties that are securely anchored in natural 18 Normative emergentists might also use this notion to dispel the alleged aura of mystery concerning the emergence of normative from non-normative properties in their account.
Naturalism: feel the width
67
properties through not only these relations of strong supervenience, but also the dependence of normative on natural properties?19 7. One rather obvious fear a naturalist might have is of course the introduction into their ontology of non-natural properties! The main advantage of reductive NA-naturalism may be said to be lost through any form of “property dualism” which allows normative properties to stand “on their own.” One worry here is causation. Views such as NANRP face questions concerning what if any causal role might be played by mental properties. But if those questions can be answered, it might seem an even greater challenge to explain the causal efficacy of normative properties. It cannot be denied that there are difficult issues here. But there are several options available to proponents of NANRN. One would be to allow that normative facts might cause certain natural facts. Parfit (2011: v. II, 306–307) suggests an evaluative example in which the world is as it is because its being that way is the best way for it to be. Or we might explain some set of physical events in terms of their being the result of a person’s noticing that she has a reason to avoid suffering by jumping from a window. A second option would be to continue to treat normative and mental properties as analogous, and perhaps allow both to be epiphenomenal. A third would be to accept that there may be some disanalogies between normative and mental properties, but to refuse to accept that a property, to be metaphysically respectable, must play some causal role. 8. Another worry is more directly epistemological. In the case of the mind, the naturalist may claim, mental properties themselves can be seen as part of the scientific conception. After all, nothing could be more available to empirical discovery, and we have already allowed in psychology and other disciplines as sciences of the mind. But isn’t the epistemology of normativity going to turn out to be unacceptably non-empirical? According to the naturalist who responds directly to this challenge, just as normative properties are anchored in the natural world through supervenience, so normative epistemology is also naturalistic. And if normative ethics is a discipline with at least the potential for providing us with access to normative truths, it should be seen as itself a natural science in its own right. To a naturalistically inclined epistemologist, natural science is usually, of course, seen as thoroughly empirical – that is, non-normative and a posteriori – and it inherits the respect due to natural science itself for its 19 As I have suggested, there is work to be done in clarifying the nature of this dependence. For an excellent introduction to the notion of dependence, see Lowe (2010).
68
ro g e r c r i sp
remarkable success as a research programme. And the obvious rationalist alternative – appeal to some kind of non-inferential capacity to grasp synthetic a priori truths – can easily seem to such epistemologists to be, as Crispin Wright (2005: 157) puts it, “empty self-congratulation.” One ethical naturalist who has pressed for an empiricist account of moral epistemology in recent years is Nicholas Sturgeon:20 It does not appear to be by intuition that we know that water is H2O or that heat is molecular motion; these appear, rather, to be theoretical discoveries based ultimately on empirical evidence. So, if the ethical naturalist is to hold … that our knowledge of ethical facts is obtained in much the same way as our knowledge of other natural facts, she will have to hold that knowledge of basic principles in ethics is also empirically based. (Sturgeon 2006: 97–98)
One example of empirical reasoning in ethics Sturgeon (2003: 544–45) offers is the consequentialist argument in favor of adopting commonsense “secondary principles” as a guide to everyday life, which (like any first-order consequentialist claim) rests on various empirical assumptions. But, as Sturgeon notes, the empirical argument here is carried on only against a background of more fundamental ethical assumptions (in particular, consequentialism itself). How plausible is it to claim that we grasp first-order ethical principles through the kind of empirical enquiry characteristic of natural science? Here Sturgeon (2006: 104–105) notes that natural science itself is “theory dependent,” relying on various background assumptions and seeking to bring experimental results into some kind of “reflective equilibrium” with those assumptions. These background assumptions themselves do not have to be construed as constituting intuitive knowledge as understood in the tradition of foundationalism. Rather, the whole story is coherentist and non-intuitionist. But even if we accept that normative ethics is analogous to disciplines more usually characterized as natural sciences in that the epistemology of each is to be understood in terms of reflective equilibrium, is that sufficient for normative ethics to be a natural science? Here we return again to the old distinction between “is” and “ought.” As Sidgwick put it long ago, “it seems clear that an attempt to ascertain the general laws or uniformities by which the varieties of human conduct … may be explained, 20 Because Sturgeon suggests that goodness itself is a self-standing natural property (2006: 98–99), some may be tempted to see his metaphysics as non-naturalist (see e.g., Finlay 2010: 335–36). But since the metaphysical boundaries of the natural are being drawn via epistemology, we should expect a naturalistic epistemology to produce a naturalistic metaphysics.
Naturalism: feel the width
69
is essentially different from an attempt to determine which among these varieties of conduct is right” (1907: 2).21 My hunch is that most naturalists will not wish to press the analogies between ethics and “positive” sciences as far as Sturgeon. For them, ethics can remain autonomous from natural science, along with other disciplines such as mathematics and logic. But the anchoring of normative properties in non-normative properties via strong supervenience and dependence enables naturalists to occupy the metaphysical common ground with those initially attracted to non-naturalism that is constituted by wide naturalism. The question of normative epistemology remains open, as does that of the epistemology of mathematics and logic; but these are other matters.
21
See also e.g., Parfit (2011: v. II, 265); Shafer-Landau (2003: 4).
ch apter 4
On ethical naturalism and the philosophy of language Frank Jackson
This essay is about an old problem that lies at the intersection of metaphysics and the philosophy of language. It comes up in many contexts. The context that will concern us is ethical naturalism. This is partly because this volume is concerned with ethical naturalism and partly because the author of this essay is a naturalist in ethics. There is a currently popular solution to the problem, one that draws on certain causal–externalist views about reference. I will explain why I reject it and offer an alternative solution, one that draws on the network account of moral terms. We start with some preliminary comments on ethical naturalism and on the representational view of language. They will help us identify our problem. Although it is an old one, there is no standard way of saying exactly what it is. It is, though, as we will see, connected to the paradox of analysis. 4 .1 W h at i s e t h ic a l n at u r a l i s m ? Roughly, ethical naturalism is the doctrine that ethical properties are natural properties. What is a natural property? I mean the kinds of properties discussed in the natural sciences, where the natural sciences are understood inclusively. For us, the current level of unemployment, the intensity of a pain, and a tendency to fall over are just as much natural properties as is the mass of a proton. In practice this means that natural properties are the kinds of properties theorists have in mind when they urge that the moral supervenes on the non-moral, and the normative supervenes on the non-normative. The natural properties are the supervenience bases for those two famous supervenience claims. Naturalism is, therefore, a doctrine in metaphysics as much as a doctrine in ethics. Just as we should reject dual attribute theories of mind because they traffic in “spooky” properties, so we should reject Moorean non-naturalism (Moore 1903) and, more generally, theories in ethics that 70
Ethical naturalism and philosophy of language
71
urge that the key ethical properties are quite distinct from those discussed in the natural sciences. They too are spooky. But, of course, it is more complicated than this, and in particular the tendentious term “spooky” calls for scrutiny. Let us review some of the complications and engage in the scrutiny. First, ethical naturalism is typically given a realist gloss. The claim isn’t merely that ethical properties are identical with natural properties. It is that, in addition, they are instantiated on occasion. In this sense, it is a version of realism in ethics. Some actions are (morally) good, some are wrong, and so on. The realist presumption is an important part of the case for ethical naturalism. The naturalist’s conviction is that no items in our world, as it actually is, have the sorts of properties anti-naturalists think goodness and rightness, for example, are. We know this much about what our world is like. It follows that realists in ethics have to be naturalists. Here we have a point of agreement with expressivism. Versions of expressivism which deny that there are any ethical properties can be versions of naturalism, provided that what is expressed (but not reported or described) by ethical terms according to the theory is given a naturalistic gloss. And this is typically the case; it was, for example, an important part of A. J. Ayer’s emotivist version of expressivism.1 However, in this essay we will be solely concerned with ethical naturalism in its manifestation as a realistically construed version of cognitivism in ethics, where what divides the cognitivist from the expressivist is that the former takes talk of ethical properties seriously. What, exactly, is it to take talk of ethical properties seriously? This is not the place for an excursion into debates over the analytic ontology of properties. For us, a version of ethical naturalism takes ethical properties seriously if it has, as a central plank, identifications of ethical properties with natural properties. Where a naturalist of the expressivist stripe might include a claim about the naturalistic nature of the state of mind expressed by “This is wrong”, and might in fact make claims about ethical properties in a suitably minimalist sense of “property,” a sense that requires little more than the existence of (reasonably behaved) ethical predicates, our naturalist identifies the property of being wrong with such and such a natural property. This property may or may not be a property that concerns, in one way or another, various mental states of humans, or idealized humans, or … The distinction we are talking about cuts across the subjective–objective distinction among ethical theories. See especially Ayer (1965).
1
72
f r a n k j ac k s on
Second, the kind of ethical naturalism that will concern us is a strong version. What makes it a strong version comes from the way it spells out what is meant by a natural property. Consider someone who argues that Being good = being of a kind that idealized versions of ourselves desire to promote. They then go on to explain that, in their view, desire, and indeed many mental states, have an ineliminable normative element, and maybe that this is especially the case when what is in play are the desires of idealized versions of ourselves. Have they naturalized being good? In one sense, yes. The property mentioned on the right-hand side is a natural property in the ordinary, inclusive sense of the term “natural.” But they have not naturalized being good in another sense. They have not identified being good with a property specified in terms they take to be non-normative, and that is the claim made by ethical naturalists which is of special interest, or so it seems to me. The normative supervenes on the non-normative, and the supervenience of the moral on the non-moral is a special case of this. The naturalist’s – the strong naturalist’s – response to this fact is to seek an account of the moral as a pattern in the non-moral. The goal is to find a truth of the form Being good = being such and such where “such and such” is a pattern in the non-moral. What is a pattern in the non-moral? It would be confused to seek an account of being good that identified it with a non-moral property. That would be to violate Leibniz’s law. What the naturalist seeks, rather, is a commonality between arrangements of elements that are themselves nonmoral. Density is a pattern or commonality among facts about volume and mass. It is the pattern thrown up when you divide mass by volume. Velocity is a pattern among facts about position and time. Shape is a pattern among facts about certain points and their location. And so it goes. Our naturalist holds that the ethical is a pattern among the non-ethical. This means that someone who offered Being good = being of a kind that idealized versions of ourselves desire to promote while adding that the notions of desire, being idealized, etc. are non-normative, would be acting as an ethical naturalist (in our sense – I will take this qualification as read from now on). They might be mistaken in their contention that the ingredients on the right-hand side are non-normative
Ethical naturalism and philosophy of language
73
but they wouldn’t be making the mistake of misreading the aim of the exercise. A pattern in the non-moral is, then, a pattern discernible in arrangements of the non-moral. In this sense, naturalists in ethics are like certain materialists about the mental. Materialists don’t identify the mental with the non-mental; to do that would violate Leibniz’s law. They identify the mental with arrangements of the non-mental. Their contention is that one can make creatures with minds, including ones with conscious mental states, by suitably putting together non-mental components, using ways of arrangement that are not in themselves mental. Each bit is non-mental, and each connection between the non-mental bits does not in itself constitute being mental. The aggregation makes up a mind but what is aggregated is one and all non-mental. In LEGOTM one can make a truck using components none of which is a truck, and using ways of putting the components together, none of which is in itself the making of a truck. Mutatis mutandis for materialism. Sometimes this way of thinking about materialism is put in terms of “what God could do.” The contention is that God could make consciousness using non-mental ingredients and modes of composition, none of which is itself mental. But of course, at a more mundane level, we materialists think that this happens every time a sentient creature is conceived and grows up. Growing into a thing with consciousness, and mental states more generally, is a process of aggregating the non-mental, using non-mental ways of aggregating. But the aggregation part of the ethical naturalist’s story is only part of the story. Properties correspond to possible patterns in nature. In saying this, I am not taking a stance on whether or not these patterns are additions, ontologically speaking, to the items that fall under the patterns. (We set aside the analytic ontology of properties earlier, and to do that is inter alia to decline to debate resemblance nominalism.) But I am insisting that there must be a pattern, even if it is a highly disjunctive one.2 When do we have a pattern in an aggregation of the non-moral? When we have in principle projectibility. So the core contention of naturalism is that moral properties are patterns – projectible similarities, in aggregations of the non-moral. (To the extent that the pattern is disjunctive, irreducible pluralism about values will be true.) Finally, there is the question of the relation of naturalism to the debate over the autonomy of ethics, on one reading of that debate. Here is where we address the issue of when a property counts as spooky.
Cf. Ayer’s discussion of knowledge in Ayer (1956: ch. 1, section ii, 10–14).
2
74
f r a n k j ac k s on
The thinking of some supporters of autonomy can be summarized roughly as follows. The supervenience of the ethical on the non-ethical tells us that the bearers of the ethical properties, taken one by one, are, in some good sense, purely non-ethical. For each item that has an ethical property, it is a mistake to think of that property as an addition to the item’s non-ethical properties. To say otherwise is to believe in spooky properties, in the sense of ones we have no reason to believe are ever instantiated; to say otherwise is to give one no way of resisting Mackie’s (1977) error theory. However, patterns can emerge in the following sense. There can be a unity between and among the non-ethical that can only be grasped by those in possession of ethical concepts. Part of gaining a mastery of ethical concepts is seeing the patterns in the non-ethical that are necessarily invisible to those who have only a mastery of non-ethical concepts. Absent a grasp of the concept of the good, the items that count as morally good will present themselves as a patternless, potentially infinite disjunction, and the same goes for rightness, etc.
Some think, or seem to think, that the line of thought of the above paragraph carves out a distinctive, non-reductive style of naturalism.3 As against this, I think that it carves out the only plausible version of naturalism (among cognitivist versions of naturalism, that is). It says what all naturalists should and must say. Let me make the point with our earlier example of density. Suppose I give you a good number of medium-sized but otherwise rather different objects, and ask you to find the interesting commonality among them. You check how rigid each is, what its mass is, its volume, and what each is made of, but initially nothing relevant emerges. Although there will be known properties in common to the objects – for example, you know that each object is of a kind I can afford to use in the experiment I am conducting, and is one that you have been given to examine – they are not the kinds of commonalities you were being invited to look for. You knew things like that from the outset. You then try dividing the mass of each object by its volume and the penny drops – the answer turns out to be the same for each object. The objects are alike in how dense they are. But, and this is the important point for us, there is no question of discerning the relevant pattern independently of grasping the concept of density. To discern the pattern is to bring the data under the concept of density. You may or may not have the word “density” in your arsenal for describing how things are, but in spotting the key unifier, you categorize the objects in terms of density.
As opposed to the reductive version to be found in, e.g., Jackson (1998: chs. 5 and 6).
3
Ethical naturalism and philosophy of language
75
The same goes for finding the patterns in the non-moral that constitute the moral. As we said above, it would be quite wrong to think of naturalism as identifying the moral with the non-moral; that would be like holding that materialism identifies the mental with the non-mental. Both theories, rather, are claims about the kinds of patterns we can make with non-moral and non-mental ingredients, respectively. But then spotting the pattern in the non-moral precisely requires a grasp of moral concepts. 4 .2 A dig r e s s ion: e t h ic a l n at u r a l i s m w i t hou t prope r t i e s ? One way to be an ethical naturalist while refusing to talk about ethical properties is to embrace a suitable version of expressivism – as we say above. However, you might think that there is another way to eschew questions about ethical properties while still being resolutely naturalistic. Insist that we do not need to talk about ethical properties in the first place! That way of talking and thinking about ethics, one might argue, is part of a mistaken tradition, and not because we should be expressivists, but because ethics is concerned with reasons for action, not with properties. What ought to be done is not that which has some property of “ought to be done-ness” (whatever that might be); what ought to be done is that for which there is a reason of a certain kind. Here we are, runs the thought, being cognitivists about what it is for there to be a reason to do so and so – to affirm that there is a reason to do so and so is to report on how things are; it isn’t to express an attitude. But we are avoiding “what are” questions about moral properties. It will of course be granted that it is acceptable English to say something like “When there is an overwhelming moral reason for doing A, then A is the morally right thing to do,” and that, when suitable bells and whistles are attached, a remark like this will be true (and, maybe, true a priori). The mistake, runs the thought, is to move from this observation to the claim that there is, metaphysically speaking, a property of being right, a property that theorists of ethics are required to give an account of. Instead, what is required is an account of when there are reasons of one kind or another (including especially the moral kind, when we are doing ethics) for acting. We have, is the thought, the old philosophers’ error of moving from the existence of a predicate to the existence of a property corresponding to it. This (interesting) thought rests, it seems to me, on a mistaken view about reasons for action, or indeed reasons for anything. Reasons don’t float free, as we might put it, from the nature of what they are reasons for.
76
f r a n k j ac k s on
Suppose that, in some specified circumstance, an act’s being the intentional killing of an innocent child provides a decisive reason why it would be morally wrong to perform the act (some will say that the qualifier “in some specified circumstance” is otiose). This will be because of the nature of the act. A reason for doing A supervenes on the nature of A. If R in circumstance C is a reason for doing A but is not a reason for doing B, this can only be the case because of a difference between A and B. This means that a theory that specifies the actions for which such and such, in so and so circumstances, are moral reasons for doing them is ipso facto an account of the nature of those actions. All cognitivists, it seems to me, have to believe in moral properties, and so need to face the question of what they are. Naturalists who hold that what is basic in ethics are the reasons, and especially the moral reasons, for doing thus and so in circumstances so and so, will have their own distinctive take on how to find true identities for the moral properties. For example, their search for the property of being right will, in one way or another, be the search for what it takes for an action to be the action one has most reason to do, or most reason of such and such a kind, but it will be a search with a genuine target. It is time to focus on our problem. 4 .3 W h y b o t h e r s a y i ng a n y t h i ng ? I am driving to a party but have become lost. You hand me a street map. I am then able to get to the party. How so? Perhaps your handing me the street map causes me to pull over to the curb in a way that attracts the attention of the police. They question me, learn of my problem and kindly escort me to the party. But that is not the normal case. In the normal case, a key part of the explanation rests on my knowing how the map represents things to be. The act of handing over the map isn’t in itself the key, nor are the lines, shapes, and colors on the map; what is crucial is that I know how the map represents things to be. That is to say, I know how the streets, etc. have to be to be as the map says they are. That knowledge, plus the map’s being accurate and my driving abilities, are what gets me to the party. If someone asks, “Why bother handing maps to people?” the answer has to be that the people receiving the maps by and large know how they represent things to be. The same goes for uttering sentences like “The right thing to do is to keep that promise,” or “It is wrong that people don’t give more money to
Ethical naturalism and philosophy of language
77
Oxfam.” For cognitivists in ethics, sentences like these represent things as being a certain way, and their impact rests on people grasping how things are being represented to be. If that were not true, one might as well speak Martian. Well, it is, as so often, not that simple. There are non-standard cases, cases like the policeman in our story about finding the party after being handed the map. Or consider a case where one hears “Insider trading is wrong” from the mouth of a government official at a seminar for share traders. The impact of hearing this sentence on a share trader may well be explained by simple self-interest. One engages in a bit of practical reasoning that might be spelt out thus: If an official says “Insider trading is wrong,” it is likely that there are penalties for insider trading. If I engage in insider trading, there is a good chance I’ll be caught. I don’t welcome being penalized. Therefore, I should refrain from insider trading. But, for cognitivists, or anyway cognitivists who aren’t error theorists, the standard case has to be the one where the rationale for giving, and for responding to, moral advice rests on knowing how moral sentences represent things to be. Arguably, error theorists can say, given their view that what the sentences say about how things are is invariably false (except in trivial cases), that the attitude to take to the question of when one should use a moral sentence, and the question of how one should respond to coming across a moral sentence, lies in the kinds of pragmatic considerations we illustrated with the example of the share trader. Expressivists will reject holus-bolus the talk of representation. But for the kind of cognitivist that the typical naturalist in ethics is, and I include myself here, when it is right to utter an ethical sentence, and how one should respond to someone else’s uttering one, must be in large part a function of what the sentence says – its content. For cognitivists, the most important element in deciding when to produce an ethical sentence and how to respond to hearing one will rest on whether or not one accepts that things are as they are being said to be. Moreover, one had better know how the sentence is representing things to be. The value of a map rests on the combination of its representing that things are thus and so and its not being a dark secret how it represents things to be. We are now in a position to state our problem.
78
f r a n k j ac k s on 4 . 4 W h y a r e n’t w e a l l n at u r a l i s t s i n e t h ic s ?
If naturalism is true, ethical properties are one and all natural properties. But then how we represent things to be when we say, as it might be, that something is right is that it has such and such a natural property. What is more, we have noted that we had better know what we and our fellow speakers of the language are saying about how things are when we use a term like “right” – there wouldn’t be much point in our using the word otherwise. There wouldn’t be much point for the same reason there isn’t much point in using (the English word) “right” when talking to monolingual Russian speakers, or to Martians. How come then, if naturalism is true, that we aren’t all naturalists? What is more, why don’t all naturalists agree on what natural property rightness is, and on what natural property moral goodness is, and so on. Why is there so much disagreement, both between naturalists and non-naturalists, and among naturalists themselves? I mentioned right at the beginning that our problem was connected with the old paradox of analysis, and so it is. But I emphasize that I have raised our problem without talking about conceptual analysis. I am sure many feel that we no longer need to worry about the paradox of analysis because, in their view, the program of seeking analyses has been discredited. Be this as it may, we all should worry about why it is worth saying things, why it is worth using words – and that is what has generated our problem. It is worth using the word “poisonous,” as in “Don’t drink that; it is poisonous,” because speaker and hearer know the property being ascribed by the word, and so how the sentence represents things to be. Mutatis mutandis for ethical terms – or anyway mutatis mutandis if cognitivism, and in particular naturalism, is true. The rest of this essay is about two very different ways to respond to our problem. One sees ethical properties as solutions to something roughly akin to a set of simultaneous equations, a set that captures the way ethical concepts form an interlocking network, a network which is, to some extent, still under negotiation; the other draws on recent work on reference to natural kinds, inspired most especially by Hilary Putnam.4 I will discuss the second first and explain why I disagree with it. 4 .5 On t h e de t e r m i n e r s of r e f e r e nc e Here is a familiar story.
Most especially Putnam (1975: 215–71).
4
Ethical naturalism and philosophy of language
79
Once upon a time we thought that the reference of a term like “water” was to whatever has enough of the properties we associate with water: being potable, falling from the sky, being liquid at room temperature, odorless, etc. We now know that this was a mistake. Reference is determined by certain causal facts. There is dispute about how exactly to spell out these causal facts, and even about whether it is a sensible project to attempt to spell them out. But there is broad agreement on the following. (i) Our use of “water” is one intended to refer to an important underlying kind with which we causally interact. (No one is saying that we couldn’t have used “water” as an abbreviation of, say, the kind that falls from the sky; it isn’t 1984. The claim concerns how we in fact use the word.) This point about the way we use “water” explains why its reference goes by certain causal facts. (ii) A signal advantage of this theory is that it allows reference to be preserved under substantial change of opinion. People’s opinions about the nature of water have changed significantly over the years, but the subject of their changes of opinion has remained the same, as has their ability to use the word “water” to pick out that subject. But that can only be true if the reference of “water” stayed the same despite the changes of opinion. Similar points apply to mass, energy, atom, etc.: our beliefs about them (including beliefs one might have thought definitive, like the belief that atoms are indivisible) have changed, but the subjects of our beliefs have not, nor has our ability to use “mass,” “energy,” “atom,” etc. to pick out those subjects. (iii) There is a sense in which the reference of “water” can be opaque to the user of the word, despite their being competent with the word. We now know that “water” refers to H2O, but it is a reasonably recent discovery. It involved discovering that the kind that played the crucial causal role involved in reference determination was H2O, something we learned in the early 1800s. All the same, speakers of English in the 1700s knew, in some good sense, what “water” meant. Again, similar points apply to “atom,” etc.
The above is a sketch that leaves many matters vague and open but says enough to make it clear why applying the same basic approach to the reference of ethical terms might seem to offer a simple solution to our problem, and, in addition, an independently attractive view of ethical terms.5 If, its supporters say, we apply it to ethical terms, we can allow that understanding them does not deliver what they refer to. No wonder then, even given the truth of naturalism, that the claim that ethical properties are natural properties is controversial, and that naturalists differ among themselves over which natural property to identify with which ethical property. True, we must allow that we understand ethical terms, but the insights from Putnam and his supporters tell us that understanding does not deliver reference. Moreover, we can allow that people can differ very 5 For two examples among many, see Loeb (1998, esp. around p. 293) (Loeb notes his debt to Boyd (1993) and Bloomfield (2001) (around p. 120 esp.). See also the discussion of the implications of Putnam’s views in ch. 6 of Brink (1989).
80
f r a n k j ac k s on
substantially over what is right, what is bad, and so on, while still being in genuine disagreement. They need not be talking past each other. A causal approach to reference, inspired by Putnam on natural kinds, allows us to hold that the Kantian and the utilitarian refer to the very same property when they use, say, “right,” despite their big differences over which items are right, and over why they are right. Of course, in carrying over the story from natural kind terms to ethical terms, one might well expect a need for substantial modifications. We shouldn’t follow the story for “water” and “atom” too slavishly. The key point is that the example of natural kind terms tells us, on the account we are retailing, that the reference of a term can depend on external factors, especially causal ones, that may be unknown to users, consistently with the users’ understanding the term in the sense of being competent with it. That is the core insight naturalists need when we think about the reference of ethical terms. For these terms are, naturalists should hold, examples of terms of precisely this kind. There is no denying the initial appeal of the picture I have just rehashed. Its popularity is no surprise. But I do not think its appeal survives probing. Our probing starts by noting an ambiguity in what might be meant by reference in these discussions. There is a sense in which we all agree that understanding is consistent with ignorance of reference. We all understand the phrase “the tallest woman” but few, if any of us, know its reference in the sense of knowing who the tallest woman is. There is, however, a sense in which we do know the reference of “the tallest woman.” We know what it takes for someone to be the tallest woman. We know, that is, the property someone has to have to be the tallest woman. In consequence, we know what we are saying about how things are when we say “Mary is the tallest woman” or “I don’t know who the tallest woman is.” That follows from our grasp of what “the tallest woman” means, and explains the phrase’s utility in passing on information about how things are. Someone who does not know what it takes to be the tallest woman does not know what information about how Mary is I am providing when I say that she is, or that she is not, the tallest woman – or what it is that I am confessing ignorance about when I say that I haven’t a clue whether she is or isn’t the tallest woman. Our interest here is in reference in the second sense. Our problem arose from the observation that coming across ethical terms in sentences informs us about how things are, or at least informs us about how producers of sentences containing the terms believe things are – or so we cognitivists do and must affirm, on pain of making the rationale for producing
Ethical naturalism and philosophy of language
81
sentences containing ethical terms a mystery. So the question we need to ask is: what might some version of the causal theory of reference for ethical terms say about the role of these terms in providing information to those who understand English? This is, however, no easy question to answer. The trouble is that there is no standard answer in the literature to our question about what one or another externalist, causal theory of reference for some term “K” says about the informational role of “K.” There is plenty about reference but often little about the connection between reference and informational value, and when the connection does come up, it is often in the context of discussions of two-dimensionalism, a controversial topic in and of itself. However, we can chart the two main options. First, one might say that an externalist, causal theory of reference has nothing at all to say about what a subject is saying about how things are when they use ethical terms. In that case, such a theory of reference is silent on our problem, and more generally is silent on the topic of the informational value of making or coming across sentences containing ethical terms. It is also, it follows, silent on why anyone should take special notice when they hear sentences containing ethical terms. Second, one might say that an externalist, causal theory of reference does tell us what a subject is saying about how things are when they use ethical terms, but what they are saying about how things are depends on matters that outrun what one gets from understanding the words. That, it might be said, is the whole point of the externalist part of the theory. The theory tells us what determines what we are claiming about how things are, but we may not know what that is. This, it might be said, is precisely the situation of those who used the word “water” in the 1700s: they didn’t know what they were saying about how things were.6 This latter claim is hard to believe, but I have said enough on that point elsewhere (Jackson 2004). Here our interest is with how the idea would play for ethical terms. There are two possibilities. One play says that, here and now, we are in the same kind of situation as those who used “water” in the 1700s. There is, runs the play, something we are claiming about how things are when we use ethical terms but we don’t know what it is. In that case, I observe, we should stop using ethical terms until we know what we are saying about how things are when we use them, and, in the context of our To clarify: it is agreed, on this option, that they knew what they were referring to in one sense of “reference,” but not in the sense that delivers the information provided by the word “water.” 6
82
f r a n k j ac k s on
problem, this looks like a pyrrhic victory if ever there was one. Also, this would seem a curious way to preserve the intuition that the Kantian and the utilitarian refer to the very same property when they use, say, “right,” despite their big differences over which items are right, and over why they are right. Yes, they are disagreeing but (unfortunately) do not know what it is that they are disagreeing about. The second play says that we do know the relevant external, causal facts that determine the reference of ethical terms, and so, on the idea under discussion, do know what we are saying about how things are when we use ethical terms. The situation is like that for “water” today. But, in that case, naturalists would have to say that there is widespread agreement about the natural property that is moral rightness, the natural property that is goodness, and so on. It is manifest that there is no such widespread agreement. There is widespread agreement today that “water” refers to H2O; however, the situation with moral terms is very different, as we all know only too well. It is time to turn positive. What follows is my suggestion for how we naturalists can solve our problem. It draws on the kind of moral functionalism – the view that the meanings of the moral terms are given by their place in a network – that I have championed elsewhere. I think the fact that moral functionalism offers a solution to our problem is a substantial point in its favor, but I am hardly an unbiased observer.7 4 .6 N e t wor k s a n d a r e a s of ag r e e m e n t a n d di s ag r e e m e n t Our problem is generated by two simple observations. There wouldn’t be much point in using moral terms unless there was a substantial amount of agreement about how they represent things to be. But, in that case, why is there so much dissension, and in particular why, if naturalism is true, is there so much dissension over any offered identifications of ethical properties with natural properties? A network account of the meaning of ethical terms is one that makes substantial agreement compatible with widespread dissension. I will now spell this out. The network account of the meaning of moral terms starts from the claim that our mastery of moral concepts involves a mastery of a network of interlocking concepts. Mastering the network is mastering the moral vocabulary.
See, e.g., Jackson (1992; 1998: chs. 4 and 5; 2005) and Jackson and Pettit (1995).
7
Ethical naturalism and philosophy of language
83
Why is this claim plausible? Think of what one would say to someone who did not appreciate such things as: that there is a close connection between goodness and rightness, between rights and duty, between being virtuous and doing the right thing, between conceding a right to do x and allowing x to be done, and between the propriety of punishment and the punished being guilty, or that equity matters for what ought to be, that morality connects intimately with action, that there is a non-accidental connection between what justice requires and what ought to be done, that the demands of morality have a special status, that there is close connection between an act’s having awful results and being wrong, and so on. What else is there to say to such a person other than that they “just don’t get it.” There is a way of conceptualizing our world, the moral way, that has passed them by. One way to bring this out is to note that it would be wrong-headed to suggest an experiment to establish that there is a connection between what is good and what is right, or between rights and duties, or to show that killing the innocent is wrong. Indeed, what would it be like to carry out an experiment to establish one or more of these claims? (Of course, one might carry out an experiment to see how widely these claims are accepted, but that’s another question.) We display our grasp of the network pretty much every time we engage in ethical debate. How do pro-choice advocates respond to the argument: abortion involves the killing of an innocent person. Therefore, abortion is wrong. They don’t respond by saying that it is perfectly all right to kill innocent people. They accept that there is some kind of conceptual link between killing the innocent and being wrong. They respond by arguing that the fetus is not a person, and so that the case is not a case of killing an innocent person. Or think of what happens when an argument like the following is up for discussion: there is a right to free speech. Therefore, censorship is only justified when what is censored would have very bad consequences. The discussion requires a grasp of the interplay between the existence of rights, what is justified, and bad consequences. I could give many more examples with lots of detail but the key point is clear. Debates over ethical issues presuppose a whole range of connections between ethical concepts, and grasping these connections is part and parcel of understanding the language of morality, and part and parcel of engaging in debates over moral questions. As I have said before, for examples to illustrate this point, one only needs to reflect on the last argument one had with friends (or the opposite) over a moral question.
84
f r a n k j ac k s on
Where, then, is the dissension? If we agree about the network, what is there to differ about? There are two places where we can differ while leaving the network relatively (relatively) undisturbed. First, the precise interpretation of almost every item on our “network” list is contestable. Consequentialists give a deflationary reading to the demands of equity. They do not simply say that equity is neither here nor there, but they do say that its relevance is derivative; it comes from the way, as a rule, equity adds to good results. Another example is the way the connection between moral judgment and action is read differently by internalists and externalists about morality. Again, the problems with punishing the innocent are treated differently by different theorists. And so on. In saying this, I am not going back on the network view. For, and this is the crucial point, each item’s place in the network is shown by the fact that it has to be debated. Although consequentialists downplay equity and rights, all sides agree that a major topic for them is to justify their downplaying, and the idea that utilitarians can properly ignore the issue about punishing the innocent would be regarded as bizarre. Externalists may insist that it is perfectly possible to judge that an act is right and yet not be motivated to do it but, all the same, they seek to explain away the contrary intuition. Each item is, if you like, a compulsory question on the ethics agenda. Every theorist has to make some sort of good sense out of each of them – has, that is, to explain why they capture some important truth, where this explaining may well involve explaining why and when there are exceptions to them. We test moral theories by examining the quality of their explanations, and how else could this be true unless something like the network view were correct? What is more, we test proffered explanations against other items in the network, that is, against other claims we find intuitively appealing. How else could we test them, given, as we note above, that experiments in the empirical sense are beside the point? An experiment to show that it is sometimes right to punish the innocent could at most show that punishing the innocent sometimes has good effects, but the question on the table is precisely whether having good effects justifies punishing the innocent. The experiment does not – cannot – answer this question; it simply creates the case the question relates to. The second place where we can have significant differences lies in whether or not there is a single underlying rationale for the network we have been discussing. We can distinguish monists from pluralists. Monists think that there is some reasonably simple, overarching conception of morality that delivers the network, give or take some wrinkles. Take a simple form of act utilitarianism that says that the moral goodness of an action is its expected “hedon” value, and that the right act is the one that
Ethical naturalism and philosophy of language
85
maximizes expected hedon value out of the available options. That delivers the relation between goodness and rightness in a nice simple way (too simple, many say). We are then given accounts of rights, justice, the virtues, and so on that see their relations to right action as explicable in consequentialist terms. There is a right to free speech because, to put it very roughly to give the general idea, societies that protect freedom of speech are typically happier than those that do not. Of course many argue that utilitarians’ accounts of rights seriously underplay their role in determining what ought to be, but this is confirmation of the network picture, as we have noted already. The very fact that utilitarianism’s capacity to make sense of the network, and the way it is assessed against other items in the network, is crucial to its acceptability confirms the centrality of the network. The system builders in normative ethics are monists. They seek some simple, central insight that, together with suitable add-ons, allows us to make sense of the network. Monists who don’t like the consequentialists’ picture try to do the job with the idea of obligations generated by implicit contracts, or with the concept of actions that display the virtues as the touchstone for being right, or with some list of prima facie duties, or by reference to the actions of ideal beings, or actions that are ideally rational, or actions that answer to what agents desire to desire, or whatever. The core notion deployed varies from one to another attempt to systematize, but what unites the otherwise diverse approaches is the conviction of the system builders that, somehow or other, there is a way to find a single, fundamental principle that allows one to recover the network, albeit at the cost of a certain amount of culling and repackaging of elements of the network. By contrast, the pluralists hold that the various parts of the network are, to one extent or another, independent. And we all know what the monists say to the pluralists. They describe cases where there is a clash between the moral verdict suggested by one part of the network with that suggested by another part, and argue that the only non-ad hoc way to arbitrate cases like these requires finding a single touchstone of just the kind monists believe in. 4 .7 W h y t h e n e t wor k ac c ou n t i s a s t rong v e r s ion of n at u r a l i s m We said early on that naturalism should be given a strong reading, a reading on which it delivers identifications of the moral properties in nonmoral terms. That honors the conviction that lies behind naturalism, the conviction that we know enough about our world to know that the moral
86
f r a n k j ac k s on
properties are not additions to the picture naturalism gives us of what our world is like (a point sometimes bolstered by arguing that the best way of explaining the supervenience of the moral on the non-moral is to hold that the moral is not an addition to the non-moral nature of our world). The network or functionalist picture delivers precisely a version of naturalism that meets this constraint. We show this by exploiting the Ramsey–Carnap–Lewis way of defining theoretical terms.8 I will say it quickly as it is a familiar story. Let T (M1, …, Mn) be the sentence that gives the network, where the ‘Mi’s are all the moral terms. We think of each Mi as defined by its place in T, so that to be Mi is to be that which is in the i-th place. That is to say, y is Mi iff (Ex 1) … (Exn) [y has xi and T (x 1, …, xn)], where each xj is in Mj’s place in T. As the right-hand side of this biconditional contains no occurrences of any Mj, the network story specifies in non-moral terms what it takes to be right, good, just and so on. We have the needed identification of moral properties in terms of the non-moral. We also have a potential ambiguity. In the philosophy of mind case, the ambiguity is over what to identify with, say, belief that snow is white. Some hold that belief that snow is white is that which fills the “belief that snow is white” place in the corresponding network for psychology; some want to say that belief that snow is white is being in the state of having the place filled; and some want to say that we should distinguish belief that snow is white from believing that snow is white, and that the first is the filler state and the second is the state of having the place filled. For what it is worth, I favor the latter position, the position of the version of the mind–brain type–type identity theory championed by David Lewis, D. M. Armstrong and J. J. C. Smart.9 A similar debate is possible in the ethics case. According to moral functionalism, an act is right if and only if it has the natural property that fills the place of “right” in the network. We can then think of rightness as the property that fills the place, or instead as the property of having the property that fills the place whatever that property may be. Nothing of importance here turns on this, as far as I can see.10
8 See, e.g., Lewis (1970). 9 See, e.g., Armstrong (1968). 10 In the mind case, the reason for favoring identifying belief, e.g., with the filler state is to preserve the causal role of belief. This reason does not apply in the ethics case. In Jackson 1998: 141, I gave a different reason for favoring identifying rightness with the filler state, but I now cannot see much force in the reason I gave there.
Ethical naturalism and philosophy of language
87
4 .8 How t h e n e t wor k v i e w e x pl a i ns t h e c om m u n ic at i v e va lu e of e t h ic a l t e r m s The differences we talked about in 4.6 are big differences, and they will show up when ethical theorists come to offer their accounts of the property of, say, being right. However, the network view allows that there is a substantial commonality in the midst of this diversity. On the network view, the ultimate aim of ethical theory is to make the best sense of the network, where this may well involve modifying the network. The network isn’t like a fixed set of simultaneous equations in mathematics waiting to be solved. The network is arguably a work in progress, evolving under the impact of critical reflection, but possibly (possibly) converging on what I call elsewhere mature folk morality (Jackson 1998: 133). At the convergence point, if it exists, the clauses of the network and the solutions for the properties that occupy its nodes are settled once and for all. All the same, there is a shared commitment to the aim of making good sense of the network, as we emphasized. This allows us to give an account of what goes on when we use ethical language, an account that fits with the cognitivist picture in that it explains the shared representational contents of moral terms, and so what we communicate to one another when we use moral terms. What we share when we say that some act is right is that the act has the property that plays the rightness role in the network, and is indeed a sharing because we share the network. Kantians and utilitarians, for instance, agree on the network in the broad. As we noted earlier, the precise readings given by different theorists to the clauses of the network will vary, and this implies corresponding differences in the meanings of moral terms. But we can live with a degree of difference. A shared place in a broadly similar network is enough to explain why we find moral terms useful for communication. The Kantian and utilitarian alike are affirming, when they say “A is right” or “B is bad,” that A has the property that plays the “rightness” role in the network, and that B has the property that plays the “badness” role in the network. What they mean by “the network” is not quite the same but it is near enough to allow useful communication to take place. Of course, sometimes nature is kind to us. The cases it throws up for adjudication are such that they get the same ethical verdict concerning what ought to be done on any reasonable reading of the clauses of the network, and any reasonable way of seeking to make overall sense of the network. If we define “x is uncontentiously right” as x is right by the lights of any defensible account of the network, then the cases where nature is
88
f r a n k j ac k s on
kind to us are the cases where there is an action which is uncontentiously right. This is the sense in which Kantian and utilitarian, to stick with that example, can agree in many cases. They are united by their commitment in the broad to the network, and their differences don’t matter in cases where any reasonable reading of the network delivers the same answer concerning what ought to be done. If (if ) it is part of the network that convergence will occur, we can go further and note a sense in which Kantian and utilitarian mean the very same by “right.” They both mean by “right,” right according to the network which is mature folk morality. To be right is to have the property that fits the constraints set by mature folk morality. But is convergence part of the network? This is an empirical question. The network is our creation. As we noted earlier, it is not 1984. My inclination has always been to take it that convergence is part of the network but others insist that, as a matter of empirical fact, it is not. And let me close by noting that this is a question that might well be addressed by experimental philosophers, while allowing myself the observation that surveys on the point would need to be done with due attention to the somewhat complex nature of the issues, in addition to the requirements for good surveys in general.
ch apter 5
Metaethical pluralism: how both moral naturalism and moral skepticism may be permissible positions Richard Joyce
5.1 I n t roduc t ion: s om e di s t i nc t ions This paper concerns the relation between two metaethical theses: moral naturalism and moral skepticism. It is important that we distinguish both from a couple of methodological principles with which they might be confused. Let us give the label “Cartesian skepticism” to the method of subjecting to doubt everything for which it is possible to do so – usually by introducing alternative hypotheses that are consistent with all available evidence (e.g., brains in vats). Let us give the label “global naturalism” to the principle that requires of any item which we admit into our ontology that it “fits” (in some manner or cluster of manners to be specified) with our naturalistic scientific worldview. One might be both a Cartesian skeptic and a global naturalist, if the latter principle is something that has survived the former test procedure. Alternatively, one might have adopted global naturalism for some other reason, while having little patience with the Cartesian method of doubt. Moral naturalism is the metaethical view that moral entities (e.g., properties like goodness and evil) fit within the scientific image of the world. The moral naturalist will probably be a global naturalist, but need not be: it is consistent with allowing non-natural entities into one’s ontology that one happens to think that moral properties are of the natural variety. Moral skepticism denies that moral entities fit within our scientific worldview. One way of denying moral naturalism is to be a moral error theorist: to hold that our moral discourse attempts to make reference to moral properties, but these properties do not exist.1 Another way of 1 The usual error theoretic strategy here is to argue that moral properties have features that no naturalistic property can have (e.g., some strong kind of practical authority) and thus the moral naturalist’s characterization of moral properties must be inadequate. (See Joyce 2001; Mackie 1977.) Alternatively, the error theorist might accept the moral naturalist’s characterization of the
89
90
r ic h a r d joyc e
denying moral naturalism is to be a non-cognitivist: to hold that our moral discourse was never really in the business of referring to moral facts or properties in the first place, and ipso facto such facts or properties are not naturalistic. In this paper, the label “moral skepticism” denotes the disjunction of these two theses. Neither the error theorist nor the noncognitivist must be committed to global naturalism, but usually will be; indeed, this commitment will often be a motivating factor of their metaethical views. The error theorist who is a global naturalist will typically deny the existence of moral properties precisely because (he thinks) these properties fail to fit with a naturalistic worldview. Likewise, those who turn to non-cognitivism often do so because they cannot see how moral properties could find a place in a naturalistic order. (There is a third way of denying moral naturalism which is not a kind of moral skepticism at all. The moral non-naturalist denies that moral entities fit within our naturalistic worldview but does not judge this a mark against their existence or our epistemic access to them; moral facts are acceptable non-natural facts. I shall not be discussing this metaethical view in this paper.) “Skepticism” is a pliable term, and the way I am using it here will not jibe with the way all others use it.2 I have already stipulated how the term will be employed, but it is probably useful to make a couple of further clarifications. First, I have described both the error theorist and the noncognitivist as denying something: that moral entities are to be admitted into our ontology. This stands in contrast to a classical kind of skeptic who neither affirms nor denies but simply doubts. The skeptic I have in mind doubts X in a strong sense of affirming the denial of X (and is thus, in the classical vocabulary, a negative dogmatist). Second, my moral skeptic is not playing the Cartesian game of doubting moral naturalism simply because alternative hypotheses can be conjured. My moral skeptic has probably accepted the naturalistic worldview – indeed, employs it as a premise for skepticism. Thus the skeptic takes himself or herself to moral (for the sake of argument) and argue that there are still no such properties. For a study of the latter strategy, see Joyce (2011b). 2 It is difficult to choose a label that denotes the disjunction of the error theory and noncognitivism. “Moral anti-realism” would serve in some people’s books, but I prefer to reserve this term to cover the error theory, non-cognitivism, and a kind of moral naturalism according to which moral facts are in some manner constructed by human attitudes (as opposed to being objective features of the world). To my terminological tastes, some moral naturalists are moral realists and some are moral anti-realists. The disjunction of the error theory and non-cognitivism is called “skepticism about moral truth” by Walter Sinnott-Armstrong in his taxonomy of moral skepticisms (Sinnott-Armstrong 2006a: 11).
Metaethical pluralism
91
have positive grounds for doubting moral naturalism beyond the unimpressive pastime of imagining far-fetched (but unfalsifiable) scenarios in which moral properties are absent. This is a naturalistic skepticism, not a Cartesian skepticism. Moral naturalism and moral skepticism are certainly not contradictories; they do not exhaust the space of metaethical possibilities. (I have already mentioned one alternative: non-naturalism.) But it is usual to think of them as contraries, for surely to embrace one position is to reject the other. In this paper I want to investigate this latter assumption closely. I will explore the possibility of a certain kind of metaethical pluralism, whereby seemingly contrary metaethical positions enjoy equal claims to legitimacy – and not just in the sense that we are not (yet) in a position to know which position is correct, but in the more interesting sense that there is no fact of the matter about which is correct. 5.2 L e w i s’ plu r a l i s m I take as my point of departure a series of intriguing and characteristically astute comments made by David Lewis at the close of his 1989 paper “Dispositional Theories of Value. Part ii.” In the course of this paper Lewis has developed and advocated a kind of moral naturalism: a dispositional theory according to which values are those things which we are disposed, under certain idealized conditions, to desire to desire. Worried that this theory makes values contingent – that we feel uneasy about a theory that allows that benevolence (say) is, but might not have been, a value – Lewis toys with identifying values with what we are necessarily disposed to desire to desire. This amendment would probably soothe the worry, he acknowledges, but it would be at a price: there is probably nothing that we are necessarily disposed to desire to desire; thus to identify moral properties with this dispositional property would leave us with a moral error theory. It is not Lewis’ dispositional theory per se that interests me here, but what he goes on to say next. The version with the necessity operator is, he thinks, that which “best captures what it would take for something to perfectly deserve the name ‘value’” (Lewis 1989: 136). But this perfect deserver leaves us with skepticism. However, “there are plenty of imperfect deservers of the name” (136). Strictly speaking, nothing shall get the name without deserving it perfectly. Strictly speaking … genuine values would have to meet an impossible condition, so it is an error to think there are any. Loosely speaking, the name may go to a
92
r ic h a r d joyc e
claimant that deserves it imperfectly. Loosely speaking … [t]here are values, lots of them. (136–37)
He goes on: What to make of the situation is mainly a matter of temperament. You can bang the drum about how philosophy has uncovered a terrible secret: there are no values! … Or you can think it better for public safety to keep quiet and hope people will go on as before. Or you can declare that there are no values, but that nevertheless it is legitimate – and not just expedient – for us to carry on with value-talk, since we can make it all go smoothly if we just give the name of value to claimants that don’t quite deserve it … Or you can think it an empty question whether there are values: say what you please, speak strictly or loosely. When it comes to deserving a name, there’s better and worse but who’s to say how good is good enough? Or you can think it clear that the imperfect deservers of the name are good enough, but only just, and say that although there are values we are still terribly wrong about them. Or you can calmly say that value (like simultaneity) is not quite as some of us sometimes thought. Myself, I prefer the calm and conservative responses. But as far as the analysis of value goes, they’re all much of a muchness. (137)
Perhaps the moral error theorist has a different temperament from that of Lewis. I confess that I have long been drawn to the error theoretic version of moral skepticism, and it would be untruthful to claim that this is entirely the product of having been persuaded by solid philosophical argumentation. Maybe it is a mischievous streak in me; perhaps it is an excitement about striding into uncharted intellectual territory having cast off familiar assumptions. Who knows? I wouldn’t attempt to publicly analyze the antecedents of my skeptic-leaning temperament any more than Lewis sought to openly scrutinize his own conservative preferences. The important points are (i) that temperaments opposed to Lewis’ exist (exhibit A = myself), and (ii) that Lewis thinks our temperaments determine “what to make of the situation.” Thus he makes one thing of the situation – moral naturalism – while I make another – error theoretic skepticism. Is this a reasonable conclusion? And, if so, then where (if anywhere) do we go from here? Lewis is certainly not saying that all there is to the debate between the moral naturalist and the moral error theorist are their respective temperaments. Both parties must first have some decent arguments on their side. For example, we can assume that no amount of “temperament” favoring the existence of magic will make a naturalistic endorsement of magic reasonable. Likewise, no amount of “temperament” opposed to the existence of big hairy spiders is going to make doubting their existence
Metaethical pluralism
93
epistemologically acceptable. From the perspective of global naturalism, some things clearly exist (spiders) and some things clearly do not (magic). Where things get interesting is in the gray area, where there are decent arguments both for and against the existence of the item, and here, Lewis thinks, temperaments might matter. Let us approach the problem with reference to Ramsey sentences. Suppose we are wondering whether Φ exists. We construct a list of platitudinous desiderata of what we think Φ is like: “Φ is P1,” “Φ is P2,” etc. (Of course, the sentences do not need to be simple atomic predications, but it makes expression easier.) Let us say we come up with twenty such desiderata. We then conjoin these sentences, replace all mention of Φ with the same variable, and bind that variable with an existential quantifier: ∃x (x is P1 & x is P2 & … x is P20).
This allows us to say that Φ exists if and only if the existential claim (the Ramsey sentence) is true. There are two ways in which it can be false: if there is nothing that satisfies all twenty predicates, or if there is more than one thing that satisfies all twenty predicates. (See Lewis 1970.) But one should not be too strict about this, for there is a requirement to accommodate the fact that we can be mistaken about certain qualities of Φ without it automatically following that Φ does not exist. We gave the name “earth” to the thing we live upon and at one time reckoned it flat (or at least a good many people reckoned it flat); but the discovery that the thing we live upon is a big ball was not taken to be the discovery that we do not live upon earth. It was once widely thought that gorillas are aggressive brutes, but the discovery that they are in fact gentle social creatures was not taken to be the discovery that gorillas do not exist. Examples abound. We make mistakes, big and small, without facing skepticism at every turn. Applying this to Lewis’ argument: one of the desiderata for “value” pertains to non-contingency. (I shan’t pause for further specificity.) But nothing that satisfies the other desiderata (some of which can be systematized into that which we desire to desire) also satisfies non-contingency, and this is why Lewis claims that “genuine values would have to meet an impossible condition.” But, he thinks, there are imperfect claimants – ones that satisfy nearly all that we want. Just as a spherical earth was an imperfect claimant of all that we might have wanted to say about the earth (but close enough to avoid earth skepticism), and gentle gorillas were imperfect claimants of all that we might have wanted to say about gorillas (but close enough to avoid gorilla skepticism), so too, Lewis thinks, his dispositional
94
r ic h a r d joyc e
theory of value is an imperfect claimant of all that we might have wanted to say about value, but close enough to avoid moral skepticism. But Lewis realizes that his dispositional theory is not close enough to silence all debate on the matter; it remains in a gray area. And the grayness persists because we have no methodology for making decisions concerning indeterminate cases. Lewis writes: “What it takes to deserve this name, not perfectly but well enough, was never settled” (1989: 136); and: “When it comes to deserving a name, there’s better and worse but who’s to say how good is good enough?” (137). Who indeed? Elsewhere I have toyed with the idea that what determines this matter is a complex counterfactual about how we would, if called upon, as a matter of fact decide (Joyce 2006: 201).3 But this idea was never meant to help us out as a decision procedure, and it is, in any case, problematic. Who, exactly, is this “we,” and in what circumstances are we being called upon to make this decision? Since I harbor no optimism that in making such decisions we collectively or individually follow any particular rules, overt or tacit, I am inclined to think that our pronouncements on such matters can be influenced by the most arbitrary of stimuli (e.g., the plot of a popular movie). Thus in circumstances A we will decide that we have had a false belief about Φ, correct that error and carry on believing in Φ (i.e., we will be Φ naturalists); while in circumstances B (which are not dramatically different from A) we will decide that our belief in the existence of Φ has been a huge mistake (i.e., we will be Φ skeptics). This to some extent problematizes the counterfactual “If we were to decide on the matter, we would decide thus-and-so.” Subsequently, I suggested that what determines the question is how we use the concepts (Joyce 2006: 201; 2007a: 65). Roughly: if concept Φ – understood as satisfying twenty desiderata – has been used in ways U1, U2, … Un, but the best imperfect claimant (call it Φ*, which satisfies, say, eighteen of those desiderata) cannot be put to the same range of uses, then this is sufficient grounds for denying that Φ* is “close enough” to count as a revision of Φ. I continue to think that this appeal to usage promises to make some headway toward solving some of the Lewisian Is-it-close-enough?-type questions. Regardless of whether one’s temperament favors Lewis’ “calm and conservative” preference for dispositional moral naturalism, if it turns out that the dispositional property in question just isn’t going to 3 I have on occasion made what I take to be the equivalent point via a thought experiment concerning translation. (See Joyce 2001: 3, 26–7; 2006: 71.)
Metaethical pluralism
95
pull its pragmatic weight in everyday life – if it can not do the work of a full-blooded all-you-could-want moral property – then this would count decisively against this form of moral naturalism. Likewise, irrespective of whether one’s mischievous temperament favors error theoretic moral skepticism, if it turns out that Lewis’ dispositional property can be pressed into service for all the practical roles that moral concepts play, then the day would go to the moral naturalist and his imperfect claimant. One interesting and possibly surprising consequence of conceptualizing the problem in this manner is that it makes the debate between the moral naturalist and the moral error theorist at bottom an empirical debate. We ask, first, “What is morality used for?” Then we ask “Could concept such and such [the imperfect claimant] be used in this manner?” Both are difficult and complex matters, but they are both empirical questions, and the former, at least, seems reasonably tractable. Yet the remaining problems are more than just the difficulty of our gaining epistemic access to these empirical truths. The problem is that the indeterminacy simply reiterates in a new place. Suppose we have used concept Φ for ten purposes – U1, U2, … U10 (idealizing horribly here, of course) – and suppose that the best imperfect claimant (call it Φ*) can be used in, say, eight of those ways. We can not use Φ* for everything that we used to use Φ for, but we can use it for most things. Well, is that close enough? I feel at this point we can only reiterate Lewis’ question: “Who’s to say?” If we can find no good answer, then we have made not only the surprising discovery that the debate between the moral naturalist and the moral skeptic bottoms out in empirical disputes, but a second surprising discovery that there may be no fact of the matter about who is correct. 5.3 C a r n a p’s pr ag m at i s m Lewis is evidently content to live with this indeterminacy between moral naturalism and moral skepticism. But we shouldn’t give in just yet; there is an obvious avenue to survey: an appeal to pragmatism. If there is an indeterminacy in what we mean by “value” (say) – in that both the moral skeptic and the moral naturalist make defensible claims about what is necessary or sufficient for something to be a value – then let the matter be decided by which is most practically expedient; let us choose a precisification on the basis of which best serves our purposes. Many philosophers, it seems to me, are drawn to this method of selecting sides when indeterminacy threatens, and, moreover, the common assumption regarding the present case is that an appeal to pragmatism
96
r ic h a r d joyc e
will favor the moral naturalist over the skeptic. My aim is to call into question both moves. First, the appeal to pragmatism is fraught with difficulties and does not obviously represent an escape route from indeterminacy. Second, I think the assumption that an appeal to pragmatism (were it to prove workable for this purpose) would favor the moral naturalist can be subjected to pressure. But before arguing these claims I want to bring the non-cognitivist on to the stage, and I will do so via a discussion of Rudolf Carnap. Carnap’s views are useful for our purposes since as well as advocating non-cognitivism he espouses an interesting pragmatic methodology. That the Carnapian notion of explication is relevant to our discussion is immediately evident from the characterization Carnap gives it in Logical Foundation of Probability: “By the procedure of explication we mean the transformation of an inexact, prescientific concept, the explicandum, into a new exact concept, the explicatum” (Carnap 1950a: 3). The test of a good explication is not whether it is “true” or “correct” (Carnap says such claims make “no good sense” [4]); but rather “whether the proposed solution is satisfactory” (4) – where this satisfactoriness is in part a pragmatic notion, centrally depending on whether the proposed explicatum is fruitful and simple. The pragmatism embodied in his notion of explication is plain elsewhere in Carnap’s philosophy. In “Empiricism, Semantics, and Ontology” (1950b) he analyzes many traditional philosophical questions as “external questions.” The question “Does the world exist,” for example, is really a practical question of whether we should adopt the “thing language” according to which objects exist at a space and time within the world. In determining the practical payoff, “[t]he efficiency, fruitfulness, and simplicity of the use of the thing language may be among the decisive factors” (208). He says the same about the external questions “Do propositions exist?,” “Do numbers exist?,” and so on. According to Carnap, these questions should be interpreted as: “Are our experiences such that the use of the linguistic forms in question will be expedient and fruitful?” (213). Carnap thus provides us two ways that we might try to break out of the indeterminacy surrounding “value” (and other moral terms). First, we might try to explicate the notion, imposing precision on an indeterminate concept in a manner guided by expediency. Second, we might treat the question “Do values exist?” as an external question, deciding whether to employ the “value language” on the basis of its utility as an instrument. Before examining these possibilities, we should have Carnap’s own metaethical view before us. His years as a leading member of the Vienna
Metaethical pluralism
97
Circle had made a moral skeptic of him – but of the non-cognitivist rather than the error theoretic variety. By 1935 he was staunchly advocating a prescriptivist version of non-cognitivism: “Most philosophers have been deceived [by grammatical form] into thinking that a value statement is really an assertive proposition and must be therefore true or false … But actually a value statement is nothing else than a command in a misleading grammatical form … It does not assert anything” (Carnap 1935: 24–25). His non-cognitivism had first become evident in 1929, in a lecture delivered to the Bauhaus in Dessau, where (according to fragmentary notes) he claimed: “Valuation (Wertung) is not the cognition of a fact but a personal attitude” (cited in Mormann 2007: 133). Interestingly, in his Aufbau, a younger Carnap had espoused a very different position, where values are a part of the world order, whose presence may be known and asserted (see section 152 of Carnap 1967 [1928]). In what follows, however, I shall take the mature Carnap’s non-cognitivism as representative of his view. (See Mormann 2007, for discussion of Carnap’s change of mind.) When Lewis outlines the range of options (in the long passage quoted earlier), the non-cognitivist is noticeably absent; all that is on his radar is a continuum of positions between the moral error theory and moral naturalism. It is clear why this is so. Lewis’ examination of “value” as a verb (roughly: desiring to desire) has provided him with license to treat “value” as a noun (roughly: that which we are disposed to desire to desire), which in effect is license to exclude non-cognitivism. Thus the question that exercises him at the close of the paper is whether the world provides a property that is close enough to deserve this noun (albeit imperfectly). The non-cognitivist, by contrast, will not let Lewis get that far. The noncognitivist endorses an account of “value” as a verb that does not provide license for treating “value” as a noun. Focusing on “valuing” as a linguistic act rather than a kind of mental attitude, Carnap tells us that value utterances are really commands, in which case the question of to what kind of entity these value utterances refer simply does not arise. When one says “X is good,” Carnap thinks, one is saying something along the lines of “Pursue X!” The goodness has evaporated on this analysis. Asking whether the world provides anything close enough to deserve the noun “goodness” is, in Carnap’s opinion, a pseudo-question, because concepts like goodness are pseudo-concepts and therefore not fit for analysis. If there is no concept of value, then a fortiori there is no such concept to be made precise via explication. That is not to say, however, that explication can play no role in Carnap’s metaethics. The concept of value accused of being an inexplicable pseudo-concept is the one associated
98
r ic h a r d joyc e
with the noun “value.” But Carnap certainly believes in the human activity of valuing, and the related concept (the one associated with the verb “to value”) will not also be accused of “pseudo-ness.” Moreover, Carnap also believes that human valuing will often take the form of issuing statements, and thus there is a noun (“value statement”) whose associated concept is available for analysis and, if necessary, explication. On the only occasion that Carnap does explicitly sketch an explication of something normative, the explicandum is, indeed, “value statement” (Schilpp 1963: 1009 ff.). Now, it is not my intention to discuss Carnap’s proffered explication in detail, but rather make some general comments about it. First, it is interesting to note that he does not take himself to be deriving the distinction between non-cognitive/evaluative utterances and descriptive/factual statements from any examination of natural language practices; rather, he is speaking of “possible kinds of meanings and the relations between these meanings” (Schilpp 1963: 1003). His aim is to establish that a noncognitive language is possible, and “to use it as a basis for the philosophical discussion of value problems” (1003). This last comment reveals that Carnap’s eye is, as usual, on the pragmatic payoff, but one should be aware of his restricted view on this matter. The “fruitfulness” of explication is understood in terms of the explicatum being “brought into connection with other concepts on the basis of observed facts; in other words, the more it can be used for the formulation of laws” (Carnap 1950a: 6). Thus, it is not any general psychological kind of practical usefulness that is being claimed for non-cognitivism; indeed, Carnap opines that whether one is a non-cognitivist or a moral naturalist “is relatively unimportant in its influence of practical life” (in Schilpp 1963: 82).4 When he does claim some advantage for his non-cognitive explication of a value statement over any cognitive rival, it is in terms of a very slight (almost, one might say, unnoticeable) additional discriminatory power in the former language: the non-cognitive language can provide an interpretation of a certain sentence of the cognitive language, but not vice versa (Schilpp 1963: 1004). However, the argument is unpersuasive, to say the least. Among its weaknesses is the fact that it claims this advantage regarding just one type of comparison; but who knows what 4 Carnap does think that clearly demarcating value questions from factual questions is practically important, and that the failure to do so “leads to confusions and misunderstandings in discussion of moral problems in personal life or in political decisions” (in Schilpp 1963: 81) – but that is not the same as thinking that the distinction between non-cognitivism and cognitivism makes a practical difference.
Metaethical pluralism
99
advantages the cognitivist language might enjoy over the non-cognitive language if we looked further afield?5 Let us conclude, then, that Carnap leaves the matter undecided. And he would, I think, agree with and even welcome this diagnosis. His Principle of Tolerance allows metalanguages to compete freely for our allegiance, to be judged not according to “truth” but according to fruitfulness.6 Permit the non-cognitivist and the cognitivist to advocate their respective languages; Carnap is tolerant of the competition. The language that best suits our theoretical purposes will be the eventual victor. “Let us grant to those who work in any special field of investigation the freedom to use any form of expression which seems useful to them; the work in the field will sooner or later lead to the elimination of those forms which have no useful function” (1950b: 221). I will sum up before moving on. I set out to explore the possibility of an irresolvable indeterminacy between moral naturalism and moral skepticism. Lewis located one potential node of indeterminacy: between moral naturalism (of a certain kind) and error theoretic moral skepticism. This is an indeterminacy over “value” as a noun. But before we get that far, we face the possibility of another node of indeterminacy: between cognitivism (which includes all forms of moral naturalism) and the other kind of moral skepticism, non-cognitivism. This can be thought of as indeterminacy over “value” as a verb. (Quite possibly there is indeterminacy in the rival non-cognitivist construals of valuing, too – between, say, Hans Reichenbach’s volitionism and A. J. Ayer’s emotivism.) Later I will further discuss the possible indeterminacy between naturalism and skepticism, but first I will investigate the potential undecidability of the debate between cognitivism and non-cognitivism. 5. 4 C o g n i t i v i s m v e r s us non- c o g n i t i v i s m Carnap, as we saw, invites open competition with his cognitivist rival. But the mere fact that one is sufficiently undecided between two theories of X (or sufficiently sportsmanlike) to be willing to allow them to “compete” 5 A couple of other weaknesses with Carnap’s arguments are the following. First, he seems to be concerned only with showing how purely non-cognitive sentences (“pure optatives,” as he calls them) are possible; but the moral cognitivist need not deny this, but rather simply deny, that moral utterances are instances of purely non-cognitive sentences. Second, Carnap has only a very restricted kind of cognitivism in his sights when he performs this comparison. 6 They are metalanguages because the Principle of Tolerance is applied after sentences in the material mode of speech have been translated into sentences in the formal mode.
100
r ic h a r d joyc e
in some manner for victory hardly means that there is a substantive indeterminacy about X. For there to be indeterminacy it must be the case that (A) this competition yields no winner, and (B) no other form of decision procedure is available. Could there really be no fact of the matter about metaethical cognitivism versus non-cognitivism? One might think that of course there is a fact of the matter. This debate is, essentially, about what we are doing when we value something, and the act of valuing is a phenomenon available to our investigation. This is why although there are plenty of advocates of an error theory about values, there are no error theorists about valuing.7 We do not doubt that valuing occurs, and this type of action has certain features which determine the answers to various questions we might ask about the phenomenon. So surely the debate between the cognitivist and the non-cognitivist must have an answer, even if we’re not sure yet what it is. But this line of reasoning is mistaken. Valuing may well be an actual phenomenon displayed before us, available to our scrutiny, but the phenomenon will yield specific answers only if we ask specific questions. There are different ways of framing the cognitivism/non-cognitivism question, but the standard way is: “Are moral judgments assertions?” Here valuing is treated as a linguistic activity, and the question concerns what kind of speech act constitutes the activity. Specific enough? Not necessarily. Questions arise as to what is meant by “assertion.” There are significant differences among the views of assertion advocated by C. S. Peirce (1934a, 1934b), Michael Dummett (1959), J. L. Austin (1962), John Searle (1969), Michael Slote (1979), Robert Brandom (1994), and Timothy Williamson (2000) – to name just a few. Can these differences be settled? Perhaps not. Perhaps different accounts of assertion work better in different theoretical environments, with there being no decisive facts to appeal to in order to settle either which is the best account of assertion overall or which is the best account of assertion to apply to the moral case. Many accounts of assertion state or imply that the person making an assertion aims at the truth, or aims to state a fact (see Dummett 1981: 300; Williams 1966). So when someone claims “Stealing is wrong,” is he or she stating a fact? It depends what you mean by “fact” – of which, it comes as no surprise, there are competing theories. One might distinguish between 7 Unless, of course, it follows from the endorsement of some grander skepticism, encompassing doubt about cats and dogs, furniture, other people, etc. Even an error theorist about all mental phenomena need not be an error theorist about valuing, since one can treat valuing as a species of linguistic act or behavioral response.
Metaethical pluralism
101
“fact” in a metaphysically robust sense, and “fact” in some more minimalist sense. This, in turn, would give us a distinction between “assertion” in a robust sense and in a minimalist sense. Indeed, we find exactly this kind of distinction appearing in metaethical debates over quasi-realism. The quasi-realist eschews the metaphysics of moral realism but thinks that language with all the trappings of realism is still permissible (Blackburn 1984, 1993). Thus, even if there is no metaphysical property of wrongness, the quasi-realist is happy to carry on claiming not only “Stealing is wrong,” but “‘Stealing is wrong’ is true,” “There is a property of wrongness (and stealing has it),” “Stealing is wrong regardless of my attitude,” “I hereby assert that stealing is wrong,” and “I really, really mean it.” A typical quasi-realist (non-cognitivist) view is that while we can claim all these things (i.e., truth, fact, property, assertion) for morality in a minimalist sense, we cannot do so in a robust sense. In this way the quasi-realist hopes to vindicate the realist trappings of moral discourse while continuing to distinguish the position from moral realism. This situation is not simply one of philosophers arguing over the correct analysis of a concept whose extension all will substantially agree to; rather, the extension may very well differ significantly depending on which conception is preferred. It is possible that the question “Are moral judgments assertions?” should receive a positive answer if using a minimalist conception of assertion but a negative answer if using a robust conception. But nor is this situation simply one of an ambiguous concept, for the parties to the dispute will usually insist that their preferred conception is uniquely correct. We will not, for example, typically find moral realists accepting that the minimalistic notion of assertion is legitimate for certain contexts; rather, they will generally despise the very idea (along with the minimalistic versions of truth, fact, etc.). Similarly, the quasirealist may well claim that the robust version of assertion is a mere philosopher’s fancy, and perhaps not a well-formed idea at all. Perhaps this dispute could one day be settled; perhaps one of the disputants is, after all, uniquely correct. But it is also possible (though I am not arguing that it is the case) that neither is uniquely correct; perhaps there is nothing that determines the correctness of one conception over the other; perhaps the concept of assertion is simply indeterminate across the relevant cognitivist/non-cognitivist gap. I want now to return to the possibility of settling this dispute by an appeal to convenience and expedience, though this time I propose to understand these matters (albeit vaguely) in a broader, more psychological, sense than Carnap intended. My worries with this suggested
102
r ic h a r d joyc e
decision procedure arise before we even begin to wonder about what the practical upshot of moral non-cognitivism is. My concern arises because the question “Expedient to whom?” seems a perfectly fair yet unanswered response. Many pragmatist suggestions speak in a cavalier fashion about “what is useful to our practical purposes” without pausing to wonder to whom the “our” refers; the common background assumption is that there is a convergence in practical requirements. But this assumption is, upon reflection, wildly implausible. What if non-cognitivism suits some people while cognitivism suits others? Then one precisification of “assertion” (say) will be practically expedient for some, while an opposed precisification will be practically expedient for others. This is reminiscent of an old complaint against the Pragmatist school of philosophy, one that Bertrand Russell heatedly made in his essay on the topic (and on every possible occasion thereafter): “One gathers … that a Frenchman ought to believe in Catholicism, an American in the Monroe Doctrine, and an Arab in the Mahdi” (Russell 1910: 97). That Russell had an accurate or fair image of his target is unlikely (see Haack 1976), but here my worry concerns nothing so grand as Pragmatism as a school of philosophy or a theory of truth. My target is just one kind of appeal to pragmatic considerations in the hope of settling a philosophical dispute, and here, it seems to me, Russell’s question remains a live one. Regarding the practical upshot of non-cognitivism’s being true or false, is it really reasonable to suspect that there will be divergence among individuals’ interests? Alasdair MacIntyre’s analysis of the Bloomsbury Group might serve as an illustration (MacIntyre 1981). MacIntyre interprets the Bloomsbury circle as developing a peculiar moral discourse: surrounded by a community that employed moral vocabulary in an assertoric way, the members of the Bloomsbury Group (unwittingly) did otherwise. Impressed with G. E. Moore’s elevation of beauty, love, and knowledge to the pedestal of “intrinsic moral values,” this group of intellectuals used moral language as a manipulative tool to persuade others of the delicacy (and thus superiority) of one’s own aesthetic sensibility. John Maynard Keynes recalls that about these matters it was “useless and impossible to argue” (Keynes [1949] in Rosenbaum 1995: 87). He goes on: In practice, victory was with those who could speak with the greatest appearance of clear, undoubting conviction and could use the best accents of infallibility. Moore at this time was a master of this method – greeting one’s remarks with a gasp of incredulity – Do you really think that, an expression of face as if to hear such a thing reduced him to state of wonder verging on imbecility, with his
Metaethical pluralism
103
mouth wide open and wagging his head in the negative so violently that his hair shook. Oh! He would say, goggling at you as if either you or he must be mad; and no reply was possible. (87)
MacIntyre’s thesis that non-cognitivism was true of the Bloomsbury Group while not true of the wider populace is an interesting but knotty claim – but it is not, in any case, our present concern. Rather, I am suggesting that it might have suited the Bloomsbury Group’s purposes if non-cognitivism were true – because, roughly, it would validate their practices – while not serving the purposes of others. Thus the proposal that we should settle any indeterminacy between non-cognitivism and cognitivism – and thus between moral skepticism and moral naturalism – by reference to “what is in our practical interests” is seen to be seriously problematic. 5.5 T h e be n e f i t s of be i ng hor r i bly w rong The pragmatic payoffs of non-cognitivism versus cognitivism may be inscrutable and divergent, but surely we can hope to do better at the other node of potential metaethical indeterminacy: moral naturalism versus the error theory? Here the naturalist might feel more confident that there is nothing to be said, pragmatically, in favor of the error theoretic position. Here the naturalist might feel satisfied with Lewis’ diagnosis, assured that if all that lies between moral naturalism and moral skepticism is “temperament,” then he, the naturalist, may declare victory. However, in my opinion, if the argument has come to this point, the moral naturalist has no particular grounds for complacency, for there are, I think, considerable pragmatic claims to be made in favor of the moral error theory. The moral error theory tells us that we have been massively mistaken about the world, about ourselves, and about the relation between the two. Finding out that one has been massively mistaken is, it can be assumed, an uncomfortable state, and being uncomfortable is unpleasant. Finding out, moreover, that our mistake has been as great as the moral error theorist would have us believe – ubiquitous across all human societies, through all history, pervading nearly every aspect of our social existence – is likely to make us uncomfortable to the point of positive distress. But one needs to look deeper. Being uncomfortable can be a useful state (after all, pain serves a good biological purpose), and distress can, ultimately, be instructive. By way of leading you in to my thinking, consider what we have learned from experimental psychology about confabulation. There are
104
r ic h a r d joyc e
various kinds of cognitive impairment (to memory, to perception, etc.) for which the subject will compensate by creating a false narrative to “fill in the gaps.” Someone suffering from Korsakoff’s syndrome fails to admit, even to herself, the severe memory loss she suffers, but rather invents elaborate stories to cover her confusion. Asked why she is in the hospital, for example, the patient might genuinely assert that she is visiting someone or applying for a job there. Sometimes stroke patients with partial paralysis will deny the paralysis, inventing sincere excuses for why that part of their body cannot perform actions. (See Hirstein 2005; Schnider 2008.) Fascinating as these kinds of case study are, they are really just the lurid side of the phenomenon of confabulation, much of which is far more routine and commonplace. We all superimpose justifications and explanations onto our actions when the real sources of motivation are hidden from us; we all string together coherent life narratives from fragmented and distorted memories; we all confabulate. (See Wheatley 2009.) This is a disconcerting truth when its full extent is apprehended (“I am confabulating nearly all the time!”). It forces us to adopt a new perspective on who we are and how we work. Experimental psychology can bring us to see that our intuitive everyday image of ourselves and our lives is, in many ways, badly mistaken.8 Yet, I claim, this unsettling apprehension is often a good thing. It is good not merely because it is true, but because it is useful. Of course, we tend to compartmentalize this analytical awareness, and quickly return to confabulating as soon as we relax our attention. And that is fine. What I am speaking in favor of is the value of regularly returning to the critical perspective from which we realize that we are more opaque and baffling beings than everyday life presupposes. Consider this scenario. You are riding in a bus through a city, absentmindedly watching pedestrians and street scenes, when the thought strikes you: “What odd creatures we are.” And suddenly you undergo a Desmond-Morris-style gestalt shift: you think of the pedestrians under the description “hairless upright social apes” – you literally see them as hairless upright social apes. Suddenly the city appears like a swarming nest of primates, commonplace urban activities come to seem like bizarre arbitrary rituals, the projects and concerns that ordinarily occupy our minds are seen as weird and futile. And perhaps the oddest thought of all is: “Hey, I’m one of those naked apes, too!” 8 Of course, it is not just the phenomenon of confabulation that shows this. The broader claim that experimental psychology has revealed that we are, in a multitude of ways, routinely mistaken about what is really going on with our motivations, emotions, and belief-formation processes, is far less contentious and far too well documented to require citations.
Metaethical pluralism
105
When one thinks along these lines, it is not that one all of a sudden forms new beliefs, for we all know (those of us that are sensible, at least) that we are hairless upright social primates, with bodies and minds shaped by Pleistocene Africa.9 But this knowledge is compartmentalized during our everyday lives; it takes an effort to attain the critical distance to feel the truth of such beliefs, to be dumbstruck at the sheer strangeness of it all. And that feeling – a feeling of alienation from the customary images of oneself and the society one inhabits – is, I contend, a valuable perspective to adopt. Like a full-blooded encounter with the aesthetic sublime, it is not an entirely pleasant experience. The sublime can make one weak in the knees with confusion and awe. Yet we seek out such encounters (sometimes) because in that feeling we sense that we are in touch with something authentic, as if we have stepped back from our familiar humdrum concerns and tasted the enormity of the universe and the fragility of our place in it. This is the benefit I would like to advertise of seeing ordinary beliefs as horribly wrong. It is good to be epistemologically shaken. It reminds us of how shockingly ignorant we are and how mysterious everything really is. It is both a corrective to epistemic complacency and a spur to intense reflection and inquiry. Discovering that everyone is horribly wrong about something fundamental to our lives is not just sobering and valuably humbling, but intellectually exciting. Endorsing an error theory puts hairs on your chest, epistemologically speaking. Well, that’s just me. If the last few paragraphs fail to speak to you, then that’s OK; that, indeed, is my central point. I am not claiming that every person would benefit from recognizing his/her profound error and therefore we should all be error theorists. Remember that the present dialectic presupposes that there are already some decent arguments in favor of the moral error theory (and decent arguments for moral naturalism), so the pragmatic considerations just raised are supposed to supplement solid arguments, not stand in place of them. But nor am I advancing the amended thesis that every person would benefit from recognizing his/her profound error and therefore if we have decent arguments in favor of the error theory then we should all be error theorists. I am in fact nowhere claiming that every person would benefit from recognizing his/her 9 This claim that we have “minds shaped by Pleistocene Africa” should not be over-interpreted as an endorsement of any strong form of evolutionary psychology. The prehistorically forged human mind may be designed to be a plastic and generalized problem solver.
106
r ic h a r d joyc e
profound error. Perhaps some people would and some people would not. My point is (again) that even in the case of moral naturalism versus the moral error theory, the pragmatic cost–benefit analysis is complex, and there is no discernible reason to expect convergence. I can imagine critics complaining that even if there do exist such intellectual benefits to be gained from an apprehension of the depths of human epistemic fallibility, surely they pale beside the losses that will be incurred from an endorsement of a moral error theory. Surely (the complaint goes) a belief in the moral error theory will lead to a breakdown of good citizenship, to a loss of motivation to enter sincerely into cooperative ventures, to cheating one’s friends when the chips are down, to selfishness and debauchery, and so forth. But these vague fears are expressed far more often than any scrap of evidence in their favor is offered. While I agree that moral belief contributes to an individual’s “cooperative motivation” (to choose an umbrella term) (see Joyce 2006: ch. 4), the idea that moral belief is the only thing that does so is a hypothesis as doubtful as it is depressing. And even if the hypothesis were true, the claim that a breakdown of cooperative motivation is the inevitable outcome of embracing the moral error theory implies the similarly doubtful hypothesis that moral belief is the only thing that can motivate cooperation. No, there is no evidence that moral skepticism leads to crimes and misdemeanors. I have met many moral error theorists, and am yet to encounter one who wasn’t perfectly civilized or was any less trustworthy than anyone else. By contrast, if one reflects upon the worst mass criminals that humanity has produced, one is likely to encounter not skeptics but individuals moved by moralistic fervor. (See Garner 2010; Hinckfuss 1987.) Of course, these observations don’t settle the matter. It is entirely possible that, despite my anecdotal evidence, in fact moral error theorists are always less flourishing and always inferior citizens when compared to moral believers. Or perhaps it goes the other way round. What I am seeking to establish is that we shouldn’t jump to any conclusions regarding these empirical matters, and in particular shouldn’t assume that there will be convergence among individuals one way or the other. Though there may well be practical costs to being a moral error theorist, I am content on this occasion to attempt to publicize some of the neglected practical benefits. The idea that our interests are best served by a methodology that always favors theories which mesh with our common sense is, for my money, shortsighted and disappointing. Perhaps some individuals require that kind of constant reassurance – perhaps many do – but not all do. Some
Metaethical pluralism
107
find the prospect that common sense is massively mistaken an object of contemplation that is, while anxiety-provoking (partly because anxietyprovoking), profoundly rewarding. Given the choice between a theory that vindicates some aspect of common sense and a theory that overturns it, some people will find the latter more valuable. I cannot resist here quoting Bertolt Brecht’s gloriously scathing assessment: “I’m not writing for the scum who want to have the cockles of their hearts warmed” (Willett 1957: 14). 5.6 Qu i n e’s s e c ta r i a n i s m a n d e c u m e n ic a l i s m I have argued that the debate between the moral naturalist and the moral skeptic – which I broke down into (i) that between the moral cognitivist and the moral non-cognitivist, and (ii) that between the moral naturalist and the moral error theorist – may be undecidable. Key concepts like assertion or value may be sufficiently indeterminate to permit all aforementioned theoretical possibilities. The prospect of avoiding this conclusion by an appeal to pragmatism has been explored and rejected; pragmatic solutions presuppose an implausible convergence in practical interests. It should be stressed that I have not argued that we do face this indeterminacy, only that we might. For all I have said here, it is possible that the moral naturalist will triumph tout court. Or perhaps the moral skeptic will.10 But let us suppose for now that these or similar arguments favoring undecidability are sound. Where do we go from here? Having discussed Lewis and Carnap, it seems apt to end with the philosopher who best connects them: W. V. Quine.11 Quine famously argues for the underdetermination of scientific theory: that logically incompatible global theories may be equally consistent with all possible evidence and equally satisfactory with respect to internal theoretical virtues (e.g., simplicity and clarity) (Quine 1960; 1975). As to the “Where do we go from here?” question, Quine vacillated. In certain moods he espoused a sectarian response, according to which we should continue to endorse our favorite theory as robustly as ever, rejecting all alternatives as false or 10 Readers might be aware that I have argued for the error theory in the past (Joyce 2001, 2006, 2011a, 2011b), and I dare say that in the future I will return to doing so. Despite my sympathetic attitude towards Lewis’ pluralistic position, I have not officially conceded that his “imperfect claimant” (or any other contender) is close enough to warrant our endorsement as a morality. But even if I were to make that concession, continuing to argue for an error theory at the level of metaethics is consistent in every way with favoring pluralism at the metametaethical level. (See note 12 below.) 11 Carnap taught Quine; Quine taught Lewis.
108
r ic h a r d joyc e
meaningless (Quine 1981: 21; 1986). But on other occasions he preferred an ecumenical response, whereby we accept that indeterminacy reveals all satisfactory though incompatible theories to be true (Quine 1981: 29; 1989). We face a similar decision regarding metaethical indeterminacy. Suppose, like Lewis, one is inclined to endorse a form of moral naturalism. And suppose the possibility that I have discussed turns out to be a reality: that another person could, with equal legitimacy, endorse some form of moral skepticism. What attitude should the first person take – to push on with moral naturalism, declaring moral skepticism bankrupt (i.e., sectarianism), or perhaps to take a more tolerant view of alternatives (i.e., ecumenicalism)? The moral skeptic, of course, faces the same dilemma. Quine became aware of this conflict in his work, and tended to plump for sectarianism. He even inserted comments into later editions of Theories and Things in order to distance himself from the ecumenical noises from his earlier self. He diagnoses his own indecisiveness on this matter by saying: “The fantasy of irresolubly rival systems of the world is a thought experiment out beyond where linguistic usage has been crystallized by use” (Quine 1990: 100). Faced with the nodes of indeterminacy highlighted in this paper, I would counsel neither sectarianism nor ecumenicalism in particular, but rather what might be called “metaethical ambivalence.”12 This perspective begins with a kind of metametaethical enlightenment. The moral naturalist espouses moral naturalism, but this espousal reflects a mature decision, by which I mean that the moral naturalist doesn’t claim to have latched on to an incontrovertible realm of moral facts of which the skeptic is foolishly ignorant, but rather acknowledges that this moral naturalism has been achieved only via a non-mandatory piece of conceptual precisification. (This describes Lewis’ tolerant view.) Likewise, the moral skeptic champions moral skepticism, but this too is a sophisticated verdict: not the simple declaration that there are no moral values and that the naturalist is gullibly uncritical, but rather a decision that recognizes that this skepticism has been earned only by making certain non-obligatory but permissible conceptual clarifications. 12 Broadly parallel to David Wong’s moral ambivalence (2006: 20 ff.). I take it that just as Wong’s moral ambivalence is ambivalence about moral systems – an ambivalence that occurs at the metaethical level – so too my metaethical ambivalence is ambivalence about metaethical theories – an ambivalence that occurs at a metametaethical level. For insightful discussion of doxastic ambivalence, see Roorda (1997).
Metaethical pluralism
109
This enlightened awareness of the legitimacy of one’s opponent does not, however, suffice for metaethical ambivalence. For the next step we can turn again to Quine, recalling that even when advocating sectarianism he does not encourage adherents of one scientific theory to blinker themselves uncritically to other possibilities. What is at stake for Quine is the permissibility of applying the truth predicate to alternative theories to one’s own; but even in those moods when he claims that one can call only one’s own favored theory “true,” he still allows that one can “oscillate between … theories for the sake of added perspective from which to triangulate on problems” (Quine 1990: 100). And this is, perhaps, indicative of the most sophisticated pluralism of all. The enlightened moral naturalist does not merely (grudgingly) admit that the skeptic is warranted in his or her views, but is willing sometimes to adopt the skeptical position in order to gain the insights that come from recognizing that we live in a world without values. And the enlightened moral skeptic goes beyond (grudgingly) conceding that moral naturalism is reasonable, but will sometimes assume that perspective in order to gain whatever benefits come from enjoying epistemic access to a realm of moral facts. Such a metaethical ambivalence will require a temperament that favors neither the conservative nor debunking attitudes mentioned by Lewis, but rather an intellectual courage to cut oneself loose from the comforting familiarity of assuming that there is always a fact of the matter.
ch apter 6
Moral naturalism and categorical reasons Terence Cuneo
Suppose you recognize that lying to your spouse simply to save face is wrong. Under what conditions does the wrongness of this act provide you with a reason not to lie? Proponents of the Humean theory of reasons answer that it does so only when and because you have desires of the right sorts. According to the Humean view, the fact that your lying is wrong does not, as such, provide you with a reason to act; it does so only because you also care about such things as doing what is right or the wellbeing of those who might be adversely affected by your lying, such as your spouse. If you were to lack desires such as these, however, you would also lack a reason not to lie. To which Humeans typically add: most of us do in fact have such desires and, hence, such reasons. There is a reliable contingent connection between an act’s being wrong and our having a reason not to perform it. Philosophers who identify themselves as both moral realists and naturalists tend to embrace the Humean theory of reasons.1 These philosophers offer arguments of various sorts for their position. Philippa Foot, for example, argues that accepting Humeanism saves realists from believing the Kantian “illusion” that all failures of morality are also failures of rationality.2 Mark Schroeder argues, by contrast, that since the Humean theory provides the most streamlined account of the conditions under which we have reasons to act, it should be accepted.3 Peter Railton contends, somewhat differently, that “for all its faults,” the Humean theory “is the clearest notion we have of what it is for an agent to have reasons to I am grateful to Tyler Doggett, Don Loeb, and the editors of this volume for their comments. 1 See, for example, Boyd (1988); Brink (1989 and 1992); Railton (1986 and 1992); and Schroeder (2007). Copp (1995) also expresses openness to the view. 2 Foot (2002: 167). Foot subsequently recanted her earlier commitment to the Humean view. 3 Schroeder (2007). Schroeder’s version of the Humean theory, I should add, differs considerably from the versions of Humeanism defended by philosophers such as Railton and Brink.
110
Moral naturalism and categorical reasons
111
act.”4 None of these philosophers, it should be noted, argues that there is something specific about a naturalist approach to philosophy that recommends Humeanism. Still, they all believe that their position is better off because it embraces the Humean theory. A commitment to the Humean theory, say these philosophers, is a virtue of their view. It is a striking fact that others who sympathize with a naturalist approach to philosophy disagree. In The Myth of Morality and The Evolution of Morality, for example, Richard Joyce contends that, far from being a virtue, a commitment to the Humean theory is a serious liability.5 The problem with Humean moral naturalism – or “moral naturalism” for short – according to Joyce, is that anything worth calling a moral system incorporates a commitment to there being reasons that apply to us no matter what desires we happen to have. In so far as moral naturalists fail to accommodate the existence of these so-called categorical reasons, they fail to offer us a description of anything worth calling a moral system. I have some sympathy with this complaint. Despite this sympathy, I propose in this essay to take up the cudgels in favor of moral naturalism. In particular, I wish to defend moral naturalism from Joyce’s contention that any philosopher with broadly naturalist inclinations should reject moral naturalism in favor of the error theory, according to which there are no moral facts at all. Joyce’s argument merits careful consideration. In my judgment, however, moral naturalists have at their disposal a reply to Joyce’s argument that is both adequate and economical. According to this reply, arguments such as Joyce’s suffer from an arbitrariness problem in so far as they arbitrarily weight certain features of ordinary moral practices while discounting others. Once we get clear on the desiderata that a good moral theory should satisfy, I further claim, we can see that not only do arguments such as Joyce’s fail to provide sufficient reasons for rejecting moral naturalism, but also that moral naturalism is preferable to the error theoretic position that Joyce claims that naturalists should accept. My aim in this essay, then, is to defend moral naturalism. This might seem a curious thing to do, since I have elsewhere defended a version of “robust” moral realism that rejects the Humean theory.6 But my wish to defend moral naturalism on this occasion is the result of neither a change of heart nor philosophical perversity. Rather, it is rooted in the conviction that it is instructive to think about what metaethical view one would accept were one to become convinced that one’s favored view is false. My 4 Railton (1986: 166). 5 See Joyce (2001 and 2006). 6 In Cuneo (2007: ch. 7).
112
t e r e nc e c u n e o
own favored view, as I have indicated, is a version of robust realism that accepts the existence of categorical moral reasons. Were I to become convinced that robust realism is false, however, I would accept a version of moral naturalism. Even philosophers need contingency plans! 6.1 T h e c a s e f or c at e g or ic a l r e a s ons Moral naturalists, we have seen, accept the claim that: The moral wrongness of an act does not, as such, provide an agent with a reason not to do it.
At first glance, this claim might take us aback. How could the wrongness of an act fail to provide you with a reason not to do it? Moral naturalists have an answer to this question. The answer consists in distinguishing the claim that: (A) Morality is categorically applicable: a person can be morally obligated to do something even if doing it fails to serve any of her desires
from: (B) Morality is categorically reason-giving: a person has a reason to perform her moral duty even if doing so fails to serve any of her desires.7
Moral naturalists claim that while (A) is true, (B) is false. To help us see how this could be so, moral naturalists such as Foot draw parallels between morality and social practices such as etiquette.8 It is plausible to believe, Foot contends, that: (A′) Etiquette is categorically applicable: a person can be obligated by the practice of etiquette to do something even if doing it fails to serve any of her desires.
But it is not at all plausible to believe: (B′) Etiquette is categorically reason-giving: a person has a reason to conform to the requirements of etiquette even if doing so fails to serve any of her desires.
The reason why we have little reason to believe (B′) is that there appear to be cases in which the requirements of etiquette apply to an agent, but she has no reason to conform to them. For example, etiquette might require that you eat at a relaxed pace, not shoveling large quantities of food into your mouth. If you are eating dinner alone in your study, however, you 7 I borrow the basis for this formulation from Shafer-Landau (2005). 8 Foot (2002). Not all moral naturalists are enamored with the appeal to etiquette. For reservations, see Railton (1992).
Moral naturalism and categorical reasons
113
might have no reason whatsoever to pay attention to these requirements. You might even enjoy, as Joyce points out, being boorish.9 Moral naturalists believe that morality is similar. Moral requirements, the moral naturalists maintain, apply to all of us simply in virtue of the fact that we are normal, mature human beings. We cannot escape these requirements simply because we lack desires of certain kinds. But these requirements do not, as such, generate reasons for us to act. If we lack desires of the right sort, the obligations apply but do not count in favor of us conforming to them. In the two books mentioned earlier, Joyce acknowledges the distinction between the categorical applicability and the reason-giving force of morality. However, he contends, moral naturalists are mistaken to think that (A) is true but (B) is not. For nothing, Joyce claims, could count as a moral system if it failed to incorporate a commitment to both (A) and (B). In this regard, Joyce says, morality is deeply disanalogous to social practices such as etiquette. Why is that? In his earlier book, The Myth of Morality, Joyce argues that the clue lies with our practices of holding others accountable. “Just consider,” Joyce writes, “our moral condemnation of the Nazis.” We do not let wrongdoers such as the Nazis off the moral hook – acknowledging that they had no reason to treat others with minimal respect – if we discover that they don’t care about morality or the welfare of those adversely affected by their actions. To the contrary: The manner in which we condemn Nazis, ignoring any unusual desires or interests that they may have, is not a peripheral element of moral discourse; it presents a kind of reprehension that is central. A system of values in which there was no place for condemning Nazi actions simply would not count as a moral system.10
When spelled out more fully, Joyce’s thought is this: our practice of holding wrongdoers morally accountable appears to presuppose that wrongdoers have moral reasons not to act as they did. If there were such reasons, however, then they would have to be categorical; otherwise wrongdoers could get off the hook by sincerely denying that they care about morality or the welfare of those they harmed. But the lack of such desires is precisely the type of consideration that those of us who engage in practices of holding others accountable find morally irrelevant. To the extent, then, that our practices of holding others morally accountable actually reveal our moral commitments, we have strong reason to believe that we are committed to Joyce (2006: 202). 9
10
Joyce (2001: 43).
114
t e r e nc e c u n e o
the existence of categorical moral reasons (and that these reasons carry considerable weight). Indeed, our commitment to there being such reasons is so deeply entrenched in our moral practices, Joyce adds, that any “system of values that leaves out categorical imperatives” would lack the authority that we expect of morality, and simply “not count as a ‘morality’ at all.”11 This is a contentious claim. We are, after all, familiar with cases in which we discover that certain claims that are deeply entrenched in a given practice are false. In some cases, we give up on the practice; the claims that we have discovered to be false, we determine, are so central to it that there is not enough left of the practice to salvage. Witness the fate of various practices of soothsaying, such as those that involve consulting the entrails of animals, for example. In other cases, however, upon discovering that various claims that are entrenched in a practice are false, we revise the practice, noting that while we must fundamentally alter our understanding of the concepts it employs, the alterations are worth it. When we became convinced that relativity theory is true, for example, we did not jettison our practice of ascribing simultaneity relations between events; we simply altered our understanding of what we are doing when making these ascriptions. Joyce is of the view that our commitment to categorical reasons is so deeply entrenched in ordinary moral discourse and practice that, were we to surrender it, whatever is left would simply not count as morality. But one might wonder why we should believe this. Why not concede that a commitment to categorical reasons appears to be central to our moral lives but also contend that, were we to discover there are none, morality would continue to go on much as it did, since we could – and perhaps must – live with the alteration? Why not believe that our moral practices are fluid in this regard, like our practices of ascribing simultaneity relations between events? In his later book, The Evolution of Morality, Joyce tackles this issue at greater length.12 In this discussion, Joyce acknowledges that, in the abstract, we probably have no way to decide whether a commitment to categorical reasons is a non-negotiable aspect of moral thought and practice. Philosophers simply have no settled procedures for determining such 11 Joyce (2001: 177). 12 Joyce (2001: 97, 157) broaches the issue but not at any length and with far less diffidence than in The Evolution of Morality. In The Evolution of Morality, I should add, Joyce voices reservations about using the conceptuality of categorical reasons, preferring to talk of “moral clout” and “practical oomph.” For the sake of expediency, I will formulate his view in terms of categorical reasons; nothing of importance, I believe, will be distorted.
Moral naturalism and categorical reasons
115
matters. Even so, Joyce contends, there is a way to make progress on the issue before us. The first step is to identify the central roles played by our moral concepts, such as our concepts of the putative moral rights, responsibilities, and obligations that comprise our moral practices. The next step is to ask whether these concepts could play these roles if moral naturalism were true. If they could not, Joyce continues, then we have grounds for doubting that a given system of putative rights, responsibilities, and obligations counts as a moral system at all.13 And what exactly are the roles that moral concepts would have to play in order to be worthy of the name? The role to which Joyce devotes most attention is what we might call their silencing function.14 An example might best illustrate what Joyce has in mind. Suppose, once more, that I am faced with the choice whether to lie to my spouse. I know that doing so would be wrong. But I also know that lying would save me from considerable embarrassment. If I genuinely understand, however, that lying on this occasion is wrong, then this should have the effect of putting out of mind rationalizations that might tempt me to lie. It should shrink the space of deliberative possibilities so that I do not engage in any sort of cost–benefit calculation. Of course I may go ahead and lie anyway. But the price will be an awareness that I deserve to be the object of disapprobation. Joyce’s thought, then, appears to be that a constellation of putative rights, responsibilities, and obligations counts as a moral system only if our concepts of them would play a silencing function of this sort. The question to put to moral naturalism is whether it can account for the silencing function of moral concepts. Joyce thinks it is clear that moral naturalism cannot. For consider the following two cases.15 Imagine, first, that I have become totally morally disaffected. I no longer care about complying with moral demands. Imagine, furthermore, that my relationship with my spouse has so deteriorated that I am no longer concerned about her wellbeing. I am now faced with the choice to lie to her, knowing that if I do, this will remain undetected. If I were aware that moral naturalism were true, then morality would not play a silencing function for me. To the contrary, in the case described, I would plainly see that I have no reason whatsoever to accede to morality’s demands. Imagine, now, a somewhat different case in which I am not totally morally disaffected. My commitment to morality and the wellbeing of my 13 See Joyce (2006: 201–202). 14 See Joyce (2006: 4.2). 15 See Joyce (2006: 6.4). I have modified Joyce’s cases for my own purposes.
116
t e r e nc e c u n e o
spouse is, however, weak. I am once again faced with the prospect of lying to her to save face. If I recognize that moral naturalism is true, I could either strengthen my resolve to do the right thing or surrender the desires that provide reasons for me to act morally. Either way, moral considerations do not play a silencing function for me. To the contrary, my recognition of the moral facts is simply an invitation to engage in exactly the sort of cost– benefit analysis that ordinary morality says should be ruled out. From cases such as these, Joyce concludes that moral naturalism cannot accommodate the silencing function of moral concepts. Since he also takes the silencing function of morality to be non-negotiable, Joyce maintains that we have strong reason to believe that moral naturalism has failed to offer us an account of anything that deserves to be called a moral system. The system moral naturalism describes, says Joyce, is simply “too wimpy to be mistaken for morality.”16 I am dubious about this. Later I will explain why. For present purposes, however, I propose that we not probe the line of argument we just considered. Instead, I suggest that we take a maximally concessive approach toward the sort of case Joyce wishes to develop. We can do this by making the following four concessions. First, let’s concede for argument’s sake that we have good reason to believe that at the heart of ordinary moral thinking lies a commitment to the reality of categorical moral reasons. Accordingly, let us assume, for present purposes, that we have decent reasons to believe that: Necessarily, if there are moral facts, then there are categorical moral reasons.
Second, let us concede, without actually reviewing the case offered in its favor, a claim about which Joyce and many moral naturalists agree, namely: There are no categorical reasons of any sort.17
Third, let us agree that: While there are no categorical reasons, there are non-moral Humean reasons, such as prudential ones. These reasons apply to agents in virtue of the fact that they have desires of the appropriate sorts.
In his discussion of moral naturalism, Joyce admits that this view is “under serious pressure” from its critics.18 Still, since both moral Joyce (2006: 208). Joyce (2001: chs. 4 and 5) presents an argument for this claim. 18 Joyce (2001: 52). 16
17
Moral naturalism and categorical reasons
117
naturalists and Joyce accept this claim, it will do no harm to grant it for present purposes. Finally, Joyce argues that if one is a naturalist, then one’s metaethical options are limited. According to Joyce, if one is a naturalist, then one should accept one of two views. Either one should accept moral naturalism or the error theory, according to which: There are no moral facts (in the realist’s sense or in any other).
And: Ordinary moral thought and discourse purport to represent moral facts but fail to do so, as there are none. In this respect, they are deeply and systematically mistaken.
At various places in The Myth of Morality, Joyce explains why he believes error theory to be more plausible than other forms of moral antirealism, such as expressivism.19 Let us suppose that these explanations are convincing. We can then grant, for the purposes of argument, this fourth claim: If there are no moral facts, then the moral error theory is true.
This last claim represents yet another claim that is common ground between Joyce and the moral naturalists. For both agree that moral thought and discourse purport to represent moral reality. As such, both agree that if there were no moral facts, then ordinary moral thought and discourse would be in deep and systematic error. It will be helpful to have a name for the position that incorporates these four claims. Let us call such a position the mixed view. Advocates of the mixed view hold that there are normative reasons aplenty, albeit no moral ones, for such reasons would have to be categorical, and these do not exist. At any rate, with the rudiments of the mixed view before us, we are now in a position to present the central argument that Joyce offers against moral naturalism and how moral naturalists might respond to it. 6.2 T h e c a s e ag a i ns t mor a l n at u r a l i s m a n d t h e mor a l n at u r a l i s t s’ r e s p ons e Moral naturalism of the sort in which I am interested comprises two claims. The first is that there are moral facts. As such, moral naturalism 19 See Joyce (2001: ch. 1). I am working with the account of the error theory that Joyce offers in The Myth of Morality.
118
t e r e nc e c u n e o
implies that there are facts such as that lying simply to save face is wrong and that Hitler was wicked. (Naturalists add that these facts are in some interesting sense natural facts. What they have in mind by this is difficult to pinpoint. But the matter need not concern us here.) The second claim is that the Humean theory of reasons is true. According to the Humean theory, recall, nothing is a reason for an agent unless and because she has desires of the appropriate types. Against moral naturalism of this variety, Joyce offers: The Categoricity Argument (1) Necessarily, if there are moral facts, then there are categorical reasons. (2) There are no categorical reasons. (3) So, there are no moral facts. (4) If there are no moral facts, then moral naturalism is false. (5) So, moral naturalism is false.20 We have conceded that we have decent reasons to accept premise (1). And we have granted that both Joyce and the moral naturalists accept (2) and (4), the argument’s remaining two premises. Obviously, however, moral naturalists cannot accept the argument’s conclusion. What, then, should a moral naturalist say in response? As a first step, let us call to mind a proposition that I have singled out for attention several times, namely: It is wrong to lie to one’s spouse simply to save face.
To this proposition, we could add others like it, such as: It is wrong to break a promise simply because one feels like it.
And: It is wrong to kill another human being simply because it gives one pleasure.
Call propositions such as these the Stock Moral Truisms. These propositions are truisms not simply because anything worth calling a moral system would have to include them. They deserve the name because they are also widely accepted and appear to be obviously true. For imagine someone were to reject the claim that it is wrong to kill another human being
See Joyce (2001: ch. 2) and (2006: ch. 6).
20
Moral naturalism and categorical reasons
119
simply because it gives one pleasure. (Such a person need not be committed to the claim that the killing is morally permissible. He might instead say that such killings, like all our activities, are simply not the sort of thing that can have moral properties.) This would be a strong prima facie reason to believe that this person lacked the concept of wrongness, suffered from some sort of serious cognitive malformation, received a deeply inadequate moral education, was not sane, was kidding, was attempting to deceive us, was deeply confused, or the like. Admittedly, such a denial would constitute only a prima facie reason to believe one or more of these things. It might be that, although this person is sincere and perfectly competent with the concepts in question, he believes that our concept of wrongness is subtly incoherent. His denial, I imagine, would still strike many of us as strange. But it would be comprehensible. And, conceivably, it might even be compelling. In that case, the Stock Moral Truisms might still appear true, although we would find ourselves with reasons not to accept them. With this in mind, let’s now return to the Categoricity Argument, noting the following feature: defenders of the argument take the appearances of ordinary moral thought and practice very seriously. The reason why we should accept the argument’s first premise, says Joyce, is because it appears to make the best sense of how we conduct ourselves in the moral domain, such as when we hold wrongdoers morally accountable. Indeed, we’ve seen that Joyce believes that this premise is constitutive of our moral practices in the sense that anything worth calling a moral system must presuppose it. Yet, it must be said, the defenders of the Categoricity Argument, such as Joyce, do not take the appearances of other dimensions of ordinary moral thought and practice with the same degree of seriousness. After all, they reject the Stock Moral Truisms. They reject the Stock Moral Truisms in spite of the fact that ordinary moral thought and practice also seem to commit us to them. Indeed, according to some, these truisms are constitutive of our moral practices in the very same sense that premise (1) of the Categoricity Argument is supposed to be: they are nonnegotiable features of our ordinary moral practices. A system that failed to incorporate these apparent truisms would simply not count as a moral system at all. Herein lies the tension I want to exploit. For suppose we agree, for argument’s sake, that there are no categorical reasons. Why, at the outset of theorizing, would we take this to be a reason for rejecting the Stock Moral Truisms rather than the first premise of the Categoricity Argument? After all, both the first premise of this argument and the Stock Moral Truisms
120
t e r e nc e c u n e o
appear to be truths deeply embedded in ordinary moral thought and practice. Indeed, I have just noted that they appear to be non-negotiable features of anything that could count as moral thought and practice. Accordingly, it is difficult to see why a person should argue that: A : It is conceptually necessary that, if there are moral facts, then there are categorical reasons. But there are no categorical reasons. So, there are no moral facts. However, if there are no moral facts, then an agent cannot exhibit a moral demerit such as having acted wrongly. Suppose, now, that in ordinary conditions, an agent in fact fails to intend to honor her promise or kills another human simply because she feels like doing these things. It follows that the Stock Moral Truisms are false: for in ordinary conditions, an agent fails to intend to honor her promise or kills another human being simply because she feels like it yet she does not thereby exhibit the moral demerit of having acted wrongly.
Rather than contend instead that: B: It is conceptually necessary that, in ordinary conditions, if an agent fails to intend to honor her promise or kills another human being simply because she feels like doing these things, then she thereby exhibits the moral demerit of having acted wrongly. Suppose that, in ordinary conditions, an agent in fact fails to intend to honor her promise or kills another human simply because she feels like doing these things. It follows that she exhibits the moral demerit of having acted wrongly. But if she exhibits the moral demerit of having acted wrongly, then moral facts exist. And since she does exhibit such a demerit, moral facts exist. There are, however, no categorical reasons. So, the first premise of the Categoricity Argument is false: there are moral facts, but there are no categorical reasons.
It was G. E. Moore who used a strategy similar to this in his reply to the skeptic about the external world.21 In his reply to the skeptic, Moore pointed out that we have no better reason to accept the skeptic’s principles rather than the claims about the external world that the skeptic rejects; the former recommend themselves no more than the latter. If so, Moore claimed, the skeptic’s case can be neutralized. We would have no more reason to reject the reasons we have to believe in the existence of the external world rather than the skeptic’s principles. Although moral naturalists have held much of what Moore had to say at arm’s length, I suggest that in this context they call upon Moore as an ally. For the Moorean strategy helps us to see that the Categoricity Argument suffers from an arbitrariness problem. Advocates of the argument help themselves to 21 See, for example, the first two chapters of Moore 1953. Rowe (1993: 86) calls this strategy the “G. E. Moore shift.”
Moral naturalism and categorical reasons
121
certain deeply entrenched components of our moral thought and practice while jettisoning others. Or so it would seem. For it might be tempting to brush off the Moorean-style response as doing little more than highlighting that one man’s ponens is another’s tollens. But I think we should not let the relative familiarity of cases in which philosophers use similar premises to argue to incompatible conclusions to obscure the following fact: given the dialectical situation, proponents of the mixed view should be able to articulate principled reasons – ones that don’t merely presuppose the truth of the position that they’re trying to support – for why they embrace argument A rather than argument B. My claim is that the reasons aren’t apparent. To make this point, let me note that ethical theories, as Joyce himself argues, are themselves assessable along two main dimensions. In the first place, one can ask of a given theory whether it commits us to types of entities that we would rather not accept if we could get away with it because they would be highly mysterious or clutter our ontology. In the second place, one can ask of such a theory whether it does an adequate job of preserving deeply embedded features of ordinary moral thought and practice.22 Ideally, a good morally theory would want to satisfy both these desiderata, although as Joyce notes, they often tend to pull against one another. Let us consider the first criterion of assessment. Does accepting the second argument just offered – what I’ve termed argument B – violate this criterion? It seems not. Were they to accept the existence of Humean moral reasons, advocates of the mixed view would not have thereby admitted the existence of a type of entity that they’d rather not accept if they could get away with it because it is highly mysterious. Nor, for that matter, would they have introduced new and unwanted complications into their view. For if moral naturalism were correct, moral reasons would simply be a species of Humean reasons. We have already seen, however, that advocates of the mixed view do not blanch in the face of Humean reasons; they have no reservations about admitting the existence of reasons of this type. If that is right, however, accepting the second argument introduces no type of entity that is new and objectionable into the picture. Turn now to the second criterion of assessment, which concerns whether a theory does an adequate job of preserving deeply entrenched features of ordinary moral thought and practice. Does accepting argument B violate
22
Joyce (2007b: 5–6). Timmons (1999: ch. 1) makes similar claims.
122
t e r e nc e c u n e o
this criterion? The issue in this case is more delicate. If we concede Joyce’s contention that our moral practices appear to commit us to the existence of categorical moral reasons, then moral naturalists will have to grant that ordinary moral thought and practice is in error. But it is important to describe the error accurately. In various places, Joyce writes as if the moral naturalists’ claim that there are no categorical moral reasons implies that moral thought cannot play a silencing function. For example, toward the end of The Evolution of Morality, Joyce writes: So the question we need to ask is whether moral discourse could carry on playing whatever role it does play if the connection between its prescriptions and the reasons people have to comply were merely a reliable contingent one. If not, then we have grounds for doubting that such a framework counts as a “moral” system at all.23
Joyce continues, arguing that if moral naturalism were true, moral thinking could not play its fundamental roles: Moral thinking has a function … and deliberations in terms simply of what we want and need will not suffice. The moralization of our practical lives contributes to the satisfaction of our long-term interests and makes for more effective collective negotiation by supplying license for punishment, justification for likes and dislikes, and bonding individuals in a shared framework of decisionmaking. It is, I submit, precisely the purported authority and inescapability of moral prescriptions that enable them to perform these functions. Thus, a value system lacking practical clout could not so effectively play the social roles to which we put morality, and thus we could not use it as we use morality.24
In passages such as these, Joyce seems to move from the moral naturalists’ claim that: All moral reasons are Humean
to the further claim that: If all moral reasons were Humean, then moral thought could not perform a silencing function.
But such an inference would be erroneous. Moral thought might tend to play a silencing function even if all moral reasons were Humean. For it might be that moral thought tends to misdescribe the nature of moral reasons, taking them to be categorical when they are really Humean. 23 Joyce (2006: 202).
24
Joyce (2006: 208).
Moral naturalism and categorical reasons
123
Let me explain more fully what I have in mind. Moral naturalists believe that there are moral facts such as those reported by the Stock Moral Truisms. They also believe that there are moral reasons, although they hold that they are all Humean. Although moral naturalists believe that moral reasons are Humean, they also believe that many – and perhaps most – of us have moral reasons to conform to the requirements of morality. In so far as they accept that Joyce has accurately described ordinary moral thought and practice, however, they concede that moral thought and practice tend to incorporate a mistake. The mistake consists in taking the moral reasons that many of us have to be categorical when they are really Humean. It is in this respect that moral naturalists (at least if they agree with Joyce’s description of ordinary moral thought and practice) believe that ordinary moral thought and practice tend to incorporate an error. To soften the blow of this consequence of their views, moral naturalists might have various things to say about why it might be useful for us to misdescribe the nature of moral reasons.25 Consider, for example, Mark Johnston’s recent description of the correct way to learn pocket billiards. According to Johnston, the correct method involves seeing a white “ghost ball” touching your object ball at the point farthest from the intended pocket. One then shoots the cue ball directly at the ghost ball, and the object ball rolls in the pocket, if all goes well. This is the correct way to learn billiards, says Johnston, even though there is no ghost ball, and even though seeing it actually gives the wrong aiming point on the object ball.26 In a similar fashion, moral naturalists might claim that thinking of some moral reasons as being categorical is not only something we find very natural, but also the best way to acquire moral reasons. Perhaps, for example, thinking of moral reasons in categorical terms helps us to care about morality and the welfare of others, thereby “setting the moral hook” in us. In this way, it is a useful device for forming and maintaining the sorts of desires that give us genuine reasons to act morally. Granted, there are important differences between the two cases just described: in the billiards case, we intentionally employ a method that we know involves false presuppositions; 25 One thing that a moral naturalist might emphasize is that their views are typically committed to so-called externalist accounts of reference, according to which it is possible to refer to something even when we have mistaken beliefs about it. What matters, according to these views, is not that we have correct beliefs about that to which refer but that our concepts have the right sort of explanatory connection to them. See Boyd (1988). 26 Johnston (2010: 17).
124
t e r e nc e c u n e o
in the moral case, by contrast, most of us presumably do not. We simply find it natural (perhaps because of our moral education) to think of moral reasons as being categorical. In any event, the important point to see is that by admitting that moral reasons are Humean, moral naturalists needn’t also claim that moral thought and discourse fail to play the types of function that Joyce believes count as components of a moral system. Earlier I proposed that we engage Joyce’s argument against moral naturalism in a concessive spirit, granting that we have decent reason to believe that moral thought and practice appear committed to the existence of categorical reasons. Proceeding in this fashion, we now see, comes at a price. For if Joyce is right, then moral naturalism does not unequivocally satisfy the second dimensions along which we assess moral theories. Moral naturalists must admit that moral thought tends to traffic in error of a certain kind, albeit of a fairly local sort. Let us now look at the alternative available to naturalists, which Joyce says is the error theory of morality. Does the error theory satisfy the two criteria with which we assess moral theories? Well, it fares no worse than moral naturalism along the first dimension. It does not introduce any new and problematic entities. Things are different, however, when it comes to the second dimension. In this case, it fares considerably worse than moral naturalism. For unlike moral naturalism, the error theory does not countenance a fairly localized error about the way we tend to think of the nature of moral reasons. Rather, it claims that there simply are no moral facts. The Stock Moral Truisms fail to express truths. In addition, it says that there are no moral reasons of any sort. Since, however, moral thought and discourse purport to represent such facts and reasons, they are, according to the error theory, in massive and systematic error. The error theory earns its name. If it were true, it is difficult to see how we could be more mistaken about moral matters. Admittedly, most philosophical views come at a cost. The task of the theorist is usually to determine which costs are, on the whole, less onerous than others. But the case at hand, it seems to me, is fairly straightforward: the theoretical cost of accepting moral naturalism is considerably less than embracing the error theory. If a good moral theory saves wellentrenched commitments of ordinary moral thought and practice, it is better to accept a view according to which there are moral reasons that we tend to misdescribe rather than one according to which there are no moral facts or reasons at all.
Moral naturalism and categorical reasons
125
In summary: suppose we assume that moral thought is committed to things having moral features of various kinds. If so, advocates of the mixed view such as Joyce who propound the Categoricity Argument are subject to a charge of arbitrariness, as they arbitrarily deploy some deeply embedded commitments of ordinary moral thought and practice while ignoring others. I’ve intimated that this accusation can, in principle, be adequately discharged. But, I’ve argued, in the case at hand it is not. The type of reason to which one might appeal to discharge the accusation – that is, that accepting argument A is preferable along the two dimensions of theory construction – is not apparent. In fact, the available reasons appear to point in the opposite direction, according to which we favor the acceptance of argument B rather than argument A. On the assumption that the charge stands, I conclude that we ought not to reject moral naturalism on the strength of anything like the Categoricity Argument. At this point, I can imagine advocates of the mixed view becoming impatient. They will protest as follows: suppose we agree that moral naturalism and error theory are the naturalists’ only two viable metaethical options. And suppose it is true that moral naturalism is in certain important respects less problematic than error theory since the nature of the error it countenances is considerably less severe than that admitted by the error theory. This, we can admit, is a point in favor of moral naturalism. Still, it does not follow that we should accept moral naturalism according to the two criteria of theory selection that we have employed. For in as much as moral naturalists reject the existence of categorical moral reasons, they fail to offer an account of a moral system at all. But the criteria for theory selection which we have employed is for assessing moral or metaethical theories – theories that describe moral systems. If so, moral naturalism may be better than error theory in some respects, but it is not a better metaethical theory, for it does not describe a moral system at all. The cogency of this objection hangs on what exactly Joyce has in mind by a moral system. In one of the more memorable passages in The Myth of Morality, Joyce says this: It is not necessary for me to claim that absolutely every piece of recognizably moral language implies the validity of categorical imperatives, only that a sufficient portion of them do, such that if we were to eliminate categorical imperatives and all that imply them from the discourse, whatever remained would no longer be recognizable as – could not play the role of – a moral discourse. Any system of values that leaves out categorical imperatives will lack the authority that we expect of morality, and any set of prescriptions failing to underwrite
126
t e r e nc e c u n e o
this authority simply does not count as a “morality” at all. Moral discourse, in other words, is a house of cards, and the card at center bottom has “categorical imperative” written on it.27
This passage is interesting because Joyce tends to slide between talking about what is necessary for a discourse or system of concepts to count as moral, on the one hand, and what it is for a system of values or set of prescriptions to count as moral, on the other. The same ambiguity, I believe, is present in much of what Joyce says in The Evolution of Morality. Let us suppose that when Joyce speaks of a system of morality being a system of values and prescriptions he means that it is a system of putative rights, responsibilities, and obligations. If so, then Joyce’s claim is that if a system of putative rights, responsibilities, and obligations were not to include categorical reasons, then it could not count as a moral system. In principle, this might be correct. But we need an argument to believe it. Joyce’s argument, recall, is that moral discourse could not play a silencing function if the system it purports to describe were not to include categorical moral reasons. I have argued that this is not so. For all we reasonably believe, our moral thinking could carry on pretty much as it always has even if there were no categorical reasons. Of course ordinary thought about this system would be erroneous in certain respects. But it might be both difficult and undesirable to avoid the error. Suppose, by contrast, that Joyce means by a system of morality a system of moral concepts. If so, Joyce’s contention is that a system of concepts could not count as moral if it did not include the concept of a categorical moral reason, which we assume has genuine application. Now, however, we need to distinguish. On the one hand, the system of concepts in question might be that employed by most ordinary people in their everyday moral deliberation. On the other, it might be that described by moral naturalists when theorizing about morality. If it is the former system, then we should not reject moral naturalism on the grounds that it offers a distorted picture of our folk moral concepts. The moral naturalists are not offering an account of such a system at all. And, so, it cannot be said that, if their view were true, our ordinary folk concepts would cease to play a silencing role. If it is the latter system of concepts that is in question, then there might be a point to press against the moral naturalists. For it might be said that moral naturalists have failed to capture enough of ordinary moral thought since, were moral naturalism accepted, then moral thought would fail to play a silencing role.
Joyce (2001: 176–77).
27
Moral naturalism and categorical reasons
127
Let us focus for a moment on this last counterfactual claim: were moral naturalism accepted, then moral thought would fail to play a silencing role. Is its truth enough to vindicate the accusation that moral naturalists have failed to provide an account of a moral system at all? I doubt it. I have no full-blown argument to establish this. But the following two considerations, I believe, should give us pause. First, consider a case analogous to the one Joyce offers. Many have thought – Plato, Hobbes, and Kant come to mind – that moral obligations must be tied to a system of rewards and punishments, typically meted out by a divine being. The idea, presumably, is that something could count as a moral system only if conformance to the rights, responsibilities, and obligations that comprise it were ultimately in one’s best interest and lack of conformance were not. (This, incidentally, is not the same idea as silencing, which Joyce invokes. According to the system of rewards, moral considerations override all others. But they do so only because conforming to them is ultimately in one’s own self-interest. If this view were true, proper moral deliberation could involve cost–benefit calculations, which reveal that acting virtuously is in one’s ultimate interest. This, recall, is precisely the sort of reasoning that silencing is supposed to rule out.) It is a plausible conjecture that, by making these claims, these philosophers were not introducing novel elements to the ordinary moral thought of their time, but accurately describing important elements of it. If so, one could imagine someone arguing that any view that rejected the system of rewards position would fail to capture enough of ordinary moral thought since, were it accepted, then moral thought would fail to represent moral obligations as overriding in the sense that conforming to them ultimately redounds to one’s own wellbeing. In fact, however, many today reject the system of rewards position. But we would not say of such people that, when they describe their view, they were describing something other than a moral system. For our moral practices appear to be supple enough to allow for difference of opinion as to whether moral considerations are tied to a system of rewards. If that is right, though, then the question to ask of Joyce’s position is why we should think that morality could survive the rejection of the system of rewards view, but not the rejection of categorical moral reasons. As best I can tell, we don’t have a good answer to this question. That moral considerations must play a silencing function and that anything worth calling virtue must ultimately contribute to one’s own wellbeing are both convictions deeply entrenched in ordinary moral thought. It is difficult to see why morality could survive the rejection of one conviction but not the other. Let me emphasize that, by drawing attention to this parallel, I do not
128
t e r e nc e c u n e o
mean to suggest that just about anything could count as a moral system. There are limits to what could count as a moral system and, hence, the changes it could absorb, although it may be difficult to say in the abstract what these changes would be. Rather, the point is that we should distinguish between a description of a moral system M according to which M has various serious theoretical blemishes since it fails to capture deeply embedded features of ordinary morality and one that fails to describe a moral system at all. I am prepared to say that moral naturalism falls into the first category. But I see little reason to believe it falls into the latter, at least when the view is cast in the most charitable light. Second, recall that the proposal we are considering is that, when elaborating their view, moral naturalists have failed to describe a moral system since were their view accepted, then moral thought would fail to play a silencing role. But one could reasonably worry that if this last counterfactual were true, it would imply that too many putatively metaethical views would not count as metaethical views at all, since they would fail to describe a system of moral concepts. Consider expressivism, for example. Expressivists believe that moral thought does not even involve predicating moral properties of things, let alone thinking that there are requirements that yield categorical moral reasons. If expressivism were accepted by the folk, then they would not think of moral considerations as generating categorical reasons which silence cost–benefit calculations. It follows that, according to the line of argument Joyce has offered, expressivism should be dismissed on the ground that it fails to offer us an account of a system of moral concepts at all. Or consider moral fictionalism, a view that Joyce himself has defended.28 Fictionalists such as Joyce propose that we should reform ordinary moral thought and discourse in such a way that those who participate in them engage in a type of pretense wherein they pretend there are categorical moral reasons. But pretending to believe and genuinely believing there are categorical moral reasons, as Joyce emphasizes, are different. They are different stances one can take toward a proposition. Now imagine a scenario under which fictionalism were widely accepted. Under this scenario, most would not believe but merely pretend to believe there are categorical moral reasons. Would pretending to believe there are moral reasons play the silencing function that Joyce believes is a non-negotiable dimension of genuinely moral thought? 28 See Joyce (2001: chs. 7–8). Actually, Joyce claims that the fictionalist position he defends is a version of expressivism. If so, then the considerations adduced earlier regarding expressivism apply to it.
Moral naturalism and categorical reasons
129
It some cases it probably would. But, then again, in other cases, it probably wouldn’t. For it is important to remember that, in Joyce’s view, those who engage in the moral pretense know that they are doing so; they are not having the wool pulled over their eyes by an elite few who propagate something like Plato’s noble lie. So, return one final time to the case with which we began: you are tempted to lie to your spouse simply to save yourself from considerable embarrassment. No one will find out if you do. Nor will doing so alter your long-term commitment to the fiction of morality. In this scenario, it is difficult to see why the thought that morality requires you not to lie – you know, after all, that you are merely pretending that there are moral reasons – would rule out your engaging in a cost–benefit analysis. To the contrary, I imagine that a moral thought of this sort would, for the clear-eyed fictionalist who is minimally committed to the moral fiction, simply be an invitation to engage in exactly the sort of cost–benefit analysis that Joyce believes is ruled out by genuine moral thought and discourse. Why is that? Well, the offense in question would not in any sense be a failure of rationality. Moreover, the clear-eyed fictionalist knows exactly why he often engages in moral pretense: it is for broadly instrumental reasons. He does so because he wishes to bolster self-control and increase social cohesion.29 But, by hypothesis, lying on this occasion would threaten neither of these goods. Earlier we saw that, in Joyce’s view, parallel considerations were enough to rule out moral naturalism. Something could not be a genuinely moral thought if it invited an agent who is minimally committed to morality to engage in a cost–benefit analysis about whether to conform to putative moral demands. If this is right, however, then it would appear that moral fictionalism should also be dismissed on the ground that it fails to offer us an account of a system of moral concepts; the normative concepts that we deploy only in the fictive mood needn’t play a genuinely silencing function. Let me summarize. I have argued that there are three ways to understand the claim that, were moral naturalism true, it would not describe a moral system. Under the first interpretation, a moral system is a system of putative rights, responsibilities, and obligations. I have claimed that if these putative rights, responsibilities, and obligations were as moral naturalists say, this would give us no reason to believe that moral thought would fail to play a silencing function. Under the second interpretation, a moral system is a system of concepts, albeit it is that employed by ordinary
Compare the discussion of Joyce (2001: 7.1).
29
130
t e r e nc e c u n e o
people when engaging in ordinary moral thought and discourse. I have contended that, if moral naturalism were true, Joyce’s argument provides no reason to believe that these concepts would fail to play a silencing function. For moral naturalists do not claim that ordinary people think of moral reasons as being Humean. Under the third interpretation, a moral system is that system of concepts that ordinary people would use, were they to accept moral naturalism. When addressing this view, I have made two replies: first, we have reason to believe that our moral practices are supple. They exhibit enough flexibility to incorporate significant changes in moral thinking, even one in which we came to think of moral reasons as being Humean. Second, if it is true that moral thought must play a silencing function, then Joyce’s argument would imply that a rather wide range of metaethical theories fail to describe a system of moral concepts. This is because were these views accepted, moral thought would not play a silencing function. For a robust moral realist who believes in the reality of categorical moral reasons this last claim would be too good to be true. A single line of argument would be sufficient to dispose of moral naturalism, expressivism, moral fictionalism, moral subjectivism, and other views, since they all fail to describe a system of moral concepts. But – and this is the point I wish to press – it shouldn’t be that easy to dismiss these rival views. I suspect Joyce would agree. After all, Joyce seems to think that if a theory yields the result that we can at best pretend there are categorical moral reasons, then that is good enough. It does not on this account fail to count as a moral theory that describes a system of moral concepts. Should a similar courtesy be extended to moral naturalism it is difficult to see why one would withold the favor.
ch apter 7
Does analytical moral naturalism rest on a mistake? Susana Nuccetelli and Gary Seay
More than a century ago, G. E. Moore famously attempted to refute all versions of moral naturalism by offering an extended inference consisting of the open question argument followed by the charge that moral naturalism commits a “naturalistic fallacy.” Although there is consensus that this extended inference fails to undermine all varieties of moral naturalism, the open question argument (OQA) is often vindicated as an argument against analytical moral naturalism. By contrast, the charge that analytical naturalism commits the naturalistic fallacy usually finds no takers at all. In this paper we argue that analytical naturalism of the sort recently proposed by Frank Jackson (1998, 2003) and Michael Smith (2000) does after all rest on a mistake – though perhaps not the one Moore had in mind when he made the naturalistic fallacy charge. Analytical moral naturalism is roughly the doctrine that some moral predicates and sentences are semantically equivalent to predicates and sentences framed in non-moral terms. One attraction of analytical naturalism is that it promises to deliver a naturalistic account of the content of moral judgment that leaves no ground for objections inspired by the OQA – which argues, in brief, that no matter how much purely descriptive information is available about an action, it’s still an open question whether that action is right or its end good, or whether we ought to perform it. Analytical naturalism attempts to accomplish this by resorting to conceptual analysis for moral predicates and sentences. Given this doctrine, it is at least possible that there are some such a priori or conceptual equivalences. If so, it is at least possible that the moral reduces to the natural in an a priori or analytical way. But our version of the OQA challenges this claim by showing that the reductions envisaged by these analytical naturalists are open to doubt on a priori grounds. We further We wish to thank Brian McGuinness, Charles Landesman, and Lynn Tyrell for helpful suggestions that have enabled us to improve this chapter.
131
132
s us a n a n uc c e t e l l i a n d g a r y s e a y
contend that, in the dialectical context created by a properly construed OQA, a “digging in the heels” defense of such reductionist strategy would in the end beg the question against the Moorean. A live option, at least since Moore’s day, for moral realists who wish to reduce the ethical to the natural has been the doctrine that moral terms and judgments could be replaced, without significant loss, by purely descriptive terms and judgments. Moore, of course, traced the doctrine back to J. S. Mill, Jeremy Bentham, Herbert Spencer, Edvard Westermarck, and others.1 But for unambiguous representatives of what we shall hereafter call “analytical naturalism” Moore needn’t have looked farther than the work of contemporaries such as Ralph Barton Perry and F. C. Sharp. Perry (1970 [1926]: 138) famously argued that “x is valuable” is semantically equivalent to “interest is taken in x.” And Sharp (1928: 409–11) defined “good” as “desired upon reflection,” and “right” as “desired when looked at from an impersonal point of view.” In Moore’s day, Allan White rightly noted, the OQA against analytical naturalism amounted to neither a straw man nor flogging a dead horse. Even so, it might reasonably be asked, what do those dinosaurs have to do with analytical naturalism now? Perhaps the reductive programs of analytical naturalists have by now been discredited? Not at all – as can be seen from the work of such formidable representatives as Jackson and Smith. Jackson (1998, 2003) develops a version of that doctrine, which he calls “analytical descriptivism,” in connection with his defense of moral functionalism. And Smith (2000) outlines a parallel version of analytical naturalism in the course of suggesting what is for him the only naturalistic moral realist account of the content of moral belief that can safely dodge the bullet aimed by the OQA against naturalistic moral realism. To both of them, this argument fails, and for similar reasons, but we’ll put that aside for the time being to consider first what their accounts have in common. For both Smith and Jackson, the most plausible version of ethical naturalism is analytical and has a realist gloss. It holds the conjunction of two theses: 1 On Moore’s view, Spencer took “conduct that is more evolved” to be semantically equivalent to “conduct that has a higher ethical sanction” (Moore 1993b [1903]: section 29, 97–98), and Westermarck advocated a form of psychological naturalism according to which “right” means “arouses feelings of approbation” and “wrong” means “arouses feelings of disapprobation” (Moore 1922b: 332). Whether analytical naturalism can be found in Mill, as Moore claims, is of course controversial (see, e.g., Hall 1950–51; Warnock 1960: 28–40; and West 1997). But Bentham is another story. He did after all write that words such as “right,” “wrong,” and “ought” have meaning only when interpreted in terms of the principle of utility. Moreover, he defined “right action” as “an action that is conformable to the principle of utility” (1988 [1789]: ch. 1, section 10 ff.).
Does analytical moral naturalism rest on a mistake?
133
(1) Some moral properties are identical to natural properties. (2) Moral predicates and sentences could be replaced without significant loss by purely descriptive predicates and sentences. Although, given thesis (1), analytical naturalism is a kind of moral realism, the doctrine is clearly incompatible with the forms of moral realism advocated by non-naturalists and by the so-called Cornell realists. True, like Cornell realism, analytical naturalism is itself motivated by the success of science, considerations of ontological parsimony pointing to both the impossibility of properties and facts other than those that make up the world as conceived by science, and the conviction that the moral supervenes on the natural. But its thesis (2) commits analytical naturalism to holding that at least some moral predicates and sentences are content-equivalent to purely descriptive predicates and sentences – a claim that is at odds with the Cornell realists’ view that the relevant relation between the moral and the natural involves properties and facts only and is therefore solely metaphysical. On the other hand, each of the analytical naturalists’ theses entails the falsity of non-naturalism, according to which at least some moral predicates and sentences are not only not replaceable without significant loss by purely descriptive predicates and sentences, but express irreducible moral properties and facts. Of concern here is a non-naturalist objection to analytical naturalism’s thesis (2) standardly raised by arguments along the lines of Moore’s OQA. If, as analytical naturalists contend, at least some moral predicates and sentences are conceptually equivalent to purely descriptive predicates and sentences, the possibility of replacing expressions in the moral vocabulary with purely descriptive expressions would be warranted a priori – although exactly which descriptive predicates and sentences might turn out to be the correct naturalistic replacements may well amount to an empirical matter of fact. Smith (2000: 29 ff.) develops a doctrine along these lines in the course of arguing that non-analytical naturalistic moral realism cannot adequately respond to the OQA. He notes that by abandoning the project of conceptual analysis in favor of a view according to which the relevant statements of identity between moral and natural properties and facts are necessary but a posteriori, Cornell’s style of naturalistic moral realism becomes unable to account for the content of moral belief. On Smith’s view, there might well be a posteriori identities between moral and non-moral properties as held by Cornell realists – say between the properties of rightness and maximizing utility. But the best argument that we could run in support of such identities would show
134
s us a n a n uc c e t e l l i a n d g a r y s e a y
“that we in fact appealed to another truth, but this time one which is supposed to be known a priori, about the relation between rightness and certain natural properties” (Smith 2000: 29). In a similar vein, Jackson has it that there are a priori or conceptual entailments between moral and non-moral predicates and sentences, which follow a priori from the moral functionalist account of moral properties and facts. On his view, “[w]hat is a priori according to moral functionalism is not that rightness is such and such a descriptive property, but rather that A is right if and only if it has whatever property it is that plays the rightness role in mature folk morality, and it is an a posteriori matter what that property is” (1998: 151 ff.). The first step in this analytical naturalist account, then, is a priori: namely, determining by analysis and reflection on our conception of certain moral predicates and sentences the conditions constraining the use of moral expressions, which would in turn constrain the naturalistic expressions that might qualify as replacements. But much of the remaining task facing the analytical naturalist is empirical, since it involves accounting for the content of moral judgment, which ultimately depends on being able to formulate correct statements of identity between expressions in both the moral and the descriptive vocabularies (Jackson 1998: 150; Smith 2000: 31). Since determining such semantic equivalences is likely to require empirical investigation and reasoned argument, there is logical space for the following reply to Moore’s OQA: given that identifying the relevant semantic equivalences is not, for analytical naturalism, an obvious matter, why should we think that the question of whether a certain equivalence holds will “always and genuinely” be an open question? That is, in response to the OQA, the analytical naturalist maintains that the correct conceptual equivalences between moral and descriptive expressions need not be trivial or obvious, but could be quite complex, and in need of negotiation (Jackson 1998: 150)2 or reasoned argument (Smith 2000: 31). Thus, if this reply is found compelling, the analytical naturalist program would appear beyond the reach of objections inspired by the OQA, none of which could therefore succeed against this version of naturalistic moral realism. After all, Jackson seems to have a point in replying to the OQA by noticing that “it is true that a sufficiently rich descriptive story leads a 2 In rejecting the OQA, Jackson writes, “[W]hat exactly is supposed to be always and genuinely an open question? Any and every identification of rightness, say, with some descriptive property? But this claim could be no objection to moral functionalist styles of analytical descriptivism. The identifications of ethical properties with descriptive properties offered by moral functionalism are one and all a posteriori” (1998: 150). See also Jackson (1998: 145, n. 10).
Does analytical moral naturalism rest on a mistake?
135
priori to an act’s being right; but this will be a clear case of an unobvious a priori or conceptual entailment, precisely because of the complexity of the moral functionalist story” (1998: 151). Likewise, according to Smith’s response to the OQA, although it is an a posteriori matter which property plays a certain moral role, the constraints on the property that could play that role can be settled only a priori because they are to be determined either by stipulation in the act of reference-fixing itself, or by reflection on the meaning of the words (Smith 2000: 29). In many cases, coming up with the correct analysis of the relevant constraints on moral concepts would be open to “reasoned argument,” and therefore might appear to be an open question. But that is simply because the task at hand is not obvious or trivial. If so, any such conceptual analysis may be correct yet open to reasoned argument. But this is only owing to the fact that its correctness does not depend on its being either open to reasoned argument or obvious. In fact, a conceptual analysis of a moral concepts is likely to be open to reasoned argument about “what the complex set of constraints on the use of the word being analyzed is and whether or not this complex set is entailed by the proposed analysis” (Smith 2000: 31). Thus there appear to be good reasons, argued independently by both Jackson and Smith, to think that the OQA has no intuitive force against the doctrine that some a priori or conceptual equivalences obtain between the building blocks of moral and purely descriptive language. Given this doctrine, other things being equal, some sentences ascribing moral predicates to actions, things, or states of affairs are content-equivalent to some sentences ascribing purely descriptive predicates instead. Needless to say, if this is plausible, then given a common assumption about the parallel between linguistic and mental content, an analytical naturalist account of mental content along similar lines would also be plausible. We turn now, for the remainder of this paper, to the analytical naturalist claim that some conceptual equivalences hold between predicates (not sentences) in the moral and non-moral vocabularies. On Smith and Jackson’s account, since such equivalences are likely to be unobvious, the theory that countenances them is therefore unaffected by Moore’s OQA. But we think that a modified inference, the OQA*, in fact raises a priori doubts about the analytical naturalist’s claim that moral expressions such as “good,” “ought,” and “right” could be replaced without significant loss by predicates in a purely descriptive vocabulary. Our OQA* begins by supposing the truth of at least one such statement of equivalence.
136
s us a n a n uc c e t e l l i a n d g a r y s e a y
Let’s suppose that the moral predicate “right” expresses a concept that is equivalent to that expressed by the purely descriptive predicate “N,” where “N” stands for whatever naturalistic predicate will turn out to denote the natural property that plays the rightness role in the ordinary conception of rightness. As far as our OQA* is concerned, what that predicate and the naturalistic property it denotes are could be currently under negotiation. Furthermore, these questions could end up being settled in mature folk morality, as Jackson predicts, by the convergence of our ordinary conceptions on the relevant predicates and properties.3 Let’s assume that now N stands for a complex (possibly infinite) disjunction where the moral predicate “right” is satisfied by action A just in case A is either utility-maximizing, or what we desire to desire, or conducive to maximal preference-satisfaction, etc. The OQA* can now be deployed to raise a priori doubts about whether a moral predicate such as “right” is content-equivalent to a naturalistic predicate (of whatever sort). For, given this argument, the claim that “right” can be replaced without significant loss by a predicate or predicates falling within N seems open to doubt on a priori grounds. To support this, the OQA* first considers a certain naturalistic predicate (or predicates) that might be a candidate for replacing “right” without significant loss. It then contends that the sort of judgment involved in assessing that putative equivalence of predicates is based initially on a self-ascriptive comparative judgment of content. Judgments of that sort are plausibly regarded as a priori in the sense that their epistemic justification requires neither evidence nor inference from evidence. Clearly, if such are the grounds for doubting a proposed naturalistic replacement, that doubt would be a priori. Once OQA* makes it plausible that there are after all a priori grounds for doubting a purported naturalistic replacement for “right,” the argument then maintains that its steps could be iterated for other suitable candidates for naturalistic replacement of “right” – or of “good,” “ought,” and the like. If this is correct, then it is open to a priori doubt whether any moral predicate could be replaced without significant loss by a naturalistic predicate or predicates.
3 On the issue of how the alleged equivalences between moral and non-moral expressions are going to be settled, Jackson seems confident that this will be the outcome of a mature folk morality. Smith is similarly optimistic in holding that, at the end of day, there will be convergence in reasoned argument about the a priori constraints that govern the ordinary conception of some moral terms. On his view, determining the actual naturalistic property (or properties) that plays the role of rightness in that conception is also an a posteriori matter (2004a: 31 ff.).
Does analytical moral naturalism rest on a mistake?
137
Obviously, we should expect a great number of currently known and unknown candidates for naturalistic replacements of “right.” Thus, perforce, the OQA*, if compelling, would render its conclusion plausible at best. To run this argument, we need not make comparative judgments of content about every candidate for naturalistic replacement of a moral predicate: as a plausibility argument, simply making them for a representative number of such candidates would suffice. In addition, there is plainly no need to make comparative judgments of content about a putative replacement of “right” with, say, “being a cabbage” (Parfit 2011). But we need to consider those candidates that have some chance of being naturalistic replacements of a moral predicate – for example: Utilitarian The moral predicate “right” can be replaced without significant loss by the purely descriptive predicate, “utility-maximizing.”
An OQA* could now be deployed to raise a priori doubts about Utilitarian. For evaluating the alleged content-equivalence triggers the standard Moorean question, construed as: Is “maximizes utility” contentequivalent to “right?” To answer this requires that one first make an a priori comparative judgment of content for both predicates, which starts out with self-ascriptive comparative judgments of content of the form: are my tokens of “M” content-equivalent to my tokes of “N?” The intuitions elicited in response to the Moorean question are initially first person, since they require that one compare one’s conceptions of the content of the predicates involved. They are therefore epistemically privileged intuitions, for under normal circumstances and in the absence of evidence to the contrary, their epistemic warrant requires neither investigation of the environment nor inference from evidence – provided of course that one understands the moral and non-moral predicates involved, and has no reason to doubt that one is a competent user of them (or to think that one’s intuitions about their contents are atypical and therefore irrelevant to the folk conception of them). Similar Moorean questions could be deployed to generate parallel doubts about other putative content-equivalences. Adequate answers to such questions would, at least initially, require access only to the concepts in one’s own mind, together with a priori generalizations of the resulting intuitions about sameness and difference in their content. There is now dialectical space to claim that the doubts about putative contentequivalences that such questions generate are warranted a priori: as argued above, they rely on intuitions that, under normal circumstances
138
s us a n a n uc c e t e l l i a n d g a r y s e a y
and in the absence of contrary evidence, require no empirical investigation (though they are defeasible by empirical evidence).4 In addition, those intuitions seem generally true, provided there are no reasons to think that one is not competent with the concepts involved. Arguments along similar lines could be run to generate a priori doubts about other putative naturalistic equivalences of “right,” “good,” “ought,” and the like. Since Moorean questions are likely to elicit a priori doubts for each proposed naturalistic equivalence of moral predicates, the burden is now on the analytical naturalists to produce reasons strong enough to overcome such doubts. Construed in this way, the OQA is beyond the reach of the responses to it offered by Jackson and Smith, given that the argument does not depend on the relevant question’s being open when it is significant or not trivial (or, alternatively, closed when it is insignificant or trivial). A priori warranted claims need not be trivial: what characterizes them instead is that they can be settled without empirical investigation. The OQA* is therefore immune to Jackson and Smith’s line of reply to Moorean arguments that starts out by observing something that Moore seems to have missed: namely, that a conceptual analysis could be correct but neither obvious nor trivial. In the case of the conceptual equivalences between moral and purely descriptive predicates countenanced by analytical naturalists, the Moorean question might appear open while being in fact closed. Does this challenge our argument? We think not, since we can concede the possibility that our self-ascriptive comparative judgments of content could on occasion be mistaken. Given resource limitations such as time and concentration-span, we could mistake for an open question one that is in fact closed. But the OQA* is not offered as a refutation of analytical naturalism: it’s merely a plausibility argument against it, resting on well known reasons in the literature on knowledge of content to the effect that, under normal circumstances and in the absence of evidence to the contrary, self-ascriptive comparative judgments of linguistic and mental contents depend on no empirical investigation and are generally not mistaken. We submit that there appear to be a priori grounds to doubt that purely descriptive candidates could replace moral predicates such as “right,” “good,” and “ought” without significant loss.
4 On our view, comparative judgments of content are a priori warranted, in the sense of being non-evidential. They might also be generally true, provided the thinker is a competent user of the concepts involved. For similar claims in the case of self-knowledge, see Boghossian (1994); Burge (1996, 1998); and Davidson (1991). But cf. Jackson 1998.
Does analytical moral naturalism rest on a mistake?
139
Again, our OQA* is a plausibility argument against the versions of analytical naturalism favored by Jackson and Smith, rather than a refutation of them. Although the idea of reconstructing the OQA as a plausibility argument against analytical naturalism is not new, we know of no previous attempt at grounding the openness of the Moorean question on the apriority of first-person, comparative judgments of content.5 Our argument may be summarized as follows: OQA*
(1) If “right” could be replaced (without significant loss) by “maximizes utility,” then whether “maximizes utility” and “right” are contentequivalent is not open to doubt on a priori grounds. (2) But whether “right” and “maximizes utility” are content-equivalent is open to doubt on a priori grounds. Therefore, (3) It is reasonable to believe that “right” cannot be replaced (without significant loss) by “utility-maximizing.” (4) Steps (1) through (3) can be iterated for a great number of purely descriptive replacements of “right.” Therefore, (5) It is reasonable to believe that “right” cannot be replaced (without significant loss) by purely descriptive terms. (6) Steps (1) through (5) can be iterated for a great number of purely descriptive replacements of “good,” “ought,” and other moral terms. Therefore, (7) It is reasonable to believe that “good,” “ought,” and other moral terms cannot be replaced (without significant loss) by purely descriptive terms. 5 A confused move in this direction is Steven Ball’s (1988). His attempt to vindicate the OQA against analytical naturalism fails on two counts: on the one hand, he neglects to invoke a priori grounds for semantic intuitions of the sort at work in the OQA. On the other, he fails to recognize that the OQA (charitably reconstructed) is at most a plausibility argument. Charles Pigden (2007) does construe the OQA as a plausibility argument, but he thinks that an adequate OQA would proceed “piecemeal” by invoking a posteriori grounds against each proposed semantic analysis. Thus he takes Moorean questions to raise a posteriori doubts about naturalistic replacements for moral terms by invoking evidence from third-person intuitions about the lack of synonymy with moral predicates. And although Terry Horgan and Mark Timmons (2009: 235) also ascribe to the OQA “considerable intuitive force,” they take it to be “closely related in spirit” to their Moral Twin Earth case, thus resulting in an argument that is stronger than the one proposed here. Cf. Nuccetelli and Seay (2007a).
140
s us a n a n uc c e t e l l i a n d g a r y s e a y
We believe that the OQA* puts the burden of argument on analytical naturalism. What turns dialectical space this way is the appeal to an a priori warrant for Moorean intuitions about the failure of contentequivalence in candidates for replacement of moral terms such as Utilitarian. Given that warrant, the burden of reason rests with the analytical naturalists, who must not only make a compelling case for the possibility of their reductive analyses, but also explain away a priori doubts generated by the OQA*. Merely claiming that correct but non-trivial analyses are possible is not sufficient to overcome those doubts. Whether moral predicates such as “right,” “good,” and “ought” are a priori or conceptually equivalent to some purely descriptive predicates, then, seems open to a priori doubt. We will now show that, in the dialectical context created by the OQA*, the claim that such conceptual equivalences hold amounts to a pattern of dialectical failure, which we shall call “the analytical naturalist mistake.” The type of error we have in mind is a dialectical phenomenon of failure that Moore would perhaps have done better to point out rather than raise the naturalistic fallacy charge against. In fact, our account of what has gone wrong with the analytical naturalists’ defense of their reductive program can make sense of Moore’s confused remarks about the naturalistic fallacy in chapter 1 of Principia Ethica (1993b [1903]: section 12, 64–66). Although his claim to have found a special type of fallacy identifiable as “naturalistic” may well be unjustifiable, his skepticism about certain analyses of ethical naturalists – as well as of other analyses featuring only purely natural terms, or where one term is ethical and the other metaphysical – might be justified on different grounds. Surely we are entitled to be suspicious of any conceptual equivalence whose proponent is unable to overcome a priori doubts about that equivalence. Interpreted in this way, Moore was clearly right in criticizing “Pleasant is the sensation of red” and “Goodness is what’s commanded by God.” Likewise, we are entitled to reject attempted definitions which, in the context of a certain debate, are dialectically abusive in some sense – as in the rather obvious case where abortion is defined as “a form of infanticide,” or “a killing of an unborn baby.” Plainly, in the context of the debate over the moral status of abortion, the claim that “killing fetuses” and “killing babies” are conceptually equivalent could not overcome a priori doubts raised by an OQA-inspired objector. On our view, the analytical naturalist commits a parallel dialectical mistake. For he maintains that, once all negotiations about the content and reference of moral predicates such as “right,” “good,” and “ought”
Does analytical moral naturalism rest on a mistake?
141
are settled (allegedly by convergence of individual conceptions), those predicates will turn out to be equivalent to certain purely descriptive predicates. By advocating such equivalences, he presupposes that he has some evidence or some reason to dismiss the a priori doubts arising from a properly construed OQA. As mentioned above, he dismisses Moore’s OQA on the grounds that a conceptual equivalence could be correct but not trivial or in need of reasoned argument. But that is not an objection to OQA*, whose Moorean question suggests that, for all we can tell now, at least from the first-person perspective, moral predicates do not seem conceptually equivalent to purely descriptive predicates. Thus, the analytical dialectical mistake is simply that of begging the question against a great number of philosophers, from non-naturalists to moral nihilists, who have found some version of the OQA persuasive. In the end, analytical naturalism might turn out to be true, but analytical naturalists must first discharge the burden of reason – which involves producing evidence or reasons that have enough force to overcome OQA*-inspired a priori doubts about the possibility of the proposed reductive program.6 Until then, non-naturalists and moral nihilists (i.e., skeptics and noncognitivists) confronted with this sort of tactic in metaethical debate can dismiss the analytical naturalist proposal by simply showing that it begs the question against reasoning along the lines of OQA*. On our view, this is what Moore should have said on the question of where analytical naturalism goes wrong.7 Recognizing occurrences of the analytical naturalist mistake has a more modest cash value than Moore’s Jackson (1998: 151) envisages the possibility of persisting doubts stemming from the OQA, but he dismisses them as remnants of an obsolete view of moral language. He writes: “It may be objected that even when all the negotiation and critical reflection is over and we have arrived at mature folk morality, it would still make perfect sense to doubt that the right is what occupies the rightness role. But now I think that we analytical descriptivists are entitled to dig in our heels and insist that the idea that what fits the bill that well might still fail to be rightness, is nothing more than a hangover from the platonist conception that the meaning of the term ‘right’ is somehow a matter of its picking out, or being somehow mysteriously attached to, the form of the right.” Given the argument we are pressing here, such a “digging in the heels” defense of analytical naturalism is unpersuasive, since it seems to beg the question against a properly construed OQA. For a different but compatible reply to the “digging in the heels” defense of analytical descriptivism, see Horgan and Timmons (2009: 235). See also Jackson’s rejoinder (2009: 442 ff.). 7 Moore did express some reservations about the problem underlying what he called “the naturalistic fallacy,” but to him they were nothing more than the matter of how to label the mistake. “I do not care about the name,” he wrote. “[W]hat I do care about is the fallacy. It does not matter what we call it, provided we recognize it when we meet with it. It is to be met with in almost every book in Ethics; and yet it is not recognized: and that is why it is necessary to multiply illustrations of it, and convenient to give it a name” (Principia Ethica, ch. 1, section 12 [1993b [1903]: 65–66]). 6
142
s us a n a n uc c e t e l l i a n d g a r y s e a y
original naturalistic fallacy charge, since we do not claim to have thereby refuted even the analytical variety of moral naturalism. We merely think that such recognition puts the onus on those favoring analytical naturalism to provide independent reasons that are strong enough to explain away a priori doubts about the possibility of replacing moral predicates such as “right,” “good,” and “ought” with purely descriptive predicates without significant loss. Our account has the consequence that, rather than its being Moore’s argument that begs the question (as W. K. Frankena famously alleged long ago), it is actually the reductive program of semantic naturalists that begs the question against the Moorean inference. But, then, why has that inference been thought by many to amount to nothing more than a petitio? For one thing, critics may have been misled by Moore’s own inflationary view about the number of reductive programs that have committed what for him appears to be an actual fallacy. As we have seen, he did claim to have found the fallacy in the work of many philosophers and social scientists, including controversial cases such as that of J. S. Mill. In addition, he seems to have conflated analytical and non-analytical varieties of ethical naturalism, charging that they all committed the alleged fallacy. On the other hand, it must be acknowledged that critics have not always given the most charitable reading to Moore’s inference. They have notoriously evaluated the naturalistic fallacy charge in isolation from its dialectical context, which is an OQA properly construed as a plausibility argument against analytical versions of moral naturalism.8 Moreover, even if it makes no sense to charge that such theories commit a naturalistic fallacy, our contention has been that they do seem to rest on a mistake akin to the broader dialectical phenomenon of begging the question. Let us now take stock of our version of Moore’s extended inference. We have argued that OQA*, construed as a plausibility argument, is nonetheless strong enough to generate a priori doubts about the possibility of replacing moral predicates with purely descriptive predicates without significant loss. If we are right, then the reductive program of analytical naturalists such as Jackson and Smith does instantiate a pattern of dialectical mistake: one committed by any argumentative strategy that assumes the 8 Critics of Moore who have read his naturalistic fallacy charge in this way include not only Frankena (1939) but also Darwall, Gibbard, and Railton (1992); Ridge (2008); and Williams (1985).
Does analytical moral naturalism rest on a mistake?
143
possibility of conceptual equivalences that are in fact open to doubt on a priori grounds. As a result, no such strategy could be cogent. A Moorean inference is thus shown to have some force against attempts to reduce the moral to the non-moral by means of taking some expressions in the moral vocabulary to be conceptually equivalent to others in a purely descriptive vocabulary.
ch apter 8
Supervenience and the nature of normativity Michael Ridge
Necessarily, no two situations can differ normatively without differing in some of their descriptive or non-normative properties. Properly understood, this so-called “supervenience” thesis is extremely plausible, perhaps even a priori true. A longstanding challenge for so-called “non-naturalist” theories of the meanings of normative predicates is to explain supervenience. Non-naturalist theories hold that normative predicates refer to irreducible non-natural normative properties. The worry is that nonnaturalism threatens to make supervenience mysterious. For why should one set of properties (the normative ones), which are utterly irreducible to some other set of properties (the descriptive and non-normative ones), necessarily supervene in this way? In this paper, I begin by sharpening this challenge. In particular, I suggest that the problem is not so much explaining why supervenience is true, but explaining how there could be any non-natural normative properties, given supervenience. I explore this challenge in the context of Ralph Wedgwood’s recent and sophisticated defense of his own novel form of non-naturalism. Although Wedgwood engages more directly with this challenge, and with more acuity and sophistication than most non-naturalists to date, I argue that his approach ultimately is implausible. Non-naturalists still cannot plausibly explain supervenience. 8.1 S u pe r v e n i e nc e a n d non-n at u r a l i s m Non-naturalism in metanormative theory maintains that (a) normative predicates refer to normative properties, (b) normative properties are utterly irreducible to any of the properties of the natural sciences (or, indeed any concatenation of descriptive or non-normative properties, I am very grateful to Susana Nuccetelli, Gary Seay, and Ralph Wedgwood for discussion of a previous draft of this material.
144
Supervenience and the nature of normativity
145
whether these are natural or not), and (c) that these properties are at least sometimes instantiated in the actual world. Most non-naturalists typically also assert that we can have substantive normative knowledge, though this epistemological commitment is not definitive of non-naturalism. In spite of its obviously ambitious metaphysical commitments, nonnaturalism is not a view without its attractions. One of the lessons of G. E. Moore’s famous “open question argument” is that attempts to define normative predicates in non-normative or purely descriptive terms seem to fit poorly with the semantic intuitions widespread amongst competent speakers.1 Moore’s own version of the argument has of course been subject to extensive debate and this is not the place to review the various moves in that complex dialectic.2 It is, though, worth noting that the basic idea behind the argument can be refined in ways that are sensitive to more recent developments in the philosophy of language.3 Of course, the putative semantic advantages of non-naturalism come at a metaphysical price. The very idea of non-natural normative properties can reasonably seem, as J. L. Mackie famously remarked, “queer.”4 One respect in which such properties can reasonably seem queer is that they would at one and the same time necessarily supervene on the non-normative and descriptive properties while being utterly irreducible to them. Before elaborating on the queerness of this combination of views, though, it is useful to characterize the relevant supervenience thesis more carefully. For reasons I have defended at length elsewhere,5 I propose to understand supervenience for the purposes of formulating the challenge for non-naturalism and other anti-reductionist views as follows: (S) Necessarily: two possible worlds cannot differ in their normative properties without also differing either (a) in their non-normative properties or (b) in their descriptive properties.
(S) is extremely plausible, to the point that someone who denied it would thereby betray incompetence with normative concepts. To deny (S) would be to allow, for all that has been said so far, that it could have been the case that the world was exactly like the actual world in all of its non-normative and descriptive features, yet such that what Hitler did was not wrong. 1 See Moore (1903). 2 For a recent statement of some of the main moves here, with useful references, see Sturgeon (2003). 3 Here I have in mind the Moorean argument developed by Terry Horgan and Mark Timmons. See Horgan and Timmons (1992). 4 Mackie (1977). 5 For the arguments for this formulation, see Ridge (2007).
146
m ic h a e l r i d g e
Since all the non-normative and descriptive facts are the same in this possible world it will still be true that Hitler killed the same people, had the same intentions, etc., but this is obviously absurd. Indeed, such bare normative differences seem inconceivable. Very plausibly, the normative facts are in some way entirely fixed by the non-normative and descriptive facts, and this is something we can know a priori. Indeed, many have argued that this idea is built into the very meaning of normative discourse, and is in that sense supervenience (à la (S)) can plausibly be taken to be analytic. Moreover, supervenience formulated in this way does not seem to beg the question against any of the leading metanormative theories. I shall therefore understand supervenience in this way. Note that (S) is what is usually referred to in the literature as a “global” supervenience thesis. This is no accident. Global supervenience theses claim that no two possible worlds can differ in supervening respects without also differing in some subvening respect. By contrast, local supervenience theses hold that two numerically distinct objects in one and the same world cannot differ in supervening respects without also differing in some subvening respect. A problem for local supervenience theses is that if the subvening base is widened far enough – say, to include spatio-temporal location – then supervenience may be trivially true. For perhaps no two numerically distinct things can be precisely the same in all of the subvening respects simply because no two things can be in the same place at the same time. In that case, it becomes trivial that no two things can differ in the supervening respects without also differing in subvening respects as well. Global supervenience theses avoid this looming threat of triviality by being couched in terms of whole possible worlds, rather than in terms of individuals within the same world. The issue then becomes which kinds of worlds really are genuinely possible, and which are not. I propose to understand supervenience here as a global thesis primarily to avoid such triviality worries. 8.2 C om pa n ions i n gu i lt ? The non-naturalist asserts that there are normative properties but that they are utterly irreducible to anything naturalistic. Moreover, the traditional inspiration for non-naturalism is the open question argument and its successors.6 The upshot of that argument, if sound, is that normative properties are utterly irreducible, full stop. As Moore clearly indicated, he 6 By “its successors” I have in mind primarily Horgan and Timmons-style “moral twin earth” arguments. See Horgan and Timmons (1992). Some contemporary non-naturalists (Parfit and
Supervenience and the nature of normativity
147
also rejects attempts to reduce normative properties to any concatenation of other non-natural properties. Attempts to define normative notions in terms of God’s will or a Kantian “noumenal self” are also ruled out. Given this line of argument, any prima facie defensible form of nonnaturalism will deny that normative properties are reducible to any other sorts of properties. It is not for no reason that the opening quotation for Moore’s Principia is Bishop Butler’s remark that “Everything is what it is, and not another thing.” Later non-naturalists seem to agree with Moore on this point.7 So non-naturalists are committed to the thesis that normative properties are not reducible to anything characterized in purely descriptive and non-normative terms. Yet the normative supervenes on the descriptive and the non-normative. Why should this be? If normative properties are utterly metaphysically distinct from the descriptive and non-normative properties, then why should it not be at least possible for there to be some normative difference without some descriptive or non-normative difference? The worry is that non-naturalism leaves this obvious necessary truth shrouded in mystery. If normative properties, non-normative properties, and descriptive properties really are, in Hume’s terms, “distinct existences,” then it is very hard to see why it should be impossible for the former to differ with no difference in the latter. In one version or another, this problem has been put to anti-reductionists many times.8 Note, moreover, that this metaphysical mystery is not inevitable. Other theories can explain supervenience without any such mystery. Most obviously, reductionist forms of descriptivism have an easy time explaining supervenience. For on these views, the normative properties just are, or are fully constituted by, a proper subset of the descriptive and non-normative properties, and every property trivially supervenes on itself. Expressivist theories also can avoid any such metaphysical mystery. Very roughly, according to expressivism, the meanings of normative predicates should be understood most fundamentally in terms of the practical role they play in allowing us to express various non-cognitive attitudes, rather than in terms of reference to non-natural or otherwise irreducibly normative properties. The crucial point here is that the expressivist’s explanatory task differs from the explanatory task facing the non-naturalist. Whereas Dancy, for example) like to distance themselves from the open question argument and instead offer arguments “from normativity.” I myself suspect that in so far as the argument “from normativity” is really distinct from any of the arguments in the tradition of the open question argument that it is not very promising, but I lack the space to argue for that here. 7 See, for example, Shafer-Landau (2003). 8 See Blackburn (1984: 182–90; 1998: 315–17); and Mackie (1977: 38–42).
148
m ic h a e l r i d g e
the non-naturalist must explain a metaphysical relationship between two potentially distinct sets of properties, the expressivist instead needs only to explain the sensibility of a practice of normative judgment governed by a supervenience constraint. Moreover, it is not hard to see how such an explanation would go. Part of the point of expressivism is that we use normative discourse to recommend courses of action on the basis of their non-normative and descriptive properties. We use normative judgments to decide what to do in any given non-normatively or descriptively characterized circumstance. There could be no sense given to the idea that we were recommending or deciding on actions on the basis of their non-normative or descriptive features if we did not obey a suitable supervenience constraint. The fact that these other theories do not make supervenience mysterious is perhaps not decisive, but it does speak in their favor. How might the nonnaturalist reply to the charge that he cannot explain how normative properties supervene? The most commonly invoked strategy here has been to pursue what are usually called “companions in guilt” arguments. The basic idea is to argue that we have supervenience without reduction in other areas, and moreover in areas in which expressivism also seems highly implausible. For example, the mental plausibly supervenes on the physical even if the mental is not reducible to the physical. Yet few would be tempted to infer from this that we should be expressivists about mental discourse. Sometimes, defenders of non-naturalism supplement this standard “companions in guilt” strategy with a positive story about supervenience. In my view, this is essential to a fully satisfying defense of non-naturalism. For the companions in guilt approach alone is deeply unsatisfying. One always wonders whether there might not be some relevant difference between the companions in virtue of which they will at some point part company. There is something a little too glib in resting easy with this line of argument, particularly when it is anyway typically deeply unclear what the best theory is in each of the companion cases anyway. One non-naturalist who has been less glib and developed a positive story about supervenience is Russ Shafer-Landau. His explanation of supervenience appeals to the idea that each token instantiation of a normative property is somehow fully constituted by some concatenation of instantiations of descriptive or non-normative properties.9 The idea is that even if each property token is fully constituted by such a
See Shafer-Landau (2003).
9
Supervenience and the nature of normativity
149
combination of non-normative/descriptive properties, the relevant normative properties themselves, construed now as types, need not be reducible to anything couched entirely in non-normative and descriptive terms. Other non-naturalists have made similar suggestions, although typically en passant, whereas Shafer-Landau develops the idea carefully and at some length.10 I have argued against the appeal to the constitution of normative property tokens by non-normative/descriptive property tokens at length elsewhere.11 I will therefore not rehearse my objections to this approach in detail here, but instead will quickly summarize them. The most central objection was that the proffered explanation depends on the assumption that if a given concatenation of token descriptive property instantiations constitutes a normative property in one context that it therefore must constitute that property in any possible context. Without this assumption, the appeal to token property constitution as an explanation of supervenience is a non sequitur, but Shafer-Landau offers no argument for it, nor is it clear how one might argue for it without presupposing what needs to be explained. I further argued that this explanatory strategy rather surprisingly commits the non-naturalist to rejecting both monism and certain forms of context-sensitivity at the level of first-order normative theory, and that these are unwelcome consequences.12 Finally, I argued that the most promising version this strategy ultimately depends on the rejection of an interesting view in metaphysics which goes under the heading of “trope nominalism,” and that this is a worrying hostage to fortune for the non-naturalist. What, though, about the companions in guilt move all by itself, bad taste notwithstanding? An obvious strategy for responding to this move is to cite some relevant difference between the normative cases and its putative companions. The most commonly invoked version of this strategy in the literature points out that the supervenience of the normative is a priori, indeed analytic, whereas supervenience in the case of the companions is plausibly neither. In particular, the supervenience of the mental on the non-mental is hardly a priori, much less analytic. In fact, the majority of human beings at various points in history may well have denied the supervenience of the 10 See, for example, Dancy (2004). 11 See Ridge (2007). 12 For the record, I am now somewhat less confident of the point about context-sensitivity. However, the other points are sufficient without this argument. I am grateful to Campbell Brown for helpful discussion of this particular argument.
150
m ic h a e l r i d g e
mental on the non-mental out of an antecedent commitment to a certain sort of dualism about the mental. For many religious traditions would naturally lead one to such a view of the mental. In my opinion, the non-naturalist has an interesting reply to this point about the analyticity of supervenience in the normative case. For the non-naturalist can argue that this difference between the normative and non-normative cases of supervenience is one which might actually make it easier for them to explain why supervenience is true – or, perhaps, to explain why it stands in need of no explanation. Suppose that supervenience is indeed analytic. Suppose, moreover, that its analyticity is axiomatic. That is, its status as analytic is primitive; it need not be derived from more basic analytic truths as a theorem. This distinction is not commonly made, but is straightforward enough. It is, the reply continues, simply a brute fact about our normative concepts that something cannot count as rightness, e.g., unless it obeys a suitable supervenience constraint. Crucially, on this view the explanation of how any given property could actually satisfy the supervenience constraint, need not be analytic, or even a priori. Here is an analogy that might help clarify the defense I am proposing on behalf of the anti-reductionist. Suppose that I just stipulate that by “shmong” I shall refer to that property which obeys such and such supervenience constraint, and which is also such that someone who fails to do that which is shmong is therefore blameworthy. There shall now be no mystery about why facts about which actions are shmong will conform to the relevant supervenience constraint – that follows trivially from the stipulated meaning of shmong. The same sort of account could, it seems, be given in the case of normative predicates. The meanings of these predicates are not given by stipulation, of course, so the analogy with my hypothetical predicate “shmong” is imperfect. Nonetheless, linguistic facts are true in virtue of conventions, and the relevant conventions could be such as to make it axiomatic by convention that normative discourse is governed by a suitable supervenience constraint. The fact that our actual conventions grew up more organically, and were not the result of an act of stipulation, does not preclude the relevant conventions having this shape. If this much is accepted then there is no mystery about why supervenience is true. The truth of supervenience is therefore no mystery at all, but a triviality – what David Hume would call a “trifling truth.” All of this seems to me to be entirely consistent with non-naturalism, in which case calling attention to the analyticity of supervenience does little to undermine non-naturalism.
Supervenience and the nature of normativity
151
Suppose we concede the reply so far as it goes, and allow that supervenience is a kind of axiomatic analytic truth. In that case, the challenge to explain why supervenience is true has been met – or perhaps a better way of putting the point is that the challenge itself has been defanged, in that there isn’t much to explain in the first place. However, the nonnaturalist must be careful not to celebrate pyrrhic victories. Recall that supervenience claims that there can be no normative difference between worlds without some non-normative or descriptive difference. So supervenience will be trivially true if there are (and could be) no normative differences between worlds. One way in which this could be true would be if there are (and could be) no normative properties – that is, if the so-called “error theory” is true. Non-naturalism, though, rejects the error theory and maintains not only that normative predicate purport to refer to normative properties but that they succeed in so referring and moreover that those properties are at least sometimes actually instantiated (and so obviously possibly instantiated). The point now is that the truth of supervenience as an analytic truth is consistent with there necessarily being no normative properties. This suggests that the real challenge for the non-naturalist is not to explain why supervenience is true. Instead, the challenge is to explain how there could be non-natural normative properties, given the truth of supervenience. In Kantian lingo, the challenge is to answer a sort of “how possible” question. Moreover, this provides a clear vehicle for the original metaphysical worry about non-naturalism. For the worry is that it is simply a metaphysical mystery how there could be properties which are entirely irreducible to the descriptive and non-normative, yet which supervene on the descriptive and non-normative. On the strength of this mystery, one might concede supervenience as an analytic truth and argue that the semantics proposed by the non-naturalist leads inexorably to the error theory, on pain of deep metaphysical mystery. However, let us back up for a moment to keep dialectical score. So far we have seen that the non-naturalist’s standard reply to the “explain supervenience” challenge is to invoke “companions in guilt.” One standard rejoinder to the companions in guilt move is to call attention to the fact that supervenience in the normative case is analytic, but that it is not analytic in the companion cases. This, though, seems not to help the opponent of non-naturalism, because the non-naturalist can now claim that supervenience is a thoroughly trifling analytic truth, and so stands in no need of explanation. This simply moves the dialectical bump in the rug, in that the challenge now is not to explain why supervenience is true,
152
m ic h a e l r i d g e
but to explain why we should suppose there are (or could be) any normative properties construed as non-natural, given supervenience. To my mind, this does help clarify the form and force of the challenge to “explain supervenience,” which really is better understood as a challenge to explain how there could be non-natural normative properties, given supervenience. However, this dialectic does mean that the opponent of non-naturalism needs a new rejoinder to the companions in guilt move. Granted there is still a metaphysical challenge to explain how there can be properties which supervene on some other set of properties without being reducible to them, but this metaphysical issue does crop up in other domains. The fact that supervenience itself is analytic in the normative domain basically seems to be irrelevant to the challenge itself, properly construed. In that case, though, the fact that the normative case differs from the companions in terms of supervenience being analytic is not a promising reply to the companions in guilt defense. The opponent of non-naturalism needs a different rejoinder. To some extent, I am here willing to rest easy with the point that the companions in guilt move is anyway deeply unsatisfying, for the reasons discussed above. Moreover, I have argued (at length in my reply to ShaferLandau, briefly summarized above) that explanations which go via the idea of the constitution of token property instantiations is not very promising in the context of normative properties, and in part this is due to features peculiar to the normative. Furthermore, the putative companions may turn out to be cases in which some form of reductionism is plausible after all, in which case the argument cannot even get off the ground.13 Finally, some of the companion cases may be ones in which the opponent of the non-naturalist can plausibly “follow the argument where it leads.” For example, so-called “essentially contested concepts”14 like the concept of democracy and art might seem to denote irreducible properties, in that they stubbornly elude reductive definition, yet still supervene on some suitably characterized set of more basic properties. Perhaps, though, these concepts are covertly normative, and as such amenable to an expressivist treatment.15 13 For a defense of a more reductionist approach to metaphysics, see Jackson (1998). Jackson also defends reductionism about ethical properties, and presumably would take the same line on the normative more generally. Although I find his arguments in the ethical/normative case unpersuasive, this is not the place to get into the details of that debate. 14 See Gallie (1955–56). 15 I in fact argue for such a view of essentially contested concepts elsewhere (in currently unpublished work).
Supervenience and the nature of normativity
153
In any event, a truly satisfying defense of non-naturalism requires some plausible positive explanation of how there could be non-natural normative properties given supervenience. So far, we have seen that the appeal to the constitution of token normative property instantiations does not succeed in its attempt to provide such an explanation. Until recently, this has de facto been the “only game in town” for the non-naturalist, in that those few non-naturalists who have taken the challenge seriously enough to offer a positive explanation have offered an explanation of this form.16 More recently, though, Ralph Wedgwood has offered a new and very different non-naturalist reply to this challenge. In the remainder of this paper, I explain his reply (in the following section) and then argue that it is implausible (in the final section). 8.3 T h e n at u r e of nor m at i v i t y ? Wedgwood’s view is sophisticated and has many “moving parts,” so I shall not be able to do it full justice in this short space. I shall instead just very briefly summarize the key points that are relevant here. I begin with an account of the semantics Wedgwood offers for normative predicates, so as to put his metaphysical views into their proper context. Wedgwood argues that, necessarily, if someone makes a judgment of the form “I ought to Φ,” and suffers from no relevant uncertainty, and Φ-ing is “of the relevant kind,” then he either intends to Φ or is irrational. The caveat about relevant uncertainty is necessary to avoid implausible claims about cases in which the agent’s judgment that he ought to do one of two things but does not know which is correct. The caveat about “of the relevant kind” restricts the thesis to those in which one’s forming an intention will have some impact on whether one in fact Φs. Wedgwood further argues that since people in general are disposed to be rational, this rational connection between normative judgment and motivation in turn entails that people have a general disposition to intend to do whatever they judge they ought to do. Finally, on Wedgwood’s account, knowing how to use normative predicates in a way that accords with this action-guiding constraint more or less exhausts what is necessary for competence with normative predicates. One advantage of this account of semantic competence is that it seems well suited to accommodating the ubiquity of fundamental normative disagreement. Indeed, given how minimal the account of linguistic
16
Here I have in mind Russ Shafer-Landau and Jonathan Dancy; see references above.
154
m ic h a e l r i d g e
competence is, it should be obvious that it can accommodate extremely ubiquitous fundamental normative disagreement. For the main worry about such disagreement is that, at least on many referential accounts of normative semantics, it threatens the idea that speakers were using normative predicates with the same meaning. Furthermore, if the account as a whole works, and Wedgwood can deduce a unique referent from the proposed conceptual role, then there will be a broad kind of continuity between the semantics for normative predicates and for non-normative predicates. That is, in both cases, we have accounts in which reference is fixed by conceptual role. Even if the account delivers all it promises, the analogy with non-normative predicates will not be precise, of course. For in the normative case the conceptual role is distinctively practical and action-guiding. So the theory’s account of the continuity of the normative and the non-normative will here hold only at the level of abstraction given by “meaning is fixed by conceptual role.” However, accommodating what is unique about normative discourse by highlighting the distinctively practical conceptual role associated with normative predicates is a virtue, rather than a vice, of the account. One worry about this approach, which I simply mention but will not pursue here, is that the proposed semantics seems “too good to be true.” In particular, the idea that we can somehow deduce a unique referent for our normative terms on the basis of such a slender semantics is hardly obvious. One way to see this point is that Wedgwood’s account of semantic competence is very much the sort of account one would expect an expressivist to give. The expressivist, though, insists that our most fundamental theory of the meanings of normative predicates should not be understood referentially at all, but simply in terms of the practical and expressive role that they play. Wedgwood has more to say about how his theory can secure a unique referent on such a slender basis. While I am not convinced, I shall not pursue this line of argument here.17 Let us suppose that Wedgwood’s theory does manage to provide unique referents for our normative terms, assuming for the sake of argument that there are irreducible normative properties for those terms to take as their referents, and see how he goes on to explain supervenience. A proper appreciation of his take on supervenience requires 17 For some discussion of this worry about Wedgwood’s theory, see Schroeter and Schroeter (2003).
Supervenience and the nature of normativity
155
some discussion of his idea that “the intentional is normative,” which he goes so far as to say is the “key to metaethics” (Wedgwood 2007: vii). Wedgwood’s view is that the individuation of intentional states of mind – that is to say, states of mind which are “about” something – states of mind that “have a content” – is irreducibly normative. On this view, neither the intentional nor the normative is metaphysically more basic than the other, but each can only be understood partly in terms of the other. For example, to be a belief on this account just is, at least in part, to be such as to be essentially generally regulated by a suitable norm of rationality. Being essentially regulated by the norms essentially associated with a given concept turns out to consist in the norms being such that an agent does not have the concept in question unless she is generally disposed, ceteris paribus, to conform to the relevant norms(s). This, however, is still not enough to fix correctness conditions, on Wedgwood’s account. On Wedgwood’s account there must also be some independently intelligible “point” or value to the norms themselves. Here, on Wedgwood’s view, the relevant further points are truth (in the case of belief) and choiceworthiness (in the case of plans). So on Wedgwood’s view the intentional is normative. This is true not in the sense that the intentional is reducible to the normative (or vice versa, for that matter). Rather, the intentional can only be understood in (at least partly) normative terms, and the normative can be understood only in (at least partly) intentional terms. The two classes of properties are metaphysically interdependent, without either having ontological priority over the other. This makes the supervenience challenge more difficult to press simply because this version of Anti-Reductionism can, it seems, at least explain why there can be no normative difference without some difference in intentional states. For on this view, normative facts in some sense just are intentional facts. Trivially, everything supervenes on itself, so in this sense it is trivial that Wedgwood’s account entails that the normative supervenes on the intentional. Note that Wedgwood’s claim that the intentional is normative is at the level of types, and not of tokens. So unlike the misguided strategy of explaining supervenience in terms of token constitution, this explanation is at the right level – it appeals to a universal and necessary fact about the essence of the normative and the intentional qua types rather than a claim about the realization of various token instantiations of normative properties.
156
m ic h a e l r i d g e
Recall that in motivating supervenience, I argued that to deny supervenience would, quite absurdly, be to allow that there could be a world just like our own save that in that world Hitler’s extermination of the Jews was not wrong. In pressing that argument, I relied on a form of supervenience according to which there can be no normative difference without there also being some non-normative/descriptive difference as well. However, Wedgwood’s theory can explain this supervenience thesis quite easily. For in my sense intentional properties are descriptive properties – they are properties picked out with what are intuitively purely descriptive terms from psychology. However, on Wedgwood’s view, normative properties just are certain psychological properties – the intentional is normative. So his account can trivially explain the supervenience of the normative on the descriptive, given that the descriptive includes the psychological. Given that the intentional will plausibly figure in empirical psychology and on Wedgwood’s account has causal powers (for Wedgwood does think that the normative has causal powers), it will also not do to formulate supervenience in terms of the normative supervening on the natural. For once again, the thesis that the intentional is normative will allow Wedgwood quite easily to explain supervenience so formulated. This does not spell the end for the supervenience challenge, though. For we can reformulate supervenience so that it is not so easily explained by the idea that the intentional is normative. As Wedgwood recognizes, the issue now will be how to explain the supervenience of the normative/ intentional on the non-normative and non-intentional – or perhaps on the physical. However, these supervenience theses are not so obviously compelling and irresistible as the ones with which I began. It is hard to use examples that are as straightforward as the Hitler example with which I motivated my original supervenience thesis to motivate supervenience couched in terms of the non-normative/non-intentional, or in terms of the physical. So this is the first point; Wedgwood’s sophisticated form of nonnaturalism interdefines the psychological and the normative in a way that makes it much harder to state the challenge to explain supervenience in a way that invokes a version of supervenience which is so immediately compelling. Second, though, and to his credit, Wedgwood himself takes the challenge to explain supervenience more seriously than previous nonnaturalists. Indeed, he is admirably clear and candid about how far the
Supervenience and the nature of normativity
157
non-naturalist in his view must go to explain supervenience in so far as it is true and otherwise mysterious. Wedgwood invokes a familiar distinction between “strong supervenience” and “global supervenience” as follows: (SS) [Strong Supervenience] For all A-properties A*, all possible worlds w, and all individuals x, if x has A* in w then for some B property B*, x has B* in w, and for all worlds v that are possible relative to w, and all individuals y, if y has B* in v, y also has A* in v. (Wedgwood 2007: 211)
Wedgwood notes that most philosophers assume that SS, so formulated, entails global supervenience, which he formulates as follows: (GS1) For any two possible worlds w1 and w2, if w1 and w2 are indiscernible with respect to B-properties then they are indiscernible with respect to A-properties as well. (Wedgwood 2007: 211)
Wedgwood then points out quite rightly that GS1 in fact follows from SS only in some modal logics, but not others. His overall strategy is then to offer a novel explanation of SS, argue against GS1, and provide a modal logic which blocks the inference from SS to GS1. In particular, the rejection of the inference from SS to GS1 commits him to rejecting S5, which most modal logicians take to be the correct modal logic. However, Wedgwood argues that this is not too great a price to pay for his Anti-Reductionism, particularly since, in his view, S5 is independently problematic. The first stage of his strategy is to explain SS. To put this explanation in its proper context, I should first say that what really troubles Wedgwood is not SS itself, but what he calls “specific supervenience theses.” These are claims of the form, “For such and such B property B*, necessarily if something has B* then it also has such and such A property A*.” Myself, I am not terribly happy with calling such claims “supervenience” theses, as they are not really of the form “no B difference without some A difference,” but that is a boring terminological point. Wedgwood is troubled by the fact that his account seems to entail that there are infinitely many such “specific supervenience theses.” Why should we not just take this infinite plurality as itself a primitive modal fact? In Wedgwood’s view, this would introduce an element of “chaos” into the modal realm, which we intuitively believe should be orderly and structured. Wedgwood therefore sets himself the task of explaining all of these so-called “specific supervenience theses” by showing how they follow from SS, and then offering a kind of explanation of SS itself.
158
m ic h a e l r i d g e
In fact, there is a dialectical reason that Wedgwood must offer some deeper explanation of these specific supervenience theses. For on his account, if these theses were the most basic metaphysical necessities, then it would seem that they are on his account not merely metaphysical necessities, but that the disjunction of all the relevant physical properties would be an essential feature of the normative/intentional property in question. For on Wedgwood’s account, the essential properties of a thing form a proper subset of its metaphysically necessary properties – those which explain all of the other necessities without themselves being explained by any deeper metaphysical truth. As Wedgwood himself notes, this in turn would support an argument (given by Kit Fine, among others) for a kind of reductionism, according to which the relevant normative/intentional property just is the associated disjunctive physical property. How, then, does Wedgwood explain SS? He begins by discussing the supervenience of pain on the physical. He argues that SS in this case is made true by the fact that, necessarily, for all individuals y, if y has physical property B, then y is in pain. He then argues that such modal facts can be explained by facts about the essential nature of pain in conjunction with a contingent fact about our world. The proposed facts about the essential nature of pain are (1) that every instance of pain must have some physical property “regularly co-instantiated” with it, where “regular coinstantiation” is non-accidental co-instantiation, and (2) that if a physical property is regularly co-instantiated with pain, then in all possible worlds in which the same fundamental physical properties are instantiated as in the actual world, anything that has that physical property is also in pain. The contingent fact is that some physical property (the firing of C-fibers, say), is in fact regularly co-instantiated with pain. The strategy is meant to generalize to all psychological facts, and ipso facto to all normative facts, since the intentional (a subset of the psychological) is normative on this view. Wedgwood then argues that if we reject the S5 axiom, according to which whatever is possibly necessary is necessary, that GS1 does not follow from SS. He therefore explains SS, which is all the supervenience he thinks needs to be explained, since GS1 can be divorced from SS by rejecting the modal logic given by S5. This is because GS1 makes a claim about all possible worlds, no matter how dissimilar they may be. If what is possible at a world can be relative to that world, then the fact that something is possible in one world does not entail that it is possible
Supervenience and the nature of normativity
159
in a physically very dissimilar world. Wedgwood further argues that the reasons for rejecting S5 also give us good reason to reject S4, which claims that if something is necessary then it is necessarily necessary. Although the point is not made explicit in his discussion, Wedgwood actually needs to reject S4 to block the inference from SS to GS1. For otherwise the necessities which hold in the actual world will also hold in all other worlds, even very dissimilar ones, but his account cannot explain why this would be. This is no objection to his overall view, though, but a statement of it, since Wedgwood does anyway argue that we should reject the S4 axiom as well as S5. Finally, Wedgwood argues that rejecting S5 is not too high a price to pay, on the grounds that S5 is anyway dubious. Wedgwood’s argument is that the only real argument for S5 relies on the premise that modal claims are really only abbreviations of claims about what worlds there are and about what is the case at those worlds. This, though, is an optional analysis of modal claims, and one that Wedgwood thinks we should reject. He defines a possible world as a maximal consistent set of propositions, but insists that some of the relevant propositions can themselves be modal. Instead of analyzing modal claims as claims about what worlds there are, he instead takes the notion of metaphysical necessity to be an ultimately primitive metaphysical notion, albeit one with “tight conceptual connections” with the notion of essence, other modal notions, and the logic of counterfactuals. Crucially, Wedgwood’s alternative account, unlike accounts which define modal claims in terms of possible worlds, is consistent with the idea that what is possible can be world-relative. Wedgwood concludes by remarking that once we give up the S5 axiom (that whatever is possible is necessarily possible) it is a short step to abandoning the S4 axiom too (according to which if something is necessary then it is necessarily necessary). Here the suggestion is that once we have allowed that contingent facts can make a difference to which modal facts obtain that it would be arbitrary to claim that they can make a difference only to what is possible and not also to what is necessary and indeed to what is impossible. This completes my summary of Wedgwood’s account of supervenience. It is not hard to see why his account is less vulnerable to the challenge to explain supervenience than other Anti-Reductionist theories. Nonetheless, his ingenious approach ultimately does not adequately meet the challenge, or so I shall argue in the following section.
160
m ic h a e l r i d g e
8. 4 S u pe r v e n i e nc e , e s s e n t i a l i s m, a n d moda l l o g ic Let us begin with the first stage of Wedgwood’s strategy – his explanation of SS. Note that Wedgwood’s proposed explanation of SS itself rests on a very specific form of essentialism about the mental. Recall that in the example of pain, the proposed essentialist theses were that it is essential to the nature of pain that (1) every instance of pain must have some physical property “regularly co-instantiated” with it, where “regular co-instantiation” is nonaccidental co-instantiation, and (2) that if a physical property is regularly co-instantiated with pain, then in all possible worlds in which the same fundamental physical properties are instantiated as in the actual world, anything that has that physical property is also in pain. For the strategy to generalize, moreover, he will need isomorphic essentialist claims to hold for all other mental states – or, at least, for all intentional mental states. The idea is then to combine these essentialist claims with analogous contingent claims about which physical property is regularly co-instantiated with the relevant mental property, and thereby explain SS. What, though, explains these putative essentialist theses about the nature of pain – and, by extension, about belief, desire, intention, and the like? Indeed, properly understood, these essentialist claims just amount to a supervenience claim. For (1) and (2) together already entail SS. So the proposed explanation of supervenience amounts to the thesis that the supervening properties are essentially such as to supervene. Without some deeper metaphysical explanation of why these essentialist theses are true, this hardly seems like a compelling explanation of supervenience. Taking a necessary truth about some class of properties and just reformulating it as a thesis about the “essential nature” of that class of properties is not explanatory progress. One is reminded of the famous explanation given by Molière’s doctor of how opium puts people to sleep – by virtue of its virtus dormitiva. Admittedly, claims like (1) and (2) do add some content to the original supervenience thesis. For on Wedgwood’s account, the notion of an essence is itself irreducible, and hence does not just amount to a claim about modality. Furthermore, (1) and (2) also say something about the way in which the mental supervenes on the physical. For (1) and (2) restrict supervenience to “physically similar” worlds. However, this last point adds little to SS, properly understood. In effect, it simply spells out what it is for one world to be possible relative to another one – relative possibility is, for these purposes, understood in terms of the physical similarity of the worlds.
Supervenience and the nature of normativity
161
The objection, however, was not that the appeal to the essential nature of the relevant states of mind did not say anything over and above supervenience. The objection was rather that a necessary supervenience thesis is not well explained simply by reformulating that modal truth and asserting that it reflects the essential nature of the supervening properties. For anyone initially puzzled by the supervenience thesis will be equally puzzled about why the essentialist theses on offer should be true. Indeed, given that mental properties like the property of being in pain are, on the proposed account, irreducible, why should they be essentially such as to be regularly co-instantiated with some physical property and such that in any physically similar world, the physical property with which they are regularly co-instantiated suffices for the presence of the relevant mental property? We must be careful to avoid begging the question against Wedgwood here. On his view, facts about the essential natures of things are themselves irreducible facts. So the challenge here cannot be to explain what makes it true that pain has such and such essential nature. For if essentialist facts are irreducible, then they may be bedrock, and not “made true” by any deeper underlying facts. Presumably, Wedgwood’s thesis is that his proposed essentialist theses are metaphysically bedrock, and not explained by any metaphysically deeper essentialist theses. We can then understand the challenge to his approach as the challenge to explain why, given anti-reductionism, we should find the specific essentialist theses Wedgwood puts forward plausible. Why, in other words, should we believe that such essentialist claims are both true and metaphysically bedrock, given anti-reductionism? It would, of course, be easy enough to see why something like the theses Wedgwood puts forward would be true if the mental/normative were somehow reducible to the physical. Given that we have rejected any such reductionism, though, why should we find it plausible that the essential nature of a given mental property is to be such as to be regularly co-instantiated with a physical property such that that physical property is sufficient, in all physically similar worlds, for the instantiation of the mental property in question? Furthermore, the fact that essentialist facts are irreducible does not by itself entail that every essentialist fact is beyond explanation. For we may explain one essentialist thesis in terms of another, even deeper essentialist thesis. Intuitively, the theses Wedgwood puts forward about pain do not seem like bedrock essentialist theses. Indeed, they seem themselves to “cry out for explanation.” In principle, they could be given a deeper metaphysical explanation, albeit not in terms of brute modal facts, given
162
m ic h a e l r i d g e
the irreducibility of facts about essences. Instead, they could, in principle, be explained by deeper essentialist theses about the nature of pain. A reductionist thesis could, for example, explain why they are true. On Wedgwood’s account, though, no such reductionist explanation is forthcoming, nor it seems is any other explanation. Yet the sorts of essentialist theses Wedgwood puts forward seem themselves to “call out for explanation,” as he sometimes himself puts it. It seems on the face of it queer that a mental property could be entirely irreducible to the physical and yet in its essential nature be such as to be regularly co-instantiated with some physical property which is in turn such that it suffices for the mental property’s instantiation in every physically similar world. Taking such facts to be metaphysically bedrock, and not such as to be explained in terms of more basic essentialist theses, leaves us with a very queer conception of mental properties. One could be forgiven for thinking that Mackie’s allegation that irreducible normative properties would have to be exceedingly “queer” was here vindicated after all. The queerness of the mental and normative properties so conceived is exacerbated by the way in which Wedgwood is forced to restrict the necessary connection between a given mental property and a given physical property to worlds which are “physically similar.” For imagine someone who accepts anti-reductionism about the mental and the normative, but is skeptical about both supervenience and Wedgwood’s essentialist theses. Such a skeptic will appeal to the intuition that there could in fact be a world physically identical to our own, but which differs in some of its mental properties. Moreover, such a skeptic will be unimpressed by the essentialist theses Wedgwood puts forward, and wonder why anyone would find those plausible, given the irreducibility of the mental/normative. The fact that even Wedgwood allows that the mental/normative property in question can in principle be instantiated without the underlying physical property with which it is in fact regularly co-instantiated does not help matters. For if this divorce between the mental/normative property in question and the associated physical property can happen in physically dissimilar worlds, the skeptic might argue, why can it not also happen in physically similar ones? Once again, we have something which intuitively cries out for explanation, but for which no explanation is forthcoming. Let us suppose for the sake of argument that this objection can be met. Even then, Wedgwood has explained only SS – Strong Supervenience, as he formulates it. He has not yet explained GS1, nor does he intend to explain it. Instead, he argues that we should simply reject GS1. He argues
Supervenience and the nature of normativity
163
that GS1 follows from SS only given certain assumptions in modal logic which he rejects – the S5 and S4 axioms. However, this leaves Wedgwood open to two further objections. First, GS1 is itself intuitively plausible in its own right. That is, anyone initially sympathetic to the supervenience of the mental/normative on the physical, or on the non-intentional/non-normative, will find GS1 intuitively plausible, or so it seems to me. I prefer the latter formulation, in terms of the non-intentional/non-normative more compelling simply because it widens the subvenient base to in principle include properties which are non-physical but also non-normative and non-intentional. To my mind, this makes GS1 considerably more intuitively appealing. Wedgwood focuses on GS1 formulated in terms of the physical because he wants to show how his view is compatible even with a very robust form of naturalism – namely, physicalism. He therefore formulates all of the metaphysical issues in terms of the physical on the grounds that if his account can be squared even with such a radical form of naturalism as physicalism then ipso facto it can also be squared with more modest forms of naturalism. However, this dialectical move works only in so far as he does indeed try to accommodate the naturalist thesis on offer. In the case of Wedgwood’s explanation of SS (Strong Supervenience), this is a fair point. If he could explain even a physicalist version of SS (my objections above notwithstanding) then that would put him in a strong position with regard to weaker forms of naturalism. However, in the case of his rejection of GS1, he makes his position easier by characterizing GS1 in terms of such a strong form of naturalism. For the fact that GS1 is formulated in terms of supervenience on the physical, rather than on the intuitively broader base of the non-intentional/non-normative, makes GS1 more controversial, and hence more plausible to reject. Presumably, Wedgwood would reject GS1 formulated in terms of the global supervenience of the normative/intentional on the non-normative/ non-intentional as well as on his canonical physicalist formulation. At any rate, none of his strategies for explaining SS seem likely to carry over to GS1 whether formulated in terms of the non-normative/non-intentional. Yet, GS1 so formulated has considerable immediate plausibility. Trivially, GS1 formulated in terms of the non-normative/non-intentional is at least as plausible as GS1 formulated in terms of the physical. For given non-naturalism, all physical properties are non-normative and non-intentional. However, nothing we have assumed so far entails that all non-normative and non-intentional properties are physical properties. So GS1 formulated in terms of the physical is strictly logically stronger than
164
m ic h a e l r i d g e
GS1 formulated in terms of the broader class of all the non-normative and non-intentional. Moreover, in so far as physicalism is itself plausible, GS1 formulated in terms of the physical already has considerable plausibility. To see how the move from the physical to the non-normative/nonintentional makes GS1 even more plausible, consider the following combination of views. First, suppose that pain and pleasure are purely qualitative states of mind, and not intentional. This is not entirely uncontroversial, as some people do defend representational views of pain, e.g., George Pitcher (Pitcher 1970). However, the idea that pain and pleasure must be understood in terms of a range of distinctive qualitative feelings is not without its plausibility. Given the intimate link between the normative and the intentional on Wedgwood’s view, we can infer from their not being intentional that the properties of being in pain and being in pleasure are also not normative properties. So, on this view, the property of being in pain and the property of experiencing pleasure are non-intentional and nonnormative. Now consider any view according to which reasons for action are entirely a function of how actions promote pleasure or reduce pain. Such a view is also not obvious, but hedonistic views have had their supporters over the years. It would, in any event, be unfortunate if the possibility of a hedonistic view were foreclosed on purely metaphysical grounds. Prematurely “closing the question” of hedonism in this way would not fit well with the spirit of the Moorean OQA intuitions which motivate AntiReductionism in the first place. On the view I have just sketched, it follows that there can be no normative difference between any two worlds without there being some difference in terms of pain or pleasure. Since the properties of being in pain and the property of experiencing pleasure are on this view themselves non-normative and non-intentional properties, this entails a version of GS1 which is couched in terms of the non-normative and non-intentional. Crucially, nothing in this line of thought presupposes that pain and pleasure can themselves be understood in physicalist terms. So a version of GS1 couched in terms of the non-normative and nonintentional is not only logically weaker than its physicalist cousin. It also can be motivated in ways that do not presuppose physicalism. Indeed, the purely qualitative view of pain and pleasure I invoked to motivate GS1 so construed does not sit very well with physicalism. Wedgwood at one point suggests that GS1 has gained more plausibility than it deserves because it has not been carefully distinguished from what he calls “GS2”:
Supervenience and the nature of normativity
165
(GS2) For any possible world w1, there is no world w2 possible relative to w1 such that w2 is mentally discernible from w1 but not physically discernable from w1. (Wedgwood 2007: 215)
Because it is formulated in terms of what is possible from a given world, this formulation is one that Wedgwood can and does accept. The suggestion is that those attracted to GS1 have not carefully distinguished it from GS2, and that this provides a kind of debunking explanation of the seeming plausibility of GS1. There are two problems with this debunking strategy, though. First, most philosophers accept S5, and so it is unlikely that their intuitions about GS1 and GS2 would diverge, in so far as their intuitions are informed by their theoretical commitments, anyway – and intuitions are to some extent informed by theory. Second, I doubt very much that once GS1 and GS2 are carefully and explicitly distinguished many philosophers would find one more or less intuitively plausible than the other, even putting the issue about S5 to one side. That is in effect a prediction about what philosophers will find intuitive, and hence rather speculative, but then Wedgwood’s debunking diagnosis is itself rather speculative as well. In any event, Wedgwood’s take on GS1 is open to two objections. First, he is mistaken in assuming that the only reason for embracing GS1, whether in its physicalist guise or in the weaker non-normative/ non-intentional guise just laid out, is that it follows from some form of SS. GS1 has considerable immediate plausibility. In which case, showing how we can logically resist the move from SS to GS1 (on either construal) does not yet show that we should reject GS1 (on either construal). Since Wedgwood’s account cannot, by his own lights, explain GS1, we need a better argument against GS1 (on either construal). Second, as Wedgwood is of course himself aware, abandoning S5, and indeed S4, does at least prima facie seem somewhat ad hoc, and like a heavy theoretical price to pay to salvage non-naturalism. Indeed, as Wedgwood himself notes, the S5 and S4 axioms are accepted by most modal logicians as figuring in the most plausible modal logic. Furthermore, Wedgwood does not directly argue that S4 and S5 are false. Instead, he argues that one way of arguing in favor of S4 and S5 is unsound. In particular, Wedgwood argues that any argument which presumes that modal claims are just abbreviated claims about what possible worlds there are, is unsound. According to Wedgwood, we should not understand modal claims in this way. Even this argument is not obviously correct, but I shall not here engage in a debate about how modal claims are best understood. For there is a much simpler point to make against Wedgwood’s argument, which is
166
m ic h a e l r i d g e
that this reductionist view of modal discourse is surely not the only reason one might endorse S4 or S5. Even those who agree that modal discourse is not reducible to discourse about possible worlds might simply find S4 or S5 immediately and intuitively plausible. In so far as it is legitimate to invoke one’s modal intuitions, and it seems on Wedgwood’s general epistemology that this will be legitimate, such intuitive evidence might be enough to support S4 and S5. Moreover, S4 and S5 do in fact have considerable pre-theoretical intuitive support. I have done no polls here, but I would be surprised if ordinary people did not find the inference from “x is possible” to “it is possible that x is possible” a plausible one, once they came to understand it. Similarly, the inference from “x is necessary” to “it is necessary that x is necessary” will seem plausible to most ordinary people. Admittedly, most ordinary people will find these claims rather weird, but that is beside the point so long as they nonetheless have clear intuitions about them once they do understand them. Of course, Wedgwood could argue that these intuitions are somehow confused, or ill-informed, or should for some other reason be abandoned. He has, however, so far offered no such argument. Instead, he has shown only how one particular argument for S4 and S5 goes wrong – an argument which invokes a controversial premise about modal discourse being abbreviated talk about what possible worlds there are. Since this is not the only reason one might endorse S4 or S5, and probably not the strongest reason anyway (the premise seems more controversial than the conclusion, which is never a good sign), some further argument is needed here. There is, moreover, something rather strange about the overall view that emerges here. For on the proposed view, if S4 or S5 did turn out to be true, then there would be a form of supervenience which, given Wedgwood’s non-naturalism, would be utterly inexplicable. This, however, would give us good reason to doubt that there are any normative properties, assuming the truth of Wedgwood’s semantics. For that semantics would in that case tell us that normative predicates purport to refer to properties which supervene globally in a way which cries out for explanation but at the same time is utterly inexplicable. That would be enough to make it reasonable to deny that there are any such properties. In that case, whether there are in fact any normative or intentional properties turns out to depend on seemingly esoteric questions in modal logic. That is, to say the least, a very strange result. Indeed, the idea that if we discovered that S4 or S5 were true we should on the strength of that
Supervenience and the nature of normativity
167
logical discovery all become both normative nihilists and eliminativists about beliefs and desires is a deeply bizarre connection. 8.5 C onc l us ion By way of conclusion, I shall not try to summarize the many twists and turns of my paper as a whole. I shall instead restrict myself to a brief summary of my objection to Wedgwood’s view, which in my view is by far the most powerful and serious non-naturalist attempt to explain supervenience to date. First, Wedgwood’s proposed explanation of SS (Strong Supervenience) is itself problematic, in that it invokes essentialist claims which are just essentialist variants on SS and which stand just as much in need of explanation as SS itself. Wedgwood also needs to reject GS1 (Global Supervenience), but here he makes his job too easy by assuming that GS1 should be understood in terms of supervenience on the physical. A sterner test of his theory would focus on whether we should also reject GS1 couched in terms of supervenience on the non-normative and nonintentional. In any event, GS1 on either construal has considerable plausibility in and of itself. So simply showing that GS1 does not follow from SS is not enough to give us any independent reason to reject GS1 on either construal. Finally, of course, the inference from SS to GS1 relies on the rejection of axioms of modal logic that most modal logicians take for granted. This is not only somewhat ad hoc, but also a large theoretical cost to pay for one’s Anti-Reductionism. Nor does the fact that we can construe modal discourse as not merely making abbreviated claims about what worlds there are show that we should reject S4 and S5. At most, it shows that one argument for S4 and S5 is unsound. However, S4 and S5 themselves have considerable immediate intuitive appeal. Note the structural analogue with Wedgwood’s treatment of GS1. Rather than offering an independent argument against GS1, or against S4 and S5, Wedgwood instead argues in each case that one argument for those theses is unsound. In so far as the arguments he criticizes are not the only grounds on which one might embrace the theses in question, this is not yet a compelling case for rejecting the theses in question. Yet without some genuinely independent argument against those theses, his position, in particular his rejection of S4 and S5, is indeed ad hoc. Finally, although this is hardly the strongest of the objections I have raised to Wedgwood’s view, there is something very strange about the
168
m ic h a e l r i d g e
idea that normative nihilism and eliminativism about the intentional would follow from something as esoteric as the truth of S4 and S5 in modal logic. Such a view seems to make the normative and the intentional hostage to what are intuitively the “wrong kinds of fortunes.” It is almost as strange as, e.g., the suggestion that the possibility of freedom of the will depends on whether some obscure mathematical theorem is sound. Perhaps truth here is stranger than fiction, but the fact that this connection seems so utterly bizarre should at least make one hesitate before taking the plunge Wedgwood implicitly invites us to take.
ch apter 9
Can normativity be naturalized? Robert Audi
Normativity is of great interest in ethics, aesthetics, epistemology, indeed in any discipline that attempts to account for certain standards by which we guide our thought and action. In the current intellectual climate, which is often dominated by naturalism and an emphasis on the scientific habit of mind, normativity presents a special challenge because normative statements – above all statements saying what we ought to do, or are justified in believing, or should consider ultimately good – do not seem to be ascriptions of natural properties to what they concern. If they are not, then they are not appraisable, or at least not fully so, using the methods of the sciences or even using those together with common-sense observations of the world. Contemporary philosophy is witnessing an age of naturalization projects: in ethics, philosophy of mind, and even philosophy of mathematics and logic – two domains one might think would pose no challenge to naturalism, since they have neither the disadvantage of being mysteriously supernatural nor – since they are essential for any scientific work – that of competing with any particular scientific approach. It is important, then, that we explore the possibility of naturalizing the normative. In doing this we cannot consider all the plausible attempts to achieve such naturalization, but it is possible to examine some major representative ones. A main focus will be on the possibility of naturalizing reasons. The paper will raise serious doubts about whether they can be naturalized; but it will also show how some of the aims of well-conceived naturalization projects can be achieved: normative properties are in some cases accessible to perception and, even apart from that possibility, may
An earlier version of this paper was given at Duke University and benefited from the discussion. For helpful comments I would particularly like to thank John Broome, Roger Crisp, Jonathan Dancy, Christian Miller, Susana Nuccetelli, Gary Seay, Walter Sinnott-Armstrong, and David Wong.
169
170
robe r t au di
be justifiedly ascribed to actions or other elements on the basis of clearly natural properties. There is no generally accepted definition of “naturalism,” but philosophical naturalists agree that the “descriptive” concepts and properties that must be countenanced to account for inquiry in the natural sciences should count as natural. What are often called “observation concepts,” paradigmatically those whose application is determinable through using the five senses, are clear and basic cases of the descriptive. With all this in mind, we might consider naturalism, in its philosophical as opposed to methodological form, to be roughly the view that nature is all there is and the only basic truths are truths of nature.1 Supernatural entities are thus ruled out, but only examination of specific versions of naturalism will tell us whether, for instance, mental phenomena must, on those views, be reduced to physical ones to count as natural. For our purposes, it will suffice to say that normative phenomena may be considered naturalized only if all normative claims can be explicated and appraised on the basis of the kinds of descriptive uses of language appropriate in natural science broadly conceived to include empirical psychology. 9.1 T h e c onc e p t of nor m at i v i t y What sorts of things can be normative? Many: reasons perhaps above all, but also properties, propositions, attitudes, including the propositional attitudes, both cognitive and conative; even language and certainly linguistic acts, such as evaluations. Normativity, then, is a property – that of being normative – which belongs to so many kinds of things that presenting a unified account of it is a major philosophical challenge. Paradigms of it are normative propositions to the effect that something ought to be done or that something has intrinsic, as opposed to instrumental, value. There are indeed many kinds of value: not just moral but aesthetic, intellectual, and spiritual value. Spiritual value includes but is not limited to religious phenomena. A spiritual experience may be deeply rewarding whether or not it is religious. The spiritual may occur in secular solitude,
1 This conception is developed and defended in Audi (2000). One persisting question is what counts as a natural science. In the text I include psychology because naturalization of the normative without dependence on psychological concepts and properties as key natural ones is apparently not possible; but strong versions of philosophical naturalism will take at least psychological properties to be at best epiphenomenal if they cannot be identified with non-psychological ones, such as neurophysiological properties.
Can normativity be naturalized?
171
even if paradigms of spiritually rewarding experiences are also in some sense religious. The concepts of justification and rationality are also paradigm of normative notions. Both are pluralistic, and both notions occur across all the categories and domains just noted. There is, for instance, justification (and rationality) of belief, of action, and of attitudes; and there is moral justification, prudential justification, and aesthetic justification. One thing that many philosophers are inclined to say of normative notions is that they are not descriptive. Descriptive properties of the most familiar sorts are observable, and, in more complex cases, the term is used for properties that, whether observable or not, are (a) causal, (b) of a kind appropriate to scientific investigation, and (c) capable of figuring directly in prediction and explanation of events. This claim is controversial, however, and I will rest no main contention on it.2 What is probably not controversial is that things possess normative properties in virtue of possessing other properties – those constituting the base of the former properties – that seem descriptive in the sense sketched. This leaves open whether normative properties are themselves ultimately descriptive. But the consequentiality (“in virtue of”) relation is not only common ground between non-naturalists and (cognitivistic) naturalists; it is also a basis for anchoring moral judgments in intersubjectively accessible “facts.” Whether the relation is even a priori, as rationalists commonly take it to be in normative cases, or empirical, as empiricists consider it to be in all cases, is a matter of continuing disagreement. It is essential here to distinguish normativity in content from normativity in upshot. Pain exhibits the latter but not the former; and as this suggests, though traits like cruelty have been called morally thick, they are, arguably, also moral in upshot but not content: they are like pain in that no exercise of moral or even normative judgment need be made to explain what they are or even to know the a priori criteria for their possession. To know that someone is cruel, we need to know only the “descriptive” fact that the person characteristically aims at causing pain in others for its own sake (hence non-instrumentally); and while knowing this gives us a reason for a normative judgment about the person, we can know someone is cruel without depending on a normative judgment.3 It is different with 2 Moore (1903) is plausibly thought to have held this. It might be denied by, e.g., Boyd (1988), Brink (1989), and Sturgeon (1985), “Cornell realists” who, even if they might grant that normative concepts are not natural, hold that normative properties are. 3 Granted, ‘cruel’ is vague owing to the indefiniteness of the degree to which the relevant desire is present. But this kind of vagueness also characterizes many non-normative terms; we need not in
172
robe r t au di
courage, which is normative (though not moral). To explain what courage is we must appeal to a proper balance between, say, steadfastness and avoidance, and to know that someone has courage we need to know that the person achieves a certain proper balance between undertaking and avoiding certain challenges. 9.2 T h e a ppe a l of n at u r a l i z i ng nor m at i v i t y One reason for the philosophical attraction of naturalizing normativity is what we might call the epistemological authority of perception. Perception is not just psychologically authoritative, in the sense that we cannot help forming beliefs given certain perceptions – and indeed cannot help trusting those beliefs unless there is strong reason for distrust. It is also normatively authoritative: a basic and pervasive kind of evidence consists in sensory impressions that are at the core of perception. For philosophical naturalists, perception is often taken to have epistemological sovereignty, in the sense that it is essential for any other kind of knowledge and is the ultimate test of claims to knowledge or justified belief. To be sure, testimony, a social source of knowledge, is also extremely important; but, depending as it does on perception for receiving it, it is not a basic source. Receiving it depends on the operation of another source of knowledge. We should also recognize the ontological authority of perception. We cannot perceive what does not exist, and sense experience provides a good ground for positing the existence of its objects. In my view, this ground is normative – evidencing (providing good reason for accepting) the existence of the object in question – as well as psychological, in tending to produce recognition of the object or at least some belief about it. In general, apart from very special circumstances, we cannot help taking to be real what we apparently perceive. The reality of a perceived object, to be sure, does not obviously entail its naturalistic status, but reflection on the kinds of objects knowable simply by the deliverances of the five senses makes clear that this status is likely to be taken as a natural implication. For understanding naturalism, the most important point about perception is that it is causal, in roughly the sense that perceiving something requires its causally impacting the perceiver. This causal character general decide, e.g., how much nasty behavior is permissible in order to determine whether someone is cruel. It is true, however, that if someone were neutrally manipulated so as to want to cause others pain for its own sake, we would normally regard the desire as excusable and the cruelty as in a way “artificial.” Does this imply that the concept of cruelty “contains” the normative concept of inexcusability or a normative notion of naturalness? I do not see that it does.
Can normativity be naturalized?
173
of perception is doubly significant. First, the most elementary paradigms of perception, or at any rate those most important for language learning, are physical: seeing properties of objects around one, feeling hard and soft things, tasting foods, and so forth. Second, it seems a constitutive element in naturalism to take what has causal power to be real and, for most versions, to presume that only the physical, and certainly just natural phenomena, are causally efficacious. A thoroughgoing naturalist may go further and hold that to be is to be the value of a causal variable. In addition to the doxastic compulsion of much sense experience, there may be (as some have thought) a dependence of thought on sense experience or imaginational experience derived from it: all ideas come from impressions, Hume famously maintained (in the Enquiry). Add to all these points a taste for simplicity, understood in part as a preference for a world sculptured with Ockham’s razor, and you have a physicalistic, even materialistic, worldview. Abstract entities such as numbers and universals may come under the razor even if eliminating the supernatural is, for most who wield it, the first priority. We seem to learn about abstracta through sense experience, and they appear to some philosophers to be explicable in terms of complicated relations to it. As Hume put it, “All this creative power of the mind amounts to no more than the faculty of compounding, transposing, augmenting, or diminishing the materials afforded us by the senses and experience” (1885 [1748]: 11). If, beyond all the points made so far, a theorist is influenced by a paradigm of knowledge of the kind produced by scientific inquiry – or technological success, which is often felt to validate science – there will be much motivation for naturalism as the view that nature is all there is and the only basic truths are truths of nature. These truths, of course, are appraisable using scientific method.4 9.3 N at u r a l i z at ion proj e c t s Naturalization projects occur in all major domains of inquiry: for instance, in ethics, epistemology, philosophy of mind, even theology. Each raises special problems, but there are commonalities, and here I consider only ethics in detail, though I will briefly bring my results to bear in some other domains. 4 I bypass here the point that this thesis itself does not seem to be a truth of nature; if naturalism is true, it is – or we might treat it as – higher order and try to square our account of philosophical justification for it with the thesis of naturalism.
174
robe r t au di
In ethics there is a longstanding division of theories of moral normativity into realism and anti-realism. Realism divides into reductive and nonreductive. Reduction can be of (among other things) concepts, properties, or facts. Here I leave aside the prospects for conceptual reduction, since this seems the least plausible of the naturalistic reduction efforts. My focus will be initially on attempts to give a reductive naturalistic account of normative properties. If this succeeds, however, then, on the plausible assumption that facts are identical when they express identical property ascriptions to the same entities, normative facts can be reduced as well. For reductivism regarding the ethical domain, we can in principle show something denied by non-naturalists and non-cognitivists alike: that moral terms express natural properties, in a sense implying that moral properties are reducible to (hence really are) natural ones. John Stuart Mill appears to have held a realist ethical naturalism. At least three versions of ethical naturalism should be distinguished. First, one might hold that moral properties are equivalent to certain natural ones. For instance, the property of being obligatory might be considered (necessarily) equivalent to (necessarily coextensive with) the property of producing at least as favorable a contribution to the proportion of pleasure to pain (in some specified population) as any alternative the agent has. Second, one might hold that moral properties are identical with certain natural ones. The same example will serve: this intermediate naturalistic reductivism would simply interpret the equivalence as representing an identity rather than just a necessary correspondence such that every instance of the first property is an instance of the other, and conversely. A stronger, identity claim is that to be a moral property is to be a causal-hedonic property of the kind just illustrated. Third, a very strong moral naturalism would take this identity claim to be conceptually true, in the sense entailing knowability on the basis of analysis of the constituent concepts. Mill apparently held not only that moral properties are equivalent to certain natural ones but also (for at least some normative terms) the strong naturalistic view. He said, for instance, “that to think of an object as desirable (unless for the sake of its consequences), and to think of it as pleasant [where pleasantness is a natural, psychological property], are one and the same thing” (1979 [1861]: 38). Mill did not make all the distinctions just presented; but this statement (in which “desirable” is used to mean roughly “intrinsically good”) at least illustrates one thesis of analytical naturalism even if, for deontic terms (e.g., “ought,” “obligatory,” and “wrong”), he held only the intermediate view.
Can normativity be naturalized?
175
This intermediate naturalism is held by some who reject strong naturalism (the kind that requires taking normative properties to be natural ones). On the theoretical reductivist view inspired by reductions in science (a view sometimes called Cornell realism), moral properties are like certain theoretical properties of a kind important in scientific theories. They are not observable, as are colors and shapes; but they are natural, causally efficacious, and understandable in a complex network of explanatory and descriptive relations to other natural properties (see in particular Boyd 1988; Brink 1989; Sturgeon 1985).5 Reductive realism is highly controversial, and a detailed assessment is not possible here.6 Such an assessment is especially difficult because we cannot simply presuppose any account of property-identity. Let us instead consider a non-reductive realism that gives a major role to natural properties but is not itself naturalistic. The apparently most plausible version of non-reductive realism is considered intuitionist, but this term can be misleading. Anti-realism is represented by non-cognitivism and, more recently, constructivism. Each of the positions sketched here deserves brief comment before I proceed to a positive view. 5 The literature exhibits terminological variability in characterizations of reductive naturalism. Like many other philosophers, I term a naturalistic view reductive if (and only if) it proposes an account of the target – say normativity or mental phenomena – on which the terms in question are at least strongly equivalent to certain naturalistic terms. The equivalence might be conceptual or, less stringent, ontic, as where property-identity is claimed for the analysans and analysandum. By contrast, Sturgeon (2006) calls a naturalistic position non-reductive provided it does not entail “the truth of some property-identity statement that has ‘goodness’ (or some other transparently ethical term) paired with some clearly nonethical term (such as, in Moore’s own examples, ‘pleasure’ or ‘what we desire to desire’) … But an ethical naturalist can deny … this implication” (98), since (to cite Sturgeon’s positive characterization), ethical naturalism is the view that moral value and obligation “are natural properties of the same general sort as properties investigated by the sciences, and … are to be investigated in the same general way” (92). Granted, this characterization does not entail the indicated claim. But without a substantive account of natural properties, Sturgeon’s view apparently leaves open that scientific investigation of, say, goodness, studies sufficient conditions for it, necessary conditions for it, and either no conditions that are (a) both necessary and sufficient or (b), even if (a), no property that is conceptually distinct from the ethical one under investigation but identical with it. Even a quite strong non-naturalist view of the kind indicated below can sustain this kind of cooperation between ethical and scientific inquiry. 6 Space does not permit appraising the “analytical descriptivism” proposed by Jackson (1998), which he distinguishes from “Cornell realism” (144–45). But the problems this paper poses (below) for instrumentalism as a naturalistic reduction program place apparently insuperable obstacles in the way of his view that “what confers value on a property ultimately comes down to facts about desires: value supervenes on the total story, actual, hypothetical, first- and higher-order, or something of this general kind, about desire … Thus to believe that A is right is to believe that A has the property it is rational to desire” (137). Even apart from the strong claim about the content of normative belief here – involving a single desirable property, for instance – the proposed account will work only if at least practical rationality (and presumably theoretical rationality as well) can be naturalized.
176
robe r t au di
In my judgment, analytical reduction of the kind Mill hoped for – naturalism the quick way, by conceptual reduction – almost certainly fails. Cornell realism is more plausible, but is not supported by an adequate theory of property-identity – also needed in philosophy of mind. It is surely not clear that ontic reduction succeeds. What of non-reductive realism? This is the consequentiality view – I prefer “consequentiality” to the more familiar term “supervenience” since it both entails strong supervenience and indicates something supervenience does not entail: the appropriate grounding relation to natural properties. This view is not fully naturalistic, and some philosophers do not approve of calling it a kind of naturalism at all, since it allows that the consequential properties are metaphysically non-natural, hence is compatible with an ontological dualism. Twentieth-century intuitionists such as Moore and Ross make clear that normative terms apply to what they characterize because it has certain non-normative properties (the base properties). This is anchoring normative properties in the natural world; it is not reducing them to causal properties. Non-reductive realism in ethics is not only realistic but, as their views illustrate, also not reductively naturalistic and, in my judgment, best understood on intuitionist lines of the kind Ross sketched (1930: ch. 2). One might think that if reduction of moral properties fails, moral naturalism fails with it. But what is not always realized is that anti-realism about normativity may sit quite comfortably with philosophical naturalism. Non-cognitivism is indeed partly motivated by the idea that we can account for normativity behaviorally – in terms of linguistic competences and accompanying motivation – and need not countenance it in the world that descriptive language is “about.” Thus, it is eliminativist about moral properties and attitudinal about normativity: normativity is not, as many ethical theorists hold, expressed by non-descriptive cognitive terms, but is carried by a kind of motivated attitude. One could perhaps call it agential: on this view, normativity is primarily a property of attitudes in virtue of which they are pro or con, for or against some phenomenon or kind of thing, such as brutal conduct. Normative judgments are then accounted for in terms of such psychological properties as motivating the appropriate kind of action. The view can even avoid subjectivism by trying to develop a (naturalistic) theory of good and bad reasons for moral attitudes. This non-cognitivist view, however, saves naturalism in ethics only by giving up not only moral properties but also moral truths.7 For a recent critical discussion of non-cognitivism, with careful attention to recently leading proponents, see Huemer (2005: ch 2). 7
Can normativity be naturalized?
177
Constructivists in ethics have a different approach. They are understandably unhappy with giving up cognitivity and with it, truth in ethics, but they agree with non-cognitivists that there are no moral facts or normative properties in the usual robust sense implying a place in our ontology. It is not clear that this project will succeed.8 A major challenge it faces is to provide a notion of construction for moral standards that is plausible enough to yield a viable ethical theory but does not collapse into taking moral properties to be natural properties, such as that of being endorsable, in one way or another, on the basis of a naturalistically describable procedure. With the consequentiality relation in mind, we should explore the constructivist route to naturalization, which has been developed in ethics but applies to the overall theory of practical reason. The idea here is to preserve the cognitivity of ethical discourse without presupposing realism. One might thereby improve on non-cognitivism while avoiding commitment to the reality of moral properties (at least if they do not turn out to be consequential in a way that gives them the kind of ontological status intuitionists can claim for them). In part, this is what at least some ethical constructivists attempt to do. Consider how John Rawls views Kant. He contrasts Kant with ethical intuitionists, for whom Rawls thinks “moral first principles and judgments, when correct, are true statements about an independent order of moral values” (Rawls 1993: 91); for Kant, “the order of moral and political values must be made, or itself constituted, by the principles and conceptions of practical reason” (99). This view goes with Rawls’ claim that “Kant is the historical source of the idea that reason, both theoretical and practical, is self-originating and self-authenticating” (100). Rawls does not explain how anything can originate or even authenticate itself. Both ideas are expressed metaphorically and, as Rawls may have realized, neither idea is clear. (Whether Kant himself is committed to relying on either one cannot be pursued here.) For our purposes, it may suffice to distinguish three things that, consistently with what Rawls says in the passages above, might be meant by “moral constructivism.” A genetic construction (in ethics) is a procedure for arriving at (roughly, formulating and accepting) moral principles or moral judgments, as where one must make (and rationalize) a decision on what to do. Genetic constructivism, then, would be the associated view that moral principles and judgments are arrived at by (or may evolve from) some procedure, such as applying the categorical imperative. They are justified when a sound
For problems confronting a wide variety of constructivist positions, see Bagnoli (2011).
8
178
robe r t au di
procedure is properly followed. This view does not automatically impose constraints on the content or status of moral principles; it concerns only how we arrive at them, not how we might know them or whether they presuppose moral realism. By contrast, an epistemic construction is a procedure, potentially but not necessarily combinable with a method of discovery or formulation prescribed by genetic constructivism, for coming to know or acquire justification for moral judgments or principles. Epistemological constructivism would be the view that such knowledge or justification – where “justification is addressed to others” (100) – must be acquired, or at least validatable, in the prescribed way. Again, the categorical imperative might provide the core of a framework for constructing a justification of a moral judgment, but here the claim is that justification is constructed. How the relevant claim is in fact arrived at is left at least largely open. To be justified in holding a moral principle might be to be able to establish it using some procedure of rational universalization that accords with a sound interpretation of the imperative. Third, a metaphysical construction is a way of creating, or at least retrospectively building, a grounding basis for one or more sound moral judgments or principles, as where, in accord with practical reason, one establishes valid moral principles or just laws, which in turn create obligations. Rawls’ terminology of self-authentication suggests that he takes Kant to be a metaphysical constructivist, one who regards the moral soundness of a judgment as created by, or in any case grounded in, its emergence (or at least derivability) from a proper construction, such as a correct application of the categorical imperative from which the judgment emerges. That lying is wrong, for instance, is not true because, owing to “an independent order of moral values,” the property of being a lie grounds the consequential property of wrongness; it is true because (for one thing) we can construct a valid veracity standard from the categorical imperative and cannot rationally endorse a constructive procedure that allows us to universalize principles that generally permit lying. How one might succeed in either epistemic or metaphysical constructions without some “independent” standards to guide one, such as the idea that human beings have value and hence may not be treated merely as means, is puzzling. It is not clear, moreover, that Kantian ethics is committed to affirming this possibility. In the Groundwork Kant posits “something whose existence has in itself absolute value,” such that “in it alone, would there be the ground of a possible categorical imperative” (1948 [1785]: section 428, p. 95). The reference to a ground suggests an
Can normativity be naturalized?
179
antecedent basis for the soundness of the imperative, indeed a basis whose reality is as robust as that of humanity itself.9 This suggestion is combinable with the view that hedonic value and promissory obligation may be as plainly real as the experiences of pain or pleasure, and the interpersonal relations, on which they are grounded. The important point here is that despite the terminology of construction, none of the constructivist positions just described is necessarily antirealist. Even metaphysical constructivism presupposes using some raw materials and can be taken to describe a method for discovering a principle true to the moral facts or at least the axiological facts, for instance the fact that rational beings have a kind of absolute value. Obligatoriness, for example, would be conceived as the property of being endorsable on the basis of a certain kind of constructive procedure: to be obligated to keep a promise would be to have the property of being such that, if we correctly viewed the facts of the agent’s case in relation to the categorical imperative, we would frame a principle of action requiring promise-keeping in the circumstances one is in. The obligation is, to be sure, a highly complex property and may be viewed as a construct from sound reasoning processes; but (a) the property of undergoing such a process and (b) the consequential moral property an act may have of being required by the conclusion of that process are real properties. Given that they essentially involve our reasoning, they need not be conceived as part of an “independent order.” But other theorists might note that in this sense of “independent,” our pleasures and pains are not part of such an order either: they are intrapersonal and depend on our experiences. If there were no experiences, there would be no hedonic value. Similarly, if there were no interpersonal relations, there would be no promissory obligations. My tentative conclusion here is that constructivism fails to steer a stable course between realism and non-cognitivism. In so far as it is plausible, it seems realistic and – if it depends on such posits as the value of persons – indeed non-reductive regarding the normative. I would speculate that it represents an attempt to naturalize Kant. In doing so, it takes him too close to Hume – a reconstructed Hume, to be sure, but still a constrained instrumentalist. On one interpretation of how such constructivistic naturalization might go, we reconstruct the desire-based theory of practical normativity by adding reflective endorsement, as Brandt (1979) – proposing a 9 Reason to take Kant to be committed to such a basis is noted in ch. 3 of Audi 1989 and also provided by Shafer-Landau (2003) (esp. p. 42) and Wood (2008) (e.g., p. 283, n. 3). The passage appears to ascribe reality to a value property.
180
robe r t au di
kind of reflective endorsement procedure – had added survival of cognitive psychotherapy as a test for rational attitudes. If, however, we hold firmly to the implicit naturalism, we fail to account for moral judgment. Desires and reflective endorsements, unconstrained by content restrictions, need not support plausible moral judgments. If, on the other hand, we build plausible normative constraints into the constructive procedure, then the view seems to take us toward an understanding of Kant that is arguably a nonreductive realism. This is a good place to explore some of the resources of a reconstructed Humean view as a naturalization project. That exploration is particularly needed if ethical constructivism, at least in some forms, is a procedurally constrained Humean instrumentalism. 9. 4 T h e n at u r a l i z at ion of pr ac t ic a l r e a s ons If there are any normative truths, then there are normative reasons. If, for instance, there is something we ought to do, there is reason to do it. If, moreover, anything has intrinsic value, there is (prima facie) reason to preserve or enhance or perhaps honor it (or to bring it about if it does not exist, as in the case of good experiences). But things are not brutely good, nor acts brutely obligatory. A thing is good as enjoyable; an act is obligatory as the saving of a child from drowning. Examples abound. These examples indicate that there exist normative reasons as considerations that ground obligations and values. Reasons are basic for normativity. Arguably, the notion of a normative reason – roughly a rationality-making reason, especially a good-making or an obligationconferring instance of such a reason – is the most basic normative notion. In the theory of practical reason, the most common and most plausible way to seek a naturalization of practical reasons is the Humean route. The basis of normative reasons is desire, specifically, non-instrumental desire. There are instrumental reasons, of course, but their authority depends on that of non-instrumental ones. I may have excellent reason to take a medicine as a means of relieving poison ivy, but if there were no reason to seek relief, then the medicine’s relieving properties would provide no reason for action. Reasons for action, then, are ultimately expressed by non-instrumental desires, in the wide sense in which desire – wanting, in a roughly equivalent terminology – is simply a motivational state of the kind that impels one to action toward its object. On the most straightforward desire-based theory, we get a maximizing account of reasons for action: an overriding reason for action is one that, relative to the agent’s alternatives, maximizes
Can normativity be naturalized?
181
desire-satisfaction. A reason for action, then, is a consideration that counts toward such maximality. This account of a reason for action is apparently naturalistic, at least on the plausible assumption that the notion of desiresatisfaction is naturalistic. Is the account of rational action successful? I think not. Consider a crucial point not always realized. Instrumentalism is foundationally content-neutral: nothing is intrinsically worth wanting. Reasons for action are naturalistically grounded in our psychology: in what we want (noninstrumentally). They are not grounded in our wanting the right sorts of things, such as things of “intrinsic value” – no such non-naturalistic posits are made. Some virtual universals of human nature provide one reason why the content-neutrality of instrumentalism is often missed or insufficiently appreciated. Our motivational starting points – the foundations of reasons for action – are often provided by our strong natural desires constituted by our (non-instrumental) aversion to the painful and the unpleasant and our (non-instrumental) attraction to the pleasurable. The relevant hedonic desires are, then, natural. Add to this that frustration of desires for pleasure is unpleasant. Thus, human nature saves instrumentalism from extensional absurdities of the most obvious kind: much of what we naturally desire is enjoyable and wanted for its pleasure-making properties; much of what we are naturally averse to – want to avoid or, if you like, “negatively desire” – is painful or unpleasant. These hedonic pursuits, however, are the kinds of things that quite different theories – including such non-naturalistic ones as Moore’s – can take to be supported by normative reasons. There are many variants of instrumentalism. Consider just the required instrumental beliefs that guide action. Some variants are subjectivist in appraising these, some objective. The simplest version is objective in the (limited) sense that it takes an action to be rational for an agent if and only if the action is truly believed to be at least as good as any alternative in achieving satisfaction of the agent’s intrinsic desires. There are various kinds of such optimal satisfaction of intrinsic desire. This may, for instance, be a matter of satisfying a single intrinsic desire whose strength is greater than the combined strength of any competing set of desires the agent has at the time, or of satisfying a set of intrinsic desires whose combined strength is greater than that of any set of such desires satisfiable by an alternative action open to the agent.10 10 Two points of clarification will help. First, I assume for the sake of argument that we may speak of combined desire strengths at least in a way that enables us to make comparative determinations of strength. Second, my focus is the rationality of an action for S, not S’s acting rationally;
182
robe r t au di
It is more plausible, however, to be less objectivistic and to conceive what is rational for us as relative to what we simply believe will optimize our intrinsic desire-satisfaction. This version of instrumentalism, like its objective cousin just stated, is also naturalistic. It construes the rationality of an action as simply a kind of fit between the action-type in question and the agent’s instrumental beliefs and basic desires. The position seems, however, both too permissive and excessively subjective. Suppose the beliefs in question are patently irrational and would be abandoned by the agent on a brief examination of their content. Here it is quite implausible to consider the action rational for the agent. The implausibility of this highly subjectivist view is confirmed by its failure to satisfy two reasonable constraints on a theory of rational action. First, such a theory should account for the sense in which a full-bloodedly rational action is a candidate for advisability, or at least minimal approvability, and, especially, for the inadvisability of certain non-rational actions, by someone who cares about the agent’s wellbeing.11 Second, it should enable us to explain why performing full-bloodedly rational actions counts toward – or at least never counts against – being a rational person. To illustrate, we would normally advise anyone we care about not to perform an action we take to be irrational. We also tend to advise against actions where we consider the person’s underlying belief(s) to be irrational (at least if, as is reasonable, we think this renders the action deficient in rationality).12 Moreover, if we take anyone to act on an irrational instrumental belief, we consider this prima facie evidence of some deficiency in the person’s rationality at the time (and the evidence seems, other things equal, to be stronger than that provided by simply having such an instrumental belief). Thus, the highly subjectivist notion of rational action just stated will not enable instrumentalism to meet the suggested constraints. As it happens, there is no standard instrumentalist resolution of the problem of what constraints to impose on the crucial instrumental beliefs. A plausible tentative formulation for a qualifiedly subjective I take the latter to be more complex, requiring that the action be based on the reason(s) in virtue of which it is rational. 11 For a naturalistic defense of a neo-Humean position on practical reason which takes advisability into account, see Hubin (2001). He says, e.g., that “there may be a fact of the matter about what action is rationally recommended for an agent … On my view, it is the agent’s values that play this role in determining what is rationally advisable for an agent. Other neo-Humeans focus on desires or preferences … at bottom, there is a fact – a brute fact – about the agent’s subjective, contingent, conative states … the property of advisability just is the property of being properly related to these brute facts” (466–67). 12 Only normally rather than always, because we might think that, e.g., by good luck, acting irrationally might bring splendid results.
Can normativity be naturalized?
183
instrumentalism might be this: an action is rational for you if and only if, given what you rationally believe on the basis of your total relevant evidence, the action optimizes your intrinsic desire-satisfaction. Alternatively, one could speak here of what you would rationally believe on appropriately considering that evidence.13 If, as I suspect, such ideas must be restricted further by a plausible instrumentalism, that should in any case not affect what follows. My main concern is to assess the basic instrumentalist – and naturalistic – idea that practical reason is an instrument of desire. If that idea is unsound, the question of rules for determining a good instrument – a producer of optimal action – is secondary. As against the simple objective formulation, the two just stated rely on the normative notion of rational belief. Instrumentalism so conceived can thus yield a fully naturalistic account of rational action only if the notion of rational belief admits of a naturalistic analysis. Reliabilist theories of justification in epistemology are perhaps the best candidates to fulfill that aim (assuming that success in naturalizing justification can be extended to rationality). I believe it is far from clear that they succeed; but here it is sufficient simply to point out that if I am right about the richness of the raw materials instrumentalism requires in order to provide a plausible account of rational action, then the project of instrumentalist naturalization of practical reason is not self-contained: its success is conditional on supplementation by a major epistemological theory.14 Let us assume for the sake of argument that epistemic normativity is naturalizable, and let us sophisticate instrumentalism by adding rational beliefs to its resources. The resulting Humean account still has untoward consequences. The manipulability of desire is one problem. Consider B. F. Skinner’s striking remark to the effect that we should seek “to design a world that will be liked not by people as they now are, but by those who live in it” (1971: 156). This view is unconstrained by any account of what people may be bought to like; it might be a drug-induced state of merely physical stimulations alternating with dreamless sleep. For instrumentalism, rejecting this disturbing view is at best difficult to justify. What instrumentalist premises preclude artificially developed desires from having as much 13 Cf. Fumerton (1990): S has most reason to X provided “The collective weight of S’s ends that might (relative to S’s evidence) be satisfied (frustrated) by X, when the value of each end is adjusted for the probability (relative to S’s evidence) of its occurring, is greater than the collective weight of … any of the alternatives” (101), where “an end for S … [is] something that S wants or values for its own sake” (94). 14 For a plausible attempt to present a detailed naturalistic epistemology see Goldman (1986). I have critically appraised the attempt to naturalize epistemology in, among other places, Audi 2002.
184
robe r t au di
normative authority as natural ones? Another problem is accounting for the grounds of moral standards. Even if we could frame necessary equivalences between moral practical reasons and desire-based ones, the latter would not provide the right kinds of grounds for moral conduct. Instrumentalist attempts to naturalize practical reason also face a different kind of problem. Let me provide some background for the problem and then indicate its bearing. Consider the notion of intrinsic desire – the kind usually viewed as equivalent to Hume’s foundational “passions.” Intrinsic desire is commonly contrasted with instrumental desire, and the latter is often taken to exhaust the category of extrinsic desire. It does not. Intrinsic desire (the kind Hume had in mind in speaking of passions), is desire for something for its own sake; this implies wanting it on account of something taken to be intrinsic to it, such as the refreshing qualities of water. But (as we shall see) a desire can be merely noninstrumental: neither intrinsic nor instrumental, as where a desire utterly isolated from one’s interests is induced by posthypnotic suggestion and one has no idea why one has it. If we keep in mind that there are two kinds of non-instrumental desire, we can see that at least for pure instrumentalism, it makes no difference whether a desire is intrinsic or merely non-instrumental. Suppose one forgets why one wanted something, such as to go to the supermarket, but still does this. The action is a response to the desire and may occur even when one has forgotten, but does not realize one has forgotten, why one is going there, yet also has no (intrinsic) desire to do it for its own sake. Since one has forgotten this, the action, although plainly desire-based, is not a candidate to be rational as a means to any further end. Still, that one desires something other than as a means (non-instrumentally) is all that matters. Such a desire – call it a residual desire – can have a clearly attainable object, and it defines a goal for action as clearly as does an intrinsic desire. We can act to realize a residual or other merely non-instrumental desire, get what we want in so acting, and do so efficiently or inefficiently in the same sense that applies to instrumental action in general. This apparent commitment of instrumentalism to the reason-giving power of merely non-instrumental desires is entirely consistent with the related psychological point that believing one does not know why one wants to go to the supermarket can eliminate the desire to, especially if one realizes that one originally wanted to do it only as a means. Normatively, the desire is, for instrumentalism, still not only not irrational but capable of providing a perfectly good reason for action, say to get out car keys. But surely it is not at the time rational, even if it does not rise to irrationality.
Can normativity be naturalized?
185
A consequence of conceding reason-giving power to a merely noninstrumental desire is this. Even though this desire is not intrinsic and (typically) would be easily given up upon realizing that one has no idea why one has it, wanting to do something purely as a means to realizing such a desire would provide a reason for this action derivatively from the “grounding” of this instrumental desire in the non-instrumental desire to which it is subordinate.15 Is there, however, any (normative) reason to perform the second action, say to ask someone to move a car so that one can drive to the market? Doing this might be reasonable on the basis of wanting to recall what one wanted (one might think it would jog memory), but not in order to fulfill the residual, merely non-instrumental desire to drive to the market.16 A major challenge for instrumentalism as a naturalization project, then, is to offer an adequate solution to the problem posed by the conative momentum these residual desires may have: why is it that merely noninstrumental desires cannot provide reasons for action? It is not clear that there is any resolution that is not either ad hoc or ultimately dependent on a different, and objectivistic, theory of practical reason. A naturalistic reduction project may draw on points made in this section by abandoning instrumentalism as an account of practical reasons and seeking a kind of objective account that yields a quite ecumenical reductivism. For anyone who agrees that hedonic considerations, promises, equality in distributions, and other descriptively identifiable states of affairs can provide reasons for action, it may be natural to claim that such normative notions as obligatoriness and intrinsic goodness are reducible to disjunctive properties. Being obligatory, for example, might be the property (of an act) of either having been promised or being a reducing of suffering or an avoidance of lying or … and so forth for all the types of basic normative grounds. But are there disjunctive properties, as opposed to disjunctive ascriptions of properties? And can we complete the list of 15 The reason is, however, entirely derivative. Otherwise, even where one wants A purely as a means to B, one could be credited with two reasons to A – that one wants to A and that one wants to B – which could outweigh an intrinsic desire to C stronger than the desire to A. At that rate one could prefer swimming to boating, but since the strength of one’s desire for the latter, together with the strength of one’s desire to rent a boat, as a means to it, could be greater than the strength of one’s desire to swim, one could have better reason to go boating. And this could be so even if one has forgotten why one wanted to go boating (say, to meet a friend across the lake) and would hardly know what to do with the boat having rented it. Here conative agitation unsettles me after all, even though I do not know its cause or how to achieve quiescence. 16 If this case of a merely non-instrumental desire is one in which an object of desire is not desired under any aspect of goodness, the case also suggests why the view that, normally, what is intrinsically wanted is wanted under the aspect of goodness is plausible.
186
robe r t au di
partially reducing properties satisfactorily? If we can, is obligatoriness such a disjunctive property – if indeed there are disjunctive properties? I doubt that an affirmative answer to either of the last two questions is plausible, especially if property-identity is a relation stronger than mere necessary equivalence.17 In any case, the burden of proof here surely lies with proponents of naturalistic reduction. 9.5 Mor a l pe rc e p t ion a n d e t h ic a l obj e c t i v i t y I have stressed the epistemological and ontological authority of perception. Once these are appreciated and seen to be a major incentive to naturalize normativity, it becomes important to consider whether moral perception is possible. If so, it can be a model for other kinds of normative perception (say aesthetic or even “evidential” perception, say seeing one proposition to support another). Moreover, any account of ethical objectivity – something ethical naturalists wish to preserve – should provide a major role for perception. Reductive naturalism does this in the most direct way. Non-reductive “naturalism,” and indeed traditional intuitionist ethics, do it indirectly by taking moral properties to be consequential on natural ones, where the latter – say violent acts – are the perceptible kind on the basis of which the former can be justifiably ascribed. Might we go further in accounting for objectivity and moral knowledge by countenancing perception that is itself moral but also has properties important for a naturalistic epistemology? Moral perception would have the causal element in sensory perception and a similar kind of normative authority but would make no commitment to supernatural authority. Let me suggest how this might be accomplished and how, without any commitment to naturalism, it may be taken to keep morality down to earth and free of supernatural dependence.18 Moral perception is not an exact analogue of physical perception, at least of seeing everyday visible objects in normal light. First, moral properties are not easily conceived as observable, in what seems (given our actual biological constitution) the most elementary way: no sensory phenomenal representation is possible for them. Intellective representations of them are possible, of course, and these may be accompanied by 17 That it is stronger, and that indeed a plausible account of it does not sustain identification of normative with natural properties, is argued in Paul Audi (“Property Identity and Non-Causal Determination,” unpublished MS). 18 The remainder of this section largely follows but also extends Audi 2010.
Can normativity be naturalized?
187
phenomenal elements.19 Second, even the perceptible properties on which the possession of moral properties is based may not be strictly speaking observable, at least in this elementary way. You can see A wrong B by, for example, witnessing A’s stealing B’s wallet. But arguably, the observable deed is just A’s removal of the wallet. Similarly, suppose you see a lie because you see and hear the speech act and you know the assertion is false. Even if you can see the basis of the falsity, such as the embers of the campfire falsely claimed to be extinguished, your seeing the wrong, as a distinctively moral perception, like your seeing the wrong in the theft of the wallet, depends on your having a kind of understanding of the significance of the discrepancy between this sight and the content of the assertion. Does one, then, in any sense see injustice or wrongdoing observationally? Or is such apparently moral perception seeing, in a way that is at least not narrowly observational, just a seeing of a set of “base properties” for wrongness – that is, of seeing properties on which wrongness is consequential? One might think that the phenomenal elements in perception properly so called must be sensory in the representational way that goes with paradigms of seeing and some of the other four ordinary senses. But might there be, for instance, a phenomenal sense of injustice that is not “pictorial” in the way the visual impression of Big Ben is? Must perceptual experience be cartographic – embodying a “mapping” from phenomenal properties, such as visualizing the evenly spaced numerals on a clock’s face, to properties of the object perceived? I doubt that this is necessary, though one could, as with visual experience of the clock, draw the wrongdoing if that means making a picture of the wrongful deed for, say, a jury. This is representation, but not of the same cartographic sort. But why is cartographic representation of a property necessary for perception of it? Consider olfactory and gustatory perception. We cannot map or draw smells or tastes; and even if smells and sounds enable us to identify their sources, they provide far less informational detail we get with sight or perhaps even touch. Thus, although, for certain cases of injustice, we may find a mapping, for the concrete deed seen to be unjust, from a perceptual sense of injustice to certain of the base properties of injustice, there is no phenomenal property that has a cartographic representational relation to the property of injustice itself.
19 Compare, however, Wynne’s view of “the perception of value.” See in particular ch. 3 of his 2005.
188
robe r t au di
Suppose, however, that we distinguish between a phenomenal – and especially, a cartographic – representation of injustice and a phenomenal representation constituted by a kind of perceptual response to it. The sense of injustice, then, as based on, and phenomenally integrated with, a suitable ordinary perception of the properties on which injustice is consequential, might serve as the experiential element in moral perception. Whether or not it is emotional, this sense is responsive to (and not merely caused by) the base properties and takes its character partly from them. A subtle shortchanging of an innumerate grocery purchaser produces a different kind of sense of injustice from the sight of a trumped up arrest of an obviously innocent protest marcher. In both cases, however, the injustice is seen by seeing the properties that ground it. Compare seeing an angry outburst that evokes the comment “I’ve never seen such anger!” Shall we say that the anger is not perceived because it is seen through what might be called constitutive manifestations of it? Granted, these can be mimicked by a good actor; but the façade of a clock may conceal an empty space behind. We should not conclude for that reason that clocks are never seen. Fallibility in sensory experience, then, does not impugn all perceptual experience, and some injustices seem as perceptible as anger. Granted, whereas someone perceptually normal cannot ordinarily view a tree in good light without the usual arboreal representational experience, many who are morally normal can view an injustice without the sense of injustice or any moral phenomenal response, such as a faint intuitive sense of injustice. But much as the light may be poor or the tree atypical, injustice can be subtle or camouflaged. Its initial imperceptibility might, then, trace to the need for greater moral sensitivity than goes with mere moral normality. Compare seeing a kitten that is sketched visibly but is not sharply distinct from background shrubbery. Someone who is perceptually normal but not an experienced viewer of paintings might not, without careful scrutiny or guidance, have any phenomenal response representing this cat. This does not imply that the cat is visually imperceptible, and the absence of a sense of injustice upon witnessing an unjust deed does not imply that its injustice is morally imperceptible.20 It should be apparent, then, that notion of perception as I construe it can accommodate moral perception by incorporating a distinction between perceptual representations of an ordinary sensory kind and 20 The view proposed is consistent with my ethical intuitionism developed in 2004. I am not here taking moral perceptions to be intuitions, but they can make singular moral propositions intuitive and can produce intuitions with the same content.
Can normativity be naturalized?
189
perceptual representations that integrate the sensory kind with a phenomenal response having a certain moral character, for instance, indignation, disapproval, and, on the positive side, a sense of justice done or of obligation fulfilled. Can this broad view of perception, however, do justice to the causal character of perception? It can. It does not do so by treating (moral) perceptual property instances like seeing injustice as causally produced or sustained by instances of moral properties. The theory is neutral regarding the possibility that moral properties are causal and, as that suggests, also neutral regarding the possibility of their turning out to be natural properties. It does, however, construe seeing a subset of base properties for injustice as – at least given appropriate understanding of their connection to moral properties – a kind of perception of a moral property; and this kind includes such ordinary perceptions as seeing a seizure of an old man’s wallet and hearing an abusive vulgarity screamed at a conference speaker. Given our psychological constitution, we may normally be unable to witness these things without a phenomenal sense of wrongdoing integrated with our perceptual representation of the wrong-making facts.21 Moreover, given how we see certain base properties sufficient for injustice, we are sometimes perceptually warranted in believing that, for instance, one person is doing an injustice to another (and in seeing the deed as an injustice). When we are thus perceptually warranted in believing this, we are often properly describable as seeing that the first is doing an injustice to the second and, indeed, as knowing this. This point does not imply that seeing an injustice is intrinsically conceptual, any more than a prelingual child’s seeing a clock is; but seeing that an injustice is done is conceptual. A child who has not yet acquired the concept of injustice can see an act that constitutes an injustice but have no sense of it as one, much less a concept of such an act. The same perception years later might immediately yield a moral conceptualization of the act or indeed moral knowledge in the child to the effect that there was an injustice. One might think that accounting for moral perception requires naturalizing moral properties so that they can figure in the causal order in the same way as the observable properties familiar in the natural sciences. I am not seeking to naturalize moral properties, nor is such naturalistic assimilationism required for explaining the data.22 At least two points 21 For related work developing a partial phenomenology of moral perception see Horgan and Timmons (2008). They also explore phenomenological aspects of fittingness. 22 For critical discussion of Sturgeon’s “Cornell realist” attempt to naturalize moral properties, see Audi (1993). That paper argues in a somewhat parallel way that naturalizing moral explanations is possible without naturalizing moral properties.
190
robe r t au di
are crucial here. First, the experiential responses to moral properties that entitle us to speak of moral perception are causally explainable in terms of a basis in the natural properties on which moral properties are consequential. Second, this is possible whether or not moral properties are themselves causal or otherwise natural. Might this account be said to yield a non-reductive naturalizing of moral perception? Perhaps so, if one bears in mind that non-reductive naturalism is not the goal of philosophical naturalism. Granted, all the constituents in the account except moral properties are natural properties and relations, and even the former properties are not presupposed to be non-natural. But moral properties themselves are not naturalized. Still, I not only take moral perception to be a causal relation but also grant that the base properties for moral properties are natural and have causal power if any properties do. The non-causal element, both in moral perception and in knowledge acquired through moral perception, belongs to conceptual capacities that go with an adequate understanding of moral concepts and with the a priori character of the relation between moral properties and the non-moral, natural ones on which they are consequential. The experience in virtue of which a moral perception counts as a perception is causally grounded in perception of natural properties, and it may be considered a causally grounded response to a moral property even if the phenomenal element in that response is not narrowly representational.23 My conclusion at this point, then, is that although moral properties are apparently not natural properties, they are constitutively anchored in natural properties and in an intimate way such that seeing or otherwise perceiving the natural properties or relations that are their base suffices, given an appropriate phenomenal response, to warrant describing certain experiences as perceptions of such moral properties as injustice or wrongdoing. When normative perceptions like these occur, the perceiver is in a position to see that something, such as an action or person, has the property in question. Such propositional perception embodies a kind of moral knowledge. Its causal element, together with consequentiality, assures both realism and the possibility of perceptual objectivity. The account also has the advantage of combining epistemologically internalist and externalist elements. To be sure, it says nothing about motivation. It is neutral on the issue of whether holding a moral judgment entails 23 Space does not permit comparing this view with moral sense theories, but I take those to be best understood as naturalizing moral properties and making them response-dependent; I do neither. For a version of this view usefully contrasting with mine see Smith (2004a).
Can normativity be naturalized?
191
motivation to act on it. It does, however, allow for both ethically relevant desires and moral emotions to play a motivational role. 9.6 Pro spe c t s f or a g e n e r a l ac c ou n t of nor m at i v i t y Recall that normativity occurs in many realms, for instance the epistemic, the prudential, the aesthetic, the religious, and of course the moral. Is it still possible to construct a unified account of normativity? If not, then assessing the prospects for its naturalization is even more difficult than we have so far seen it to be. In so far as normativity can be accounted for by a theory of theoretical and practical reasons, a unified account of it does seem possible. What I propose (and have developed in Audi 2001) is an experientialist account on which the qualities of experience constitute the ultimate grounds of theoretical and practical reasons. Our reasons for action or belief rest on internally accessible grounds of a kind that are common in human experience, say sense experiences or memory impressions. As in the case of practical reasons, one might, on the model considered above for obligatoriness, take the structure of theoretical reasons – those that are prima facie justificatory – to enable us to construct a disjunctive analysis of, say, being justified in believing. But consider justification for believing a proposition p. Is such justification the property of (say) either sensing that p or intuiting that p or apparently remembering that p or believing p on the basis of inference to the best explanation? I see no better prospect of reduction of the normative here than in the theory of practical reason. A more positive approach is along the following lines. Consider theoretical grounds, such as perceptual experience – these need not presuppose a theory but are theoretical in belonging to the realm of theoretical rather than practical reason. Theoretical grounds, for instance perceptual, memorial, or intuitive grounds, confer believability (roughly, credibility) on certain propositions or predications; parallel grounds, such as the pain of immersion in icy water, confer desirability on certain states of affairs. Such points suggest that the good and the true are the broad standards to which normative reasons are, as it were, responsible. There is much truth in the analogy: valuing is to the valuable as believing is to the true. The suggested (partial) account may seem to represent normativity as in some way passive. It does not. Normative properties figure in normative practices such as guiding and criticizing action. These are a kind of motivated behavior and often based on desires quite natural to us, such as
192
robe r t au di
the desire to educate our children. Such normative practices are pluralistic in their governing principles; they reflect the many kinds of value and obligation we have considered. Does such a realist view lose the motivational element in self-addressed normative judgments (e.g., that one should answer objections to one’s view), an element heavily stressed by Humeans? It does not. It accounts for motivation in rational, integrated persons, which is the crucial thing. Such persons tend to be motivated by their normative judgments. Noncognitivists, following Hume, may hold that, contrary to what moral realism apparently implies, self-addressed moral judgments are intrinsically motivating. But this has not been shown and seems mistaken.24 Once we restrict attention, as we must here, to rational agents, motivation comes back in. The reality of moral and other normative properties does not prevent our responses to these properties – and to the “external” reasons they ground – perceptual or attitudinal, from motivating us to act in the ways those reasons support. Note too the motivational potentialities of representative experiential grounds. Painful experiences powerfully motivate action to reduce or avoid the pain. The prospect of pleasure is characteristically a pervasive motive in human life. And there may be an almost as natural aversion to seeing such gross immorality as violence to defenseless innocent people. There may be an evolutionary explanation for the prominence of such motives, but that is not of concern here; the point here is that the realist, experientialist account of normativity focuses not on otherworldly elements or elements whose power to motivate is in doubt, but on common experiences that play prominent roles in both our explanatory and our critical discourse. We have seen incentives to naturalize the normative, and I have sketched some leading attempts to do so. These attempts apparently fail, but they do help us to see some of the reasons why a good normative theory should have what I call a naturalistic anchor. Normative properties are consequential on properties that it is plausible to call natural, and this relation provides the appropriate anchoring. These base properties are not identical with the normative ones grounded in them, but that leaves open that the former may yet be a kind of natural property. I see no good reason to conclude this, though I cannot claim to have established its falsity. We may, however, account for the objectivity of ethics and meet 24 In Audi (1997) I have argued in detail for a weaker view of the relation between moral judgment and motivation and connected both notions with weakness of will.
Can normativity be naturalized?
193
other challenges, such as the motivational challenge non-cognitivists pose for realists, without reduction or elimination of normative properties.25 On my view, a sufficiently ramified experientialist theory of reasons, both theoretical and practical, can provide the needed resources. It enables us to see normative truths on the basis of grasping descriptively accessible facts. It makes possible rational disagreement in normative matters but, in many cases, also provides a way to adjudicate such disagreements. It even accounts for the possibility of moral perception. That possibility may justify speaking of partially naturalizing one kind of moral knowledge. But it does not imply that there are no irreducibly normative properties, properties that do not simply describe the world as scientific theorizing may enable us to do. Nature provides facts that ground normative knowledge, but it may well not be all there is. 25 I have left open, to be sure, that the concept of knowledge and related external notions can be naturalized, but if knowing does not entail being justified, this should not seem beyond possibility even if justification is not naturalizable.
c h a p t e r 10
Ethical non-naturalism and experimental philosophy Robert Shaver
One thing – perhaps the main thing – that the classic non-naturalists wanted to protect was a specific epistemology.1 Say I am tempted to think that X is right because X maximizes pleasure and whatever maximizes pleasure is right. I need to reflect on cases such as these: (1) Say that X maximizes whatever is good, but involves breaking a promise. If I keep the promise, only slightly less total good will result. Should I perform X? (2) Say that worlds A and B have equal amounts of pleasure, but A has more knowledge, or more virtue, or a more just distribution of the pleasure. Do A and B have the same amount of value, or is there more value in A? Ross, who gives these examples, hopes that in (1) one thinks one should keep the promise, and in (2) one thinks there is more value in A. If one agrees with Ross, one does not think that whatever maximizes pleasure is right; the right is not simply a function of the good, and there are goods in addition to pleasure (Ross 1930: 34–35, 134–35, 138–40). The correct way to decide whether X is right is to elicit the intuitions of “the best people,” “thoughtful and well-educated people,” about thought experiments such as these, rather than for anyone to perform, exclusively, purely empirical work (Ross 1930: 41). The non-naturalists took naturalists to deny this. My concern is what impact, if any, recent work stemming from the psychology and experimental philosophy literatures should have on this debate. Some of this work, especially from epistemology, is hostile to appeals to intuition, and there are parallel empirical findings in ethics that could justify similar hostility to the appeals to intuition prized by the non-naturalists. I am grateful to Darcie Fehler, Tom Hurka, Tara Murray, Jen Robinson, and an audience at Manitoba, for discussions, and to Ben Caplan, Jason Christie, Joyce Jenkins, Susana Nuccetelli, Brendan Ritchie, Gary Seay, and Jonathan Weinberg for comments on earlier drafts. 1 For support, see Shaver (2007). By the classic non-naturalists, I intend Sidgwick, Rashdall, Moore, Prichard, Carritt, Ross, Ewing, and, in most moods, Broad.
194
Ethical non-naturalism and experimental philosophy
195
In the first section, I argue against a general empirical objection to appealing to intuitions. It does not follow, however, that empirical findings are irrelevant. In the second section, I argue that they may be relevant to particular arguments some non-naturalists give – in particular, to the argument of Prichard, Carritt, and Ross that egoism, Kantianism, and consequentialism give the wrong reasons for some moral judgments. One preliminary: one might object that discovering intuitions is itself an empirical business (Prinz 2008a: 191–96). The objection is reinforced by noting how often many of the non-naturalists – especially Prichard, Carritt, and Ross – take themselves to be uncovering “what we really think” (e.g., Ross 1930: 18–19, 23, 38–39; Ross 1939: 88, 91, 97–98, 100–102, 104–107), our “natural moral consciousness,” “ordinary consciousness,” “our reflective moral sense” (Ross 1939, 90, 95, 109). But this is a merely terminological point. What the non-naturalists wanted to preserve was the primacy of appeals to intuition. If a naturalist agrees with this primacy, but is concerned to label the process “empirical” or “a posteriori,” the non-naturalist would not mind. In any case, given the ridicule heaped by the naturalistically inclined on “the armchair,” I suspect that most naturalists are more aggressive than this (e.g., Prinz 2008a). In much of the experimental philosophy literature – usually focused on conceptual analysis, and in areas other than ethics – subjects respond to thought experiments in unsettling ways. Their responses are often influenced by the order and wording of the questions, and they often do not respond as philosophers expect. This has led some to doubt the worth of appeals to intuitions (e.g., Machery et al. 2004; Swain et al. 2008; Weinberg et al. 2001). In ethics, conceptual analysis is not usually the issue. Sidgwick, Broad, and Ross are clear that the relevant claims are synthetic a priori (Broad 1971: 122–23, 189; Ross 1939: 320; 1954: 42; Sidgwick 1907: 374–79).2 But the empirical literature seems to support the same negative conclusion about intuitions.3 For example: (1) Petrinovich and O’Neill found that when the trolley problem was described in terms of switching or not switching “resul[ting] in the death” of the one or five, subjects were significantly more likely 2 Some critics do not see, or perhaps take seriously, the possibility of the synthetic a priori. Thus Prinz writes that if “happiness has intrinsic value” is “not a conceptual claim” or “conceptual truth,” “it is presumably an empirical claim” (Prinz 2008a: 195). 3 For philosophers’ worries about framing in ethics, see Doris and Stich (2005: 138–41); SinnottArmstrong (2006b: 352–56; 2008b).
196
robe r t s h av e r to find switching the trolley wrong than when the problem was described in terms of switching or not switching to “save” the one or the five. Subjects were also significantly more likely to find switching the trolley right when the trolley problem is given before, rather than after, a variant in which the trolley is stopped before hitting the five by pushing the one in front of it. These framing effects show that responses are influenced by irrelevant differences in wording and order.4
Subjects were not only prone to framing effects. They also made puzzling distinctions and were often poor at justifying their judgments. This again suggests that their intuitions are poor evidence. For example: (2) Greene et al. found that 31 per cent say it is permitted to push one man off a bridge to stop a trolley that would otherwise hit five, but 63 per cent say it is permitted to pull a lever that opens a trap door that deposits the one in front of the trolley to save five (Cushman et al. 2006; Greene 2008b: 112; Prinz 2007: 25). (3) Waldmann and Dieterich (2007) found that most permitted saving nine by throwing a bomb away from them, killing one, but most did not permit pushing the one onto the bomb to save the nine. One might reply that this marks the significant distinction between using and not using the one as a means. But, in a variant, there was no significant difference in the (fairly high) level of permission to switch a trolley onto a side track, saving ten and harming two, and switching a trolley onto a looping side track, on which the two are the means of stopping the trolley from returning to the main track and harming the ten. Further, in both cases people were significantly more willing to permit redirection than in a case in which a series of cars would push a bus containing the two in front of the trolley, saving the ten. Waldmann and Dieterich conclude that subjects are 4 For a different example of framing, see Bartels and Medin (2007). Presumably framing effects partly explain the disagreement between different studies about the rightness of switching the trolley in the standard sidetrack case – Hauser et al. (2006), Waldmann and Dieterich (2007) and a BBC poll (www.newsvote.bbc.co.uk/2/hi/uk_news/magazine/4954856.stm) find 75–90 per cent agreement that switching is right, whereas Petrinovich and O’Neill and Sinnott-Armstrong (2006b) find much weaker agreement, even in cases designed to elicit agreement. It is not clear, however, how the framing works between experiments. One problem, not due to framing, is that Petrinovich and O’Neill ask “What would you do?” rather than “What is the right thing to do?” (1996: 149). Some may think they could not bring themselves to do what they take to be the right thing; see Baron (1992). For some philosophers’ speculations about order effects, see Kamm (2008: 131–32); Norcross (2008: 66–68); Otsuka (2008: 109–10); Unger (1996: 88–94).
Ethical non-naturalism and experimental philosophy
197
sensitive to what is redirected, rather than to whether the victims are used as means.5 (4) Haidt found that 80 per cent of subjects thought consensual, enjoyable, secret sex, using contraception, between a brother and sister, was wrong, even when it had only good effects on their long-term relationship. When asked why, they gave justifications that failed, given the details – there would be deformed offspring, the siblings would be traumatized, others would be encouraged. The experimenters pointed out the failure of these justifications, and the subjects agreed, but only 17 per cent changed their verdict (Haidt 2001: 814; Haidt and Bjorklund 2008: 196–98; Haidt and Hersh 2001; Prinz 2007: 30). The subjects were “dumbfounded.” (5) Many say that redirecting the trolley onto a sidetrack, killing one to save five, is permitted, while pushing one off a bridge to stop the trolley from hitting five is not permitted. Hauser et al. found that 70 per cent of these subjects provided justifications that “failed to identify a factual difference between the two scenarios” (Hauser et al. 2006: 13, 14; also Hauser 2006: 128; Hauser et al. 2008: 131).6 (Hauser et al. 2006 were [intentionally] generous about what counted as a successful justification. Any factual difference that would produce the different verdicts counted as successful, e.g., the presence/absence of physical contact.) Some thought it permissible to redirect the trolley such that it saved the five by hitting a heavy object, with the trolley hitting one person standing in front of the object, but impermissible to redirect such that it saved the five by hitting the one directly. Of these subjects, twenty gave justifications that failed and three gave justifications that succeeded (Hauser et al. 2006: 15; also Hauser et al. 2008: 133).7 5 One might question whether this is an example of a puzzling distinction, since it is one that Thomson, at one time, took to be significant (though she added that “I do not find it clear why there should be an exemption for, and only for, making a burden which is descending onto five descend, instead, onto one” [Thomson 1986: 108]). But I take Kamm to have shown that the distinction is insignificant, and Thomson did go on to give a very different account; see Kamm (1996: 163–65) and Thomson (1990: ch. 7). 6 Actually, things are worse than this. Two hundred and sixty-seven of 597 gave no justification or added assumptions to the cases; the 70 per cent is of the 330 remaining. Similarly, in the case noted next in the text, forty-five of sixty-eight gave no justification or added assumptions (Hauser et al. 2006: 14, 15). 7 Note, however, that Hauser et al. asked for justifications only from those who gave different verdicts on the sidetrack/pushing cases or on the hitting the object/hitting the person cases (2006: 12). (The same goes for Cushman et al. 2006: 1083.) Those who gave the same verdicts for either pair could easily produce a sufficient justification. Since only 12 per cent thought it was permissible to push, this would be a small number for the sidetrack/pushing pair. But the percentage who agreed for the hitting the object/hitting the person pair would be much higher.
198
robe r t s h av e r
These results might suggest that, as Swain et al. write, “intuitions … are ill-suited to do the work philosophers ask of them” (2008: 140). But I do not think these results are a problem for the appeals to intuition employed by the non-naturalists. Ross took the intuitions of the “thoughtful,” “well-educated,” and “best” people to count. One might at least want the intuitions of the subjects in better conditions. The empirical literature does not impugn these, for several reasons.8 In some of the studies, steps were taken to prevent the subjects from comparing their responses and so achieving consistency. Waldmann and Dieterich gave subjects either scenarios in which the trolley is redirected or scenarios in which the (smaller number of) victims are redirected. The reason is that “pilot research showed that [giving both scenarios to the same subjects] would make the similarity of the dilemmas salient and lead to transfer of the moral evaluations between the agent- and patient-intervention versions, thus eliminating the predicted effect” (Waldmann and Dieterich 2007: 249). The result is that subjects who might on reflection on both cases see that what is redirected is irrelevant are prevented from this reflection. Hauser et al. initially gave the hitting-object-then-person and hitting-person cases to different groups, since if the same subjects received both they “were unlikely to judge both cases differently in a single test session” (only 5.8 per cent did when given both) (Hauser et al. 2006: 10, 14). They speculate that “viewing such superficially similar scenarios one after the other induced subjects to provide identical responses” (Hauser et al. 2006: 14). It is important to prevent the “possibility that as subjects read and think about their answers to prior dilemmas they may well change their strategies to guarantee consistency” (Hauser et al. 2008: 127). But again the result (of the initial survey) is that subjects were prevented from reflecting on their replies by comparing them with the relevantly different cases.9 Reflection matters: for example, 8 For a similar conclusion about cases of grasping concepts, see Ludwig (2007); Sosa (2007); and especially Cullen (2010); Kauppinen (2007); and N. Pinillos, N. Smith, G. Nair, P. Marchetto and C. Mun, “Philosophy’s New Challenge: Experiments and Intentional Action” (2011). In what follows, I try to specify, more than Kauppinen does, the respects in which the particular empirical studies take place in poorer epistemic conditions. 9 Those who, in the initial survey, thought hitting the object was permissible or hitting the person impermissible were contacted later and given the case they had not seen. Thirty-three per cent gave differing judgments about the cases. This requires one to give a very weak gloss to the conclusion that Hauser et al. draw, namely that “the principle of the double effect is operative across a wide range of sub-populations” (2006: 10). Of those given both cases, where the distinction double-effect relies on is clearest, at worst 5.8 per cent and at best 33 per cent see a difference. Even in the initial survey, 56 per cent thought hitting the person permissible (72 per cent thought hitting the object permissible) (Hauser et al. 2006: 8; for somewhat different numbers, see Greene (2008b, 112–13); Mikhail (2008: 88); a majority thought it permissible to do what double-effect forbids. (Oddly, Hauser et al. at one point report that hitting the person is “most likely not” permitted [2008: 129].)
Ethical non-naturalism and experimental philosophy
199
I have found no one who, when presented with both the pushing and trapdoor versions of the trolley problem, gives different verdicts.10 It is also common to think, in cases of reasoning, that subjects score higher in a within-subjects format.11 (In a within-subjects format, the same subject responds to all of the cases of interest [say, both pushing and trap-door cases]. In a between-subjects format, each subject responds to only one of the cases of interest.) For example, consider the sure-thing principle: if I prefer A to B, knowing that C, and I prefer A to B, knowing that not-C, then I should prefer A to B even when I do not know whether C obtains. Subjects violate this principle in a between-subjects format but not in a within-subjects format (Shafir 1998: 71). Similarly, when subjects were given one scenario or the other, they gave much higher compensation to a victim injured in a robbery while shopping at a store he first visited on that occasion (because his regular store was closed), than to a victim injured in the same way at his regular store. Ninety per cent, however, when given both scenarios, gave equal compensation (Kahneman 1994: 7). This suggests that there is a significant difference between the epistemic conditions of the studies and of typical philosophical thought experiments, in which subjects each receive multiple variations on some scenario. When this reply – distinguishing the desired epistemic conditions from those in the studies – is made in response to cases which test for concepts, it is sometimes dismissed quickly by noting that if the philosophers have different concepts, it is unclear why one should be interested in these (e.g., Machery et al. 2004: 9; Nichols 2008: 397; Weinberg et al. 2001: 452). Perhaps taking philosophy classes simply teaches a certain understanding 10 Jason Christie noted that subjects in Haidt’s incest case were permitted a great deal of reflection, with freedom to revise, yet still did badly. I consider a possible reply – they did not do badly – in the second section of the present chapter. 11 For references, see Stanovich and West (2008: 673). Shafir and Kahneman note the difference between daily events, best modelled by a between-subjects format, and the within-subject thought experiments of philosophers. See Shafir (1998) and Kahneman (1994) (and, briefly, Kamm 1998: 469). Kahneman writes that the “estimation of weights [explanatory factors uncovered by psychologists] requires judgments of multiple cases considered one at a time, either in a sequential format that makes explicit comparison difficult or in a between-subjects design. In contrast, rules about the relevance of a factor in judgment [reason-giving considerations that subjects endorse] are evaluated by comparing cases that differ only in that variable. The rules are not necessarily evoked by cases considered in isolation … The philosophical analysis of [a] moral problem typically involves a search for rules guided by a comparison of cases … This design is suitable for the study of moral intuitions about rules, but the weights that determine intuitions about cases seen in isolation are necessarily beyond the reach of this method of inquiry … [T]he intuitions evoked by carefully crafted thought experiments will not … reliably yield correct predictions of the responses to cases seen in between-subjects designs” (Kahneman 1994: 8, 15). The point here is that the reverse is also true.
200
robe r t s h av e r
of a concept or weeds out those with a different understanding (Weinberg et al. 2001: 438). Whether or not this is a good reply, it does not bear on the issue of whose intuitions should be taken as the evidence for what is right or wrong.12 It is also sometimes objected that there is no reason to think philosophers, or Ross’ “best” people, are different from the college students who are the usual subjects (e.g., Sinnott-Armstrong 2008: 101; Swain et al. 2008: 149–50). All parties agree that it would be nice to have empirical studies of philosophers here. But in their absence, I think there is reason to believe that philosophers are different.13 After all, we think we are teaching our students various skills that we have and they, in varying degrees, lack. One important skill is to be consistent and to judge cases differently only on the basis of relevant differences. Haidt, one of the most prominent critics of ordinary moral reasoning, concedes that philosophers are “one of the few groups that has been found to reason well” (Haidt 2001: 819; also Haidt and Bjorklund 2008: 188, 193, 196). Kuhn found that in her informal reasoning tests, philosophers showed “perfect performance,” radically unlike all other groups tested (Kuhn 1991: 258). Best of all, Pinillos et al. (2011) found that greater intelligence, awareness that one’s initial response may be mistaken, and seeing both of the relevant scenarios before answering each made subjects less likely to display the (puzzling to philosophers) Knobe effect, in which subjects’ judgments of whether someone acted “intentionally” depend on whether the action was harmful or helpful. Improving epistemic conditions made the judgments of the subjects more like those of philosophers. One might object that superior reasoning skills are irrelevant to intuitions about cases, if these intuitions are intended to be immediate and unreflective. But that is presumably not the sense of “intuition” that the experimental philosophers (as opposed to some psychologists) intend, since if they did, their findings would be irrelevant here. One might object that philosophers are untrustworthy because they are often committed to a theory that guides their intuitions (Doris and Stich 2005: 138–39). This is certainly a threat (though not one specific to intuitions or philosophy). But in many cases, it seems possible to judge
12 It also seems much less likely that non-philosophers have a different concept of “wrong” or “permitted” than that they have a different concept of “knowledge.” (This point also makes it unattractive to argue that, in the ethics case, non-philosophers can be dismissed as confused about the concepts involved.) 13 For an empirical account of the difference, in terms of reflectiveness, see Livengood et al. (2010).
Ethical non-naturalism and experimental philosophy
201
whether a given judgment should be discounted. Thus although utilitarians may think it right to push the one in front of the trolley, they do not use this as an argument for utilitarianism. Instead, they try to explain away the contrary intuition. And often philosophers admit intuitions that raise problems for their theories. Thus although deontologists are caused difficulties by the trolley problem – especially the loop version – few change their intuitions to fit their theory. Instead, even when it makes their life much harder, they try to accommodate the intuitions (see especially Kamm 2008). Finally, one might rightly object that while this may address puzzling distinctions and justifications, it does not address framing. But consider the (conflicted) literature on framing. Framing seems to go away when the cases are elaborated (Druckman 2001; Jou et al. 1996; Kühberger 1995).14 Group discussion makes framing go away “when the group members learn about other frames for viewing a problem” (Kühberger 1998: 44). (Paese et al. 1993 found that discussion sometimes made framing more extreme, but made it less if a reframing occurred.) Asking subjects for justifications and/or requiring them to think for longer periods of time reduced or eliminated framing (Miller and Fagley 1991; Sieck and Yates 1997; Takemura 1994). Against this, some found that asking for justifications made no difference to framing (Fagley and Miller 1987; LeBoeuf and Shafir 2003; Levin and Chapman 1990). Framing persisted in a within-subjects format (Frisch 1993; Kühberger 1998; LeBoeuf and Shafir 2003). Within-subjects, subjects were usually consistent between cases, but the wording of the first case encountered usually dictated their response to both cases (LeBoeuf and Shafir 2003; Levin et al. 1987; Loke 1989; Stanovich and West 1998).15 One worry about these studies of within-subjects format is that subjects were not permitted to change their answer to the first problem after seeing the second version of it. There was thus no test for whether seeing both 14 In the Asian Disease case, subjects differ in their ranking of policies according to whether, knowing 600 lives are concerned, they see “200 saved” or “400 die.” Elaborations given for “200 saved” were “200 saved and 400 die,” “200 saved and 400 not saved,” or an explanation of how saving the 200 dooms the 400. Frisch (1993) gave “200 saved” and “400 die” consecutively, and found that 31 per cent of those who earlier framed when presented with both cases (separated by other questions) said the cases were different. They are presumably importing information prevented by these elaborations. 15 LeBoeuf and Shafir close by suggesting that “[u]pon immediate presentation of a conflicting frame, those who think more carefully will be able to detect and avoid a potential contradiction,” and avoid a framing effect (LeBoeuf and Shafir 2003: 86, 89). It is unclear whether their suggestion is that subjects will be consistent (but still framed) or not framed at all.
202
robe r t s h av e r
frames resulted not only in consistency but also in an absence of framing for the initial cases seen. The within-subjects design was not permitted to have an effect on first-problem framing. Another worry is that Sieck and Yates found that framing was reduced only by producing a written justification before making a choice – merely knowing that one must provide a justification after making the choice had little effect – whereas LeBoeuf and Shafir (and others) asked subjects for a choice followed by a justification.16 Findings on individual differences also vary. Many found that experts did no better at avoiding framing (Loke and Lau 1992; Loke and Tan 1992; McNeil et al. 1988; Tversky and Kahneman 1986; for more references, see LeBoeuf and Shafir 2003: 88). But Smith and Levin (1996) found that those with a “tendency to engage in effortful thought” framed less or not at all. McElroy and Seta (2003) found that an “analytic” thinking style, one not relying on fast, unreflective heuristics, reduced or eliminated framing. (Shiloh et al. (2002) found that those who scored highly on both “rational” and “intuitive” thinking framed, while those who scored highly on one or the other style but not both did not frame.) Again on the contrary, LeBoeuf and Shafir (2003) found that the tendency to engage in effortful thought did not help, even when combined with asking for justifications, though it did increase consistency (see also Stanovich and West 1998) – but again, their format may have prevented thought from working. Now consider philosophy. Philosophers use an “analytic” thinking style and score highly in measures of the tendency to engage in effortful thought. When they consider thought experiments, the experiments are usually elaborated, often by considering small variations on them; justifications are sought; a great deal of time is spent and a great deal of discussion occurs; the cases are considered within-subjects, in many different wordings and orders, immediately after one another, repeatedly, with freedom to revise one’s verdicts. Think of discussion of the trolley, or Thomson’s burglar or Singer’s pond or Rachels’ bathtub examples. (If one asks what it is, in the realm of intuitions about thought experiments, that philosophers are expert at, it is that they can devise a host of relevant thought experiments and take the correct answer to depend on fitting 16 There are also other ways to address framing. Cullen made the framing noted by Swain et al. vanish by presenting the survey as “testing prospective questions” and prefacing each case with an instruction to consider each case independently (Cullen 2010). Hauser et al. claim that framing effects “have small effects on overall judgments as long as the wording is held constant across a set of different dilemmas (e.g., if a permissibility question is framed as ‘saving,’ then all contrasting dilemmas use ‘saving’ as well)” (Hauser et al. 2008: 143).
Ethical non-naturalism and experimental philosophy
203
these cases [see Livengood et al. 2010].) This is very different from the survey experience. It is possible that the initial order of presentation has lasting effects through this process, but that is not proven by the surveys. Here are two modest surveys. (1) I gave half of a class the standard trolley problem, discussed it, gave the pushing-off-the-bridge version, discussed it, then had them judge the permissibility of action in the two cases. Further discussion followed, with a chance to revise one’s judgment. I then did the same with the other half (absent earlier), with the order of cases reversed. The discussions were fairly lengthy – it took an hour and a half in total. In each case, a few changed their minds. If there was to be a framing effect, one would predict that seeing the standard version first would make one more favorable to pushing, and seeing pushing first would make one less favorable to switching. I found the opposite. Unfortunately, the numbers were small (twenty-five), and some had seen the cases in previous classes. It would be helpful to replicate the test in better conditions. (2) Jason Christie tested for order effects for standard/pushing cases, testing 373 first- and second-year philosophy students at Manitoba. Two hundred and thirty had not been taught the cases beforehand and answered each question as it came, without a chance to revise. One hundred and fifteen had been taught the cases beforehand, and were instructed to read all the cases and think for thirty seconds before answering any. (Twenty-eight failed a control question.) For the former group, those who saw the standard case first strongly favored redirection (1.91) and opposed pushing (3.1); those who saw pushing first mildly opposed redirection (2.83) and strongly opposed pushing (3.95). For the latter group, those who saw the standard case first mildly favored redirection (2.44) and opposed pushing (3.32); those who saw pushing first also mildly favored redirection (2.4) and opposed pushing (3.35). The combination of prior exposure, considering all cases before answering, and time eliminated one order effect. It does not follow that Ross is right to rely on the “best people,” or that only the intuitions of professional philosophers count. In my experience, with a little prodding, non-philosophers are quite good at making their positions consistent, making relevant distinctions, disowning judgments due to framing, etc. The experimental philosopher could get similar results by, say, conducting interviews, or designing surveys that, for example, made one’s pattern of responses clear to one, with freedom to
204
robe r t s h av e r
revise. But these are moves in the direction of replicating the appeals to intuition employed by the non-naturalists, and so far experimental philosophy has not shown that these appeals are ill-suited to their task.17 Nor, I think, are the results of experimental philosophy surprising. Many of the worries raised for appeals to intuition (such as framing) are familiar from decades of social psychology. They are not special problems for intuitions, but rather part of the general picture social psychology presents of belief formation.18 The real issue all along was under what conditions, if any, we are more reliable. There may, however, be more local conclusions to be drawn from the empirical literature. Here is one example. Prichard, Carritt, and Ross do not argue only that rival normative theories, whether utilitarian, Kantian, or egoist, give the wrong verdicts about the rightness of some actions. They also argue that rival normative theories sometimes give the wrong reasons for their verdicts. (Call these the wrongverdicts and wrong-reasons arguments.) Thus Prichard: “Suppose we ask ourselves whether our sense that we ought to pay our debts or to tell the truth arises from our recognition that in doing so we should be originating something good, e.g., material comfort in A or true belief in B, i.e. suppose we ask ourselves whether it is this aspect of the action which leads to our recognition that we ought to do it. We at once and without hesitation answer ‘No’” (Prichard 2002: 10; also 136). Similarly, “no one … really thinks that the fact that a certain action would be advantageous … to himself constitutes the action a duty on his part” (Prichard 2002: 123; also 26, 30, 122, 171, 188). “[W]henever in ordinary life we think of some particular action as a duty, we are not simply thinking of it as right, but also thinking of its rightness as constituted by the possession of some definite characteristic other than that of being advantageous to the agent … [For example] in thinking of our keeping our promise to X as a duty, we are thinking of the action as rendered a duty by its being the keeping of our promise” (Prichard 2002: 29). Against Kant he notes that “[n]o one could suppose that the reason why an act ought to be done consists in the fact that everyone could do it” (Prichard 2002: 59). Carritt asks of Joseph “Do we really think that the reason why we ought not to cheat this man of his hire is that, if there should ever be a society where all paid their debts, their lives would be better for it?” (Carritt 1937: 77). Ross agrees with Prichard that “[e]goistic hedonism is put out of court by the fact which stares us
For the same point for concepts, see Kauppinen (2007: 106, 107, 109); Ludwig (2007: 151). I owe this point to Brendan Ritchie.
17 18
Ethical non-naturalism and experimental philosophy
205
in the face, that it is consideration for the rights or interests of others, far more often (to state the matter very mildly) than consideration for our own interests or rights, that makes us think it our duty to behave in a certain way” (Ross 1939: 65; also 277). Utilitarian explanations of gratitude face the objection that “Is this the way in which anyone not intent on defending a theory really thinks about the return of benefits? Is it really the wish to keep alive our benefactor’s good-will, or to save him from becoming patronizing, that makes us think we ought to return benefits? Is not our actual thought more truly expressed in the simple phrase ‘one good turn deserves another’?” (Ross 1939: 101). And Ross’ final argument against ideal utilitarianism is that, even if the ideal utilitarian can generate Ross’ verdicts by including things such as promise-keeping as goods, this still gives the wrong reason: “while a good man will feel satisfaction at a second man’s fulfilling his promise to a third [and so the state of affairs is good], that satisfaction presupposes the thought that the promiser has, by making the promise, put himself under an obligation to the promisee … [T]he rightness of the act does not depend on the goodness of the result produced” (Ross 1939: 289; also 107–108).19 Separating wrong reasons and wrong verdicts might seem nitpicky. But the arguments work very differently. The wrong verdicts argument requires showing that the rival theory indeed produces the wrong verdicts. This requires careful study of the rival theory, and in some cases might not give Prichard, Carritt, and Ross what they want. For example, as Ross concedes, if promise-breaking is itself bad, the ideal utilitarian might be able to get the right verdicts. The wrong-reasons argument avoids all this labor. We know “at once and without hesitation” that the rival theory gives the wrong reasons, even without demonstrating this by showing that wrong verdicts emerge. There are two potential objections to the wrong-reasons argument when it is given without support from wrong verdicts. Both objections owe something to the empirical literature. Objection 1: Prichard, Carritt, and Ross usually consider single, uncontroversial cases. They suppose that by reflecting on, say, a case of promise-keeping, one sees that the reason it is (prima facie) right to keep the promise is just that one has promised. (This is also how doubts are to be 19 Contrast Sidgwick: “it is a common observation that the judgments of practical men as to what ought to be done in particular circumstances are often far sounder than the reasons they give for them” (Sidgwick 1998: 21). For a view of moral intuitions as often tracking relevant factors, unknown to those who have the intuitions, see Woodward and Allman (2007, esp. 190–91, 197–98).
206
robe r t s h av e r
resolved: “if we do doubt whether there is really an obligation to originate A in a situation B, the remedy lies not in any process of general thinking, but in getting face to face with a particular instance of the situation B, and then directly apprehending the obligation to originate A in that situation” [Prichard 2002: 20].) But one might think that the correct explanation of our judgments cannot be found by reflecting on a small number of judgments, each taken in isolation. Consider a standard teaching pattern. One asks whether it is wrong to cut up one innocent man to harvest organs to save five. The reply is that it is wrong because it is killing. One asks whether it is permitted to abandon saving one to save five. The reply is that this is permitted, because one is letting the one die rather than killing. One asks whether, in a standard trolley case, it is permitted to turn the trolley to hit one rather than let it hit five. The reply is that this is permitted, but the reason cannot be that this is letting die rather than killing. Perhaps the reason is that the one is not used as a means in the trolley case. One then asks whether it is permitted to turn the trolley onto a looping sidetrack on which the one will be hit and die, stopping the trolley from returning to the main track where five lie. The reply is that this is permitted, but the reason cannot be that the one is not used as a means … One lesson is that the justification one would endorse on reflection is not shown by looking at one case, but by numerous cases designed to see what matters. This is supported by the empirical literature. As noted, results differ between within-subjects and between-subjects formats. Where there is a clear right answer, subjects are usually more likely to reach it with a within-subjects format. This suggests that we are more likely to discover what we really think, at least in the reflective sense Prichard and Ross intend, by considering many related cases at once. Ross does examine more than one case in his initial argument against ideal utilitarianism (Ross 1930: 17–18). We think we ought to keep the promise just because we have promised. We do not think we should keep the promise if disastrous results would ensue. But the second case need not force us to think that in the first, on reflection, we think we ought to keep the promise because of the results. We can instead say that there is a prima facie duty to keep promises, just because we have made them, which is sometimes defeated by a prima facie duty to bring about good results. This is preferable to thinking that there is only a duty to bring about good results, since we are not indifferent between producing X amount of good by keeping a promise and producing X by breaking the promise. But note that here what does the decisive work is a claim about
Ethical non-naturalism and experimental philosophy
207
verdicts – that taking there to be only a duty to promote the good gives the wrong verdict in the X amount case. Without that case, Ross has a stand-off about the reasons. The wrong-reasons argument does not stand alone. Objection 2: Haidt’s subjects failed to justify their judgments about incest. Hauser et al.’s subjects often failed to produce justifications that explained their judgments. Sometimes they produced justifications that succeeded, but which they then disowned as irrelevant. Hauser sums up: “people are largely incoherent when it comes to explaining their judgments … [P]eople are clueless with respect to their justifications … Most people have no idea why they draw a distinction between these cases [standard trolley vs. pushing off bridge] … When we ask subjects for justifications [regarding taboo cases], they are hopeless” (Hauser 2006: 124–25, 128, 158). More generally, there is a large literature showing the unreliability of introspection (e.g., Nisbett and Wilson 1977; Wilson 2002: ch. 5). The literature concentrates on cases in which subjects incorrectly introspect the causes of their responses. Subjects typically overlook the cognitive process in them which is the cause, even when they know about the efficacious stimulus, and even when, after the experiment, the experimenter describes how the stimulus could cause the response. Subjects instead cite different cognitive processes (which are plausible, but not operative in the case at hand).20 The wrong-reasons argument relies on the plain man’s ability to report the thoughts leading him to think it is right to keep a promise. This is a report of a cognitive process, rather than a report of the result of one – exactly what the anti-introspection literature targets. Unless this introspection is more reliable than the introspections of causes of responses in the literature, there is little reason to believe what people say. They may well, as in the literature, simply cite plausible reasons for keeping promises, whether or not those are the cause of their judgment.21
20 For example, insomniac groups A and B were given placebos. A were told that it would produce the same symptoms as their insomnia. B were told that it would reduce their symptoms. As expected, A slept better than usual and B slept worse. The cause is presumably that A took their symptoms to be caused by the pill, rather than the troubles causing their insomnia, and B took their troubles to be so serious that the troubles overcame the relaxing effect of the pill. But when asked for explanations, subjects cited factors such as increases or decreases in stress from exams, relationships, etc. They denied thinking about the pills. When told of what seems the cause, they again denied its presence and even suspected that no one could have gone through that thought process (Nisbett and Wilson 1977: 238). 21 Haidt (2001: 822) also makes this suggestion, though does not apply it to Prichard, etc.
208
robe r t s h av e r
Ross’ final argument against ideal utilitarianism seems especially vulnerable. The goodness of the state of affairs in which a promise has been kept is not the reason why one ought to keep it, because we do not think of the goodness to justify the ought-claim. Rather, we use the ought-claim to justify the claim that the state of affairs is good. The latter attributes to the plain man a remarkably sophisticated view about the relation between ought- and goodness-claims.22 (Ross might be able to back up his ascription of this view by presenting “paradox of deontology” cases, in which, say, I think I ought not to break one promise to prevent two other promises from being broken. If the goodness of the state of affairs in which promises are kept explains duty, then my duty is to break, contrary to what some think. But this again invokes wrong verdicts to support wrong reasons.) The plain man given philosophical training would do better than the empirical literature indicates. Perhaps this is why Ross writes of the “best people,” “thoughtful and well-educated people.” But this is dangerous. When Prichard and Ross make their arguments concerning our reasons, they typically write of “plain” men, or “we” “in ordinary life.” Ross opens against the utilitarian by writing that “[w]hen a plain man fulfils a promise because he thinks he ought to do so, it seems clear that he does so with no thought of its total consequences, still less with any opinion that these are likely to be the best possible” (Ross 1930; 17; also 20–21n.; 1939: 186). This makes it more plausible that the reasons cited are (sometimes) not utilitarian, Kantian, or egoistic. But the argument is then less trustworthy. Thoughtful and well-educated people are exactly those who give these sorts of reasons. In reply to the worries about introspection, Prichard, Carritt, and Ross might recast the wrong-reasons argument. Rather than asking for the thoughts that led to the verdict that one ought to keep the promise, the argument could simply ask for a judgment about the plausibility of a certain sort of reason for the verdict. The plain man would not justify keeping a promise by noting, say, that doing so would profit him, or produce more good than any alternative action. He would instead justify keeping it simply by noting that he has made that promise. In effect, what the objection treats as a process – I think “I’ve made a promise” and from that alone reach the verdict that I ought to do what fulfills it – can 22 Ross, following Carritt, does give one piece of evidence for attributing this view. We do not think it is good that the promisee gets the item promised when this occurs without the agency of the promiser. This shows that the goodness depends on a duty’s being fulfilled (Ross 1939: 287–89; Carritt 1928: 72–73). But the ideal utilitarian can reply that what is taken to be good is not the promisee getting the item promised, but that that is brought about by the promiser.
Ethical non-naturalism and experimental philosophy
209
be treated as itself a verdict, and so becomes immune to the arguments against introspection. There is again danger in this recasting, however. The plain man is now reflecting on his verdict that the promise ought to be kept. Justifications that perhaps would not occur in the daily keeping of the promise are now on the table. It is not so clear (supposing it was clear in the first place) that egoist or utilitarian or Kantian justifications would be dismissed. The plain man is coming closer to the position of the philosopher, and so to the justifications philosophers offer. Prichard, Carritt, and Ross might, however, give a more successful reply, at least to the worry about satisfactory justifications. Cushman et al. found that subjects were quite successful at justifying their verdicts in some cases, namely those in which the paired cases were instances of acting versus omitting to act (e.g., pulling a lever to dump one off a bridge in front of a trolley versus refraining from pulling a lever that would prevent the one from falling off the bridge, in both cases with five saved). Baron also found that those in within-subjects designs were successful at justifying their verdicts in a range of cases designed to reveal non-consequentialist thinking (e.g., distinguishing between acting versus omitting, punishing in sub-optimal ways, compensating differentially on the ground that the injury was caused by humans or nearly avoided, resisting coerced improvements) (Baron 1994: 9). In Haidt’s incest case, one might think that the subjects found the incest intrinsically wrong. Perhaps they tried to give further reasons because they did not take merely saying “incest is wrong” to be what was wanted, or sufficiently explanatory.23 Or perhaps, as Haidt himself suggests, the problem was that subjects had competing justifications between which they could not decide. On the one hand, harmless, consensual, private acts are permitted; on the other hand, “unnatural,” untraditional acts are suspect.24 If so, the subjects did have access to their reasons. They lacked access to how they weighed the reasons, but this might not be such a failing, if, as Prichard and Ross for example think, there is nothing general to be said here. 23 Prinz suggests that subjects may think of incest as a “basic [dis]value,” though he sometimes runs this together with treating the verdict as an emotional reaction (Prinz 2007: 31–32; also Prinz 2008b: 160; 2008c: 428–29). The basicness suffices to explain the inability to justify, without mentioning emotion. Cushman et al. suggest the principle “‘it is wrong for a brother and sister to have intercourse’” as explaining the verdict (2006: 1087). 24 Haidt does not suggest this for incest, but rather for conservative reactions to homosexuality. Conservatives and liberals did not differ much for incest, so one might think the explanation for conservative dumbfounding about homosexuality applies to both groups for incest (Haidt and Hersh 2001: 214–15). Even liberals invoked how things are “supposed” to be in the incest case (Haidt and Hersh 2001: 206).
210
robe r t s h av e r
Moreover, Hauser’s claim that plain men are clueless seems an overgeneralization from trolley and taboo cases. Similarly, Prinz, following Haidt, infers from Haidt’s examples that the justifications people give are “often superficial and post hoc. If the reasons are successfully challenged, the moral judgment often remains … Arguments for that judgment are usually contrived after the judgment is made” (Prinz 2007: 29). What may be true about incest and other taboos need not show anything about what is often or usually the case for other moral judgments. Indeed, when Haidt gave Kohlberg’s Heinz example (should Heinz steal a drug to save his wife’s life?), along with various taboo examples, he found that for Heinz subjects “did seem to use some reasoning, and they were somewhat responsive to the counterarguments given by the experimenter” (Haidt and Bjorklund 2008: 198).25 The worry that Prichard, Carritt, and Ross should not rely on single, uncontroversial cases remains, but things are not as bleak as Hauser, Prinz, and Haidt claim. Empirical concerns might suggest other problems for non-naturalists. For example, the worry that people simply do not agree is an old one, but one which empirical work has given new life (e.g., Doris and Stich 2005: sections 4–5).26 I have argued only that concerns about framing and odd responses do not discredit their reliance on intuition, at least not without more empirical work. Empirical work may, however, discredit a particular sort of argument, the wrong-reasons argument.
25 Haidt also now restricts his account to “moral judgments,” by which he means judgments of others, and not “moral decision making,” by which he means deciding what I ought to do. For the latter, he does not argue for dumbfounding (Haidt and Bjorklund 2008: 242–44). 26 They would not mind Walter Sinnott-Armstrong’s argument against “intuitionists” that, since one needs to know that framing effects are absent to trust an intuition, one’s intuitive verdict is an inference, with this knowledge as a premiss, rather than something non-inferential (SinnottArmstrong 2008c: 70–72, 99, 104; also Sinnott-Armstrong 2006b). Sinnott-Armstrong takes this to be a problem because he presents non-inferential beliefs as a way of avoiding a sceptical regress argument. Making them inferential makes them unsuited to this task. The classic non-naturalists, however, typically show little interest in adopting non-inferential beliefs as a way of defeating scepticism. Indeed, they show little interest in scepticism. For example, neither Sidgwick nor Ross is concerned to justify particular moral judgments by deducing them from what they take to be non-inferentially known; both deny this possibility (Ross 1930: 30–32; Sidgwick 1907: 379). And they agree with Sinnott-Armstrong that, in order to trust a judgment, one needs to know that it has been formed in conditions that make for reliability. Consider Sidgwick’s four tests for whether a given non-inferential judgment has “highest certainty.” He supposes that one must have justified beliefs about whether the tests have been passed (Sidgwick 1907: 338–42). The point in characterizing some moral judgments as non-inferential is simply that some moral judgments seem not to be inferred from either other moral judgments or wholly non-moral judgments, but rather are seen to be true just from understanding their content.
ch apter 11
Externalism, motivation, and moral knowledge Sergio Tenenbaum
1 1.1 I n t roduc t ion For ethical naturalists of a certain stripe, externalism about moral motivation is an attractive option. An influential form of ethical naturalism takes moral properties to be natural properties, while maintaining that moral concepts are not reducible to descriptive or non-moral concepts. Moral properties are, on this view, natural properties, and moral terms that refer to these properties play an essential role in the explanation of natural, empirically observable phenomena. These properties are epistemologically accessible to us in the same way that, for instance, physical or chemical properties are accessible to us. According to this view, moral facts are known a posteriori, and the explanatory role of moral terms plays an essential role in moral epistemology; many advocates of such a view will take moral knowledge to be warranted, inter alia, on the basis of “inferences to the best explanation.”1 On the other hand, an internalist about moral motivation thinks that there is a necessary, conceptual connection between judging that “x is morally right” and being motivated to x. A non-analytic ethical naturalist of this kind3 claims, roughly, that moral properties are natural, objective properties and that we learn about their instantiation in the world due to their causal powers2. But how could the judgment (or belief) that a certain natural property of this kind is instantiated by a certain action type (or some instances of this type) or a certain consequence of acting in a certain way be conceptually connected to a completely different item in 1 Although I am ignoring some complications, some advocates of roughly this kind of view include Boyd (1988, 2003); Brink (1989); Miller (1985); Railton (1986). 2 This is only one version of ethical externalism. For a long list of possible versions, see Brink (1989). 3 Since this is the only kind of ethical naturalism with which I will be concerned in this paper, I will just refer to it as “ethical naturalism,” unless the context is misleading, in which case I will refer to it as “non-analytical ethical naturalism.”
211
212
s e rg io t e n e n b au m
our psychological economy; namely, a motivation to act? Even if one were not inclined to accept the Humean thesis that beliefs cannot motivate on their own in its full generality, the belief or judgment that a certain natural, explanatory property is instantiated by, or is the expected effect of, certain possible actions seems to be the wrong kind of thing to exhibit this kind of conceptual connection to motivation. Thus ethical naturalists tend to accept externalism and explain virtuous and moral behavior by the existence of a desire to act morally, or a desire to perform only actions that are morally right. So moral motivation is explained in terms of a self-standing desire to be moral, which, when coupled with the belief that certain actions are morally right or morally wrong, will engender motivation in a completely unmysterious, Humeanly acceptable way.4 However, Michael Smith has mounted an important objection to this form of ethical externalism.5 According to Smith, reliance on such a desire to be moral would make the virtuous agent liable to a charge of fetishism; the virtuous agent on this picture would care for those properties of an action that make the action morally right only because, in virtue of having these properties, they make the actions right, rather than caring for the properties that make the action right directly. So rather than being motivated to help because she cares about people in dire need, the virtuous agent in this picture cares for what is morally right, and by realizing that helping people in need is morally right, she cares, derivatively, to help those in need. According to Smith, this gives the virtuous agent “one thought too many,” and does not adequately capture the motivation of such an agent; the virtuous agent should have a direct motivation to help those in need. I will argue that Smith’s objection ultimately fails; in particular, I will argue that Smith is wrong to think that the non-analytic naturalist cannot allow that the virtuous agent is typically and reliably motivated by a direct motivation to pursue the specific ends that morality enjoins us to pursue. In responding to this objection, I hope we get a better understanding of how non-analytical naturalism must see the connection between moral knowledge and moral motivation in this picture. My paper does not conclusively show that this account of moral motivation is adequate; in particular, Smith is right that non-analytic naturalism requires that the virtuous agent be directly motivated by a desire to do whatever turns out to be morally right, and for all we know, there is 4 For examples of naturalists who accept ethical externalism, see, for instance, Brink (1989) and Railton (1986). 5 See Smith (1994).
Externalism, motivation, and moral knowledge
213
something problematic about having this motivation.6 But if there is a problem in this area, it is not the fact that externalism forces the virtuous agent to have “one thought too many.” The externalist has room in her theory to say that the virtuous agent also cares directly for justice, honesty, pain relief, etc. just as well as anyone’s virtuous agent. 1 1.2 S m i t h’s obj e c t ion t o e x t e r n a l i s m According to Smith, an ethical externalist cannot escape being committed to an account of the motivation of the virtuous agent that is incompatible with certain ordinary views. Smith points out that anyone needs to explain why a rational, moral agent reliably changes his or her motivation in accordance with changes in her moral beliefs. Consider a moral, rational agent (Mary), who is persuaded that, contrary to what she used to hold, it is morally right to do X rather than Y when choosing between X and Y. We would expect to encounter a corresponding change in Mary’s motivation; in so far as Mary is virtuous, her change in belief would be accompanied by a corresponding change in motivation. That is, Mary will now be motivated to choose X rather than Y. According to Smith, the externalist can explain such reliable changes in the virtuous agent’s motivation in accordance with changes in their moral beliefs only in terms of a non-derivative desire to do what is morally right, read de dicto. That is, on this picture, we can explain why Mary’s motivational changes follow her changes in moral beliefs only if we think that her motivation to X is derived from her desire to do what is morally right (read de dicto). Given that, according to the externalist view, there is no internal connection between moral belief and motivation, it would be an incredible coincidence if every time the moral agent acquires a belief to the effect that X is right, she also acquires a non-derivative desire to do X; the externalist can explain such reliable changes only by attributing to the virtuous agent a non-derivative de dicto7 desire to do whatever happens to be morally right. So if Smith is right, on the externalist’s account, when Mary learns that she ought to provide her children with fun and enjoyment (rather
In fact, I argue that this kind of motivation is incompatible with some important aspects of moral commitment in “Moral Faith and Moral Reason” (unpublished MS). 7 I’ll use sometimes the less precise expressions “de dicto desire” and “de re desire” for simplicity; but, of course, we are not talking about two kinds of desires, but two different ways of attributing or reporting desires. 6
214
s e rg io t e n e n b au m
than, say, concern herself only with developing their talents8), she does not thereby acquire a non-derivative motivation to play with her children de dicto; her motivation to play with her children is derived from her motivation to do what is morally right and her newly acquired belief that playing with her children is morally right. Smith’s point, of course, shouldn’t be understood as a claim that the very existence of a desire to do what is morally right, read de dicto, is objectionable. On the classic way to understand the distinction in belief contexts, someone holds a de dicto belief, at least typically, if the believer would assent to the proposition in the that-clause.9 Extending this point to desire, we would expect that to desire something de dicto would also be implied, typically, by the fact that the agent would want the object of the desire as described in the content of the desire ascription.10 Let us suppose that Larry has a daughter Mary who also happens to be the captain of St. Mary’s high school football team, and who is in the running for the MVP award for the 2010 football season. Larry, as a good father, wants Mary to do well. Let us now compare the following sentences: (1) Larry wants Mary to get the MVP award. (2) Larry wants his daughter to get the MVP award. (3) Larry wants the captain of the St. Mary’s high school football team to get the MVP award. If Larry knows that Mary is his daughter and that she is the captain of the high school football team, in a very clear sense all of (1)–(3) are true of Larry even on a de dicto reading; these are all things that he wants and he is disposed to pursue the objects of these desires under any of these descriptions. So, for instance, if he reads that the captain of a high school team can win an MVP award only if she has a signed form from her principal, he will be disposed to make sure that the principal signs the relevant form. Similarly, if Larry comes to believe that Mary wins the award, all the following will be true of him even when read de dicto: (1) Larry believes that Mary got the MVP award. 8 A view inspired by Roger Scruton’s comment about his bringing up his son: “It goes without saying that Sam will not enjoy his childhood … But that is not the point. Childhood is not an end in itself but a means to growing up.” Cited in Heller (1999). 9 Obviously there are other ways to understand the distinction. There is also an understanding of “de re beliefs” as a relation to an actual object, or as a difference of the scope of the existential quantified relative to the attitude verb, but these readings are obviously even further removed from the sense that Smith intends in the discussion of the fetishism objection. 10 I am using this more convoluted formulation to avoid commitment to the view that the desires and wants are attitudes whose content is necessarily propositions or sentences.
Externalism, motivation, and moral knowledge
215
(2) Larry believes that Larry’s daughter got the MVP award. (3) Larry believes that the captain of St. Mary’s high school football team got the MVP award. It follows that on any account the virtuous agent has a de dicto desire to do what is morally right. For, presumably, the virtuous agent believes (read de dicto) that it is right to do what is morally right. Thus it is a consequence of Smith’s practicality requirement that if the agent now does not suffer from any relevant form of irrationality, as is the case of the virtuous agent, she will also be motivated to do what is morally right.11 But now there seems to be no reason to think that she would not recognize this as a description of what she wants. In other words, we can ascribe a desire to be moral to a virtuous agent even in opaque contexts without making the agent less virtuous for this reason.12 Moreover, on everyone’s view, the virtuous agent has a de dicto desire to pursue X, once she recognizes that X is morally right. On both views, once the virtuous agent recognizes that, say, helping the poor is morally right, then, she has thereby a de dicto desire to help the poor, at least because she recognizes “helping the poor” as a correct description of what (or of one of the things) she desires. So we cannot find any difference regarding de dicto and de re desires when it comes to Smith’s views or the externalist’s with respect to the virtuous agent. Moreover, given that, as I’ll argue shortly, the point seems to be about the basing or the inferential status of one’s desire, specifying that the desire is de dicto is unnecessary; after all, the contexts in which we are examining the agent’s inferences or the basing of her beliefs are typically contexts in which the attitudes are being ascribed de dicto. Even though all views in question attribute the same desires to the virtuous agents, they don’t attribute it in the same way. According to Smith, the externalist has trouble explaining how the virtuous agent could have 11 More precisely, the practicality requirement says: “If an agent judges that it is right for her to Φ in circumstances C, then either she is motivated to Φ in C or she is practically irrational” (Smith 1994). But since the antecedent is presumably true of the virtuous agent when we replace “Φ” with “act morally” or “do that which is morally right” and “C” with any circumstance, it follows that the virtuous agent has a de dicto desire to do what is morally right. 12 Jonas Olson, for instance, argues that a virtuous agent could have both kinds of desires (Olson 2002). If I am right, Smith is already committed to accepting that the virtuous agent has both kinds of desires. But I do not see why he needs to deny this; the point is that only the virtuous agent can reliably acquire de re non-derivative desires. Olson also points out that often moral action can be motivated only by a de dicto desire to do what is morally right in a non-objectionable way. This is probably correct, but all that Smith needs to make his point is that the virtuous agent needs to be reliably motivated by a non-derivative de dicto desire to do X when X is morally right, not that she must be always so motivated.
216
s e rg io t e n e n b au m
certain non-derivative motivations that are obviously included in our ordinary understanding of the motivational makeup of a virtuous agent: Good people care non-derivatively about honesty, the weal and the woe of their children and friends, the well-being of their fellows, people getting what they deserve, justice, equality, and the like, not just one thing: doing what they believe to be right, when this is read de dicto and not de re. (Smith 1994: 75)
But the question is how to spell out the notion of “derivative” and “nonderivative” that gives rise to a serious difficulty for the externalist, or at least for our kind of non-analytic naturalist externalist. As we will see in the next section, this is not as easy as it might seem. 1 1.3 De r i vat i v e , non-de r i vat i v e , a n d i ns t ru m e n ta l mo t i vat ion We can start by assuming that “derivative” refers to some kind of rational or inferential connection (or a causal connection that obtains in virtue of, or is constitutive of, the agent’s rationality); it concerns the basing of a desire, or how it is inferentially arrived at. After all, the fact that a certain motivation just has a certain causal genesis that involves other motivational attitudes should not be, at least not systematically, enough to disqualify it from being the right kind of motivation. So, if Larry fell in love with his wife because he was attracted to chess players, or because they met at a time when he was emotionally vulnerable and thus disposed to engage in charged conversations with recent acquaintances, then these motivations are in the causal path to his motivation to help his wife whom he now loves. But the fact that these items figure in the causal genesis of Larry’s motivation to help his wife does not make his desire to save her life “derivative” in any objectionable way.13 On the other hand, an agent’s motivation to save his wife might be objectionable if the agent cares for his wife only in so far as morality demands that he cares for his wife; this kind of motivation seems objectionable exactly because of the way in which the agent’s desire to help his wife is inferentially arrived at,14
13 Of course, there might be cases in which the causal genesis of a desire might make the motivation problematic in various ways. However, it would be hard to claim that it makes it “derivative” in this way. 14 “Inferentially arrived at” here needs to be understood in a weak sense; we do not want to say that the desire is objectionable only if it is arrived at by explicit deliberation. I will leave it open how to understand “inferentially arrived at” more precisely; all that my argument requires is that not all causal paths leading to a desire count as inferential paths.
Externalism, motivation, and moral knowledge
217
or based on,15 the desire to perform only actions that are morally right. We can think about two “extremes” about what would make a certain desire or motivation derivative: [MAXIMUM] A motivation to ϕ is derivative if it is inferentially arrived at from (or based upon) any other mental states;
and: [MINIMUM] A motivation to ϕ is derivative16 only if it is inferentially arrived at from (or based upon) a belief that ϕ-ing is causally necessary to bring about that p together with a desire that p.
In a nutshell, MAXIMUM counts as derivative all but what Nagel would call “unmotivated desires,”17 while MINIMUM counts as derivative at most one’s instrumental desires.18 But we can see that even given the setup of the objection to externalism, one is already committed to rejecting MAXIMUM. After all, in our example Mary is persuaded that it is morally right to provide her children with fun and enjoyment. But now it seems that the beliefs that she formed in arriving at the conclusion that it is morally right to provide her children with fun and enjoyment are part of the basing of her motivation to provide her children with fun and enjoyment. In fact, [MAXIMUM] makes her motivation inappropriately derivate even when her motivation to do so is simply based on her normative belief that she should provide her children with fun and enjoyment; arguably, even Smith himself is committed to this view. This would make not only the externalist account of moral motivation defective, but also any account that took moral motivation to be typically based on normative beliefs. Obviously, this does not show that the objection fails, but that [MAXIMUM] is far too strong a condition. On the other hand, [MINIMUM] would deliver Smith’s argument only if the naturalist were committed to the view that, for instance, “helping the poor” is a causal means to “doing what is morally right.” But the externalist 15 That is, the desire to do what is morally right is part of the reason or the basing of the desire to save his wife. Here, too, the question of how to understand the basing relation is a rather controversial one, but, again, all that my argument requires is that the basing relation cannot be understood as just. 16 Of course, we are defining “derivative” here with the assumption that derivative in this sense is a form a derivative motivation that would be in some way inappropriate for a moral agent; that is, the kind of derivative motivation to do those things that are morally right that we would not expect to find in a virtuous agent. 17 See Nagel (1970). 18 David Brink at some points seems to claim that the fetishism objection would apply only to instrumental desires. See Brink (1997).
218
s e rg io t e n e n b au m
need not say that the relation between “helping the poor” and “being a morally right action” is of this kind; the externalist need not say that by performing an action that instantiates the property “helping the poor” one has brought about, as a separate effect, the instantiation of another property; namely, “being a morally right action.” The more plausible view is that “helping the poor” is just a way of, or an instance of, “being a morally right action.” No externalist is committed to the view that helping the poor is an instrumental means to being a morally right action, and the non-analytic naturalist is particularly unlikely to make this move; the naturalist will probably identify the property of being morally right with a complex property that will have things like “helping the poor” among its constituents.19 One obvious intermediate option is to restrict the class of derivative motivations to motivations inferentially arrived at from, or based on, other desires or motivational states. We can define this intermediate position as follows: [INTERMEDIATE 1] A motivation to ϕ is derivative if it is inferentially arrived at from, or based on, a motivational state to ψ (or that p) and certain beliefs that connect20 ϕ-ing with ψ-ing (or p).
So this would still make the externalist version of the virtuous agent a case of inappropriately derivative motivation. However, it is difficult to accept [INTERMEDIATE 1]. Let us look at one of Smith’s rational requirements of motivation, the requirement to acquire more general desires that provide systematic justification for our existing particular desires.21 So in so far as an agent acquires new motivation in virtue of the realization that adding a new desire to her motivational set would give her a more coherent motivational set, her new desire would be derivative under [INTERMEDIATE 1]. But suppose Shoshana improves her moral outlook exactly by making her motivation more coherent; suppose she notices that she treats teachers inexplicably differently from the way she treats lawyers, and, as a result of this realization, and independent of any moral 19 In The Moral Problem (1994), Smith carefully uses the terms “derivative” and “non-derivative” to describe what the externalist is committed to. In a later paper, Smith uses “instrumental” and “non-instrumental,” which does suggest that he takes “derivative” in the sense described by [MINIMUM]. See Smith (2004c). But, again, the externalist need not be committed to the claim that the more specific desire is derivative in the sense defined by [MINIMUM]. 20 This has to be left vague, since the whole point of this condition is to rule out any inferential move from one kind of motivation to another. 21 Smith (1994: 159–61). Smith explicitly classifies at least the more general desires arrived at in this way as “underivative” (159–61).
Externalism, motivation, and moral knowledge
219
argument, she changes the way she treats teachers. It seems that Shoshana is just as virtuous as if she had changed the way she treated teachers by an explicit argument to the effect that her differential treatment of lawyers and teachers was morally wrong. But [INTERMEDIATE 1] would classify Shoshana’s new motivation as derivative. Another obvious candidate is to extend the category of means in [MINIMUM] to include all constitutive means and make it a sufficient, rather than necessary, condition for a motivation being derivative: [INTERMEDIATE 2] A motivation to ϕ is derivate if it is inferentially arrived at from, or based on, a belief that ϕ-ing is necessary (causally or constitutively) to bring about that p (or to ψ) together with a desire that p or to ψ.
The first problem with [INTERMEDIATE 2] is that it is difficult to see why some typical cases of constitutive means make the motivation derivative. For instance, suppose you are spending the afternoon with your daughter and she asks: “Why did you take the time off work to go for a walk with me in the park?” Suppose you say: “because I want to spend more time with you,” and your daughter says, disappointed: “I thought you just wanted to go for a walk with me in the park.” There is a way of interpreting your answer that makes your daughter’s disappointment justified. If you have this grand plan of being a better parent, and you realize that being a good parent involves among other things, spending more time with your daughter, and you realize that today there is a good opportunity to put the plan into action and thus take your daughter for a walk in the park, while at the same time you feel jealous about your colleagues who are at this moment in a meeting discussing next year’s departmental budget, there is no doubt that your daughter could, and perhaps should, be expecting more from you. But if you simply enjoy spending time with your daughter, and you notice that you’d like to do more of this, and one of the activities that would count as “spending time with your daughter” that you particularly enjoy is taking walks with her away from the city, and you feel particularly eager do this today and in the park, then your daughter probably could not and certainly should not expect more from you. Smith would like to classify the externalist type of motivation as falling into the first category, but [INTERMEDIATE 2] does not discriminate between the two. In fact, it is not clear that the difference between the two can be cast in terms of one motivation being more derivative than the other, or derivative in a different way. “Helping the poor” can be constitutive of the end of “helping the needy” or of the end of “expressing one’s charitable disposition.” Arguably, the latter is a
220
s e rg io t e n e n b au m
case of “one thought too many” but not the former.22 Defining “derivative” in such a broad manner cannot distinguish between these two kinds of motivations. It was worth looking at another, rather obvious, problem with [INTERMEDIATE 2]. At the limit, one of the things that is constitutive of helping the poor is helping the poor itself; and thus [INTERMEDIATE 2] threatens to trivialize the distinction between “derivative” and “not derivative.” Of course, it is not hard to think of various ways one can try to fix this problem, but it is not clear that we can avoid trivialization without jeopardizing the criticism against at least the naturalist externalist. For, after all, the non-analytic naturalist typically claims that the property of being morally right is identical with a certain natural property. But if this natural property is the one that we should want to instantiate for its own sake, then the naturalist would not be guilty of making moral motivation derivative in any objectionable way. So, for instance, if the only property that virtuous agents are responding directly to is the maximization of general expected utility then by responding to the property of being morally right, a virtuous agent is responding to the property of maximizing expected utility. But this might lead us to a more perspicuous way of putting the potential problem for the externalist in this vicinity. One might argue that morality calls for acting on motivations whose content is, at least potentially or for all the agent knows, distinct from what morality actually requires. But the externalist cannot explain how a virtuous agent can have her motivation reliably track her moral views and at the same time allow that the virtuous agent will, often if not always, be motivated by desires whose objects are indifferent to changes in her moral views. In fact, in answer to objections, Smith says: But if the only mechanism that exists for causing and sustaining moral motivation … is the desire to do what is morally required, then it seems … that they [sc. morally virtuous people] will positively eschew any non-instrumental concern [for their family and friends]. After all, a non-instrumental desire to look after family and friends wouldn’t be kept in check by a desire to do the right things under conditions of moral belief revision. (Smith 2004c: 286) 22 This means that “one thought too many” might be a misleading way to express the problem. One might think that since Smith restricts his claim to changes of mind regarding “fundamental values,” these issues are irrelevant. But this is not correct; fundamental values can be constitutive of other fundamental values. Note that Smith’s own list of things that the virtuous agent cares non-derivatively about (cited above) includes fundamental values that seem to stand in this relation (e.g., “people getting what they deserve” and “justice”).
Externalism, motivation, and moral knowledge
221
One might think that this argument sidesteps the need for coming up with a more precise understanding of “non-derivative” or “direct” motivation. For we already know that the motivation attributed by the externalist cannot be of the right kind, since in the externalist picture, the virtuous agent can be reliably motivated to do what is morally right only if she eschews the right kind of motivation. But this kind of criticism underestimates the resources available at least to the naturalist version of externalism, or so I will argue. 1 1. 4 A r e ply t o S m i t h’s a rgu m e n t Let us use a toy naturalist theory in which “morally right” is identified with the complex property that is instantiated by any possible action which gets at least a certain minimum score in a weighted sum of various graded properties. So, for instance, “causing pain relief to degree d” might be one of these properties, and “instantiates helping a friend to degree d” might be another one, and, of course, some properties might have negative weights (“causing suffering”, etc.);23 the weighted sum also settles disputes about what is morally better. For simplicity’s sake, I’ll assume that, on this theory, there is always one morally right action, the one that is morally best.24 Although I will focus on this toy theory, I hope that it will be clear that various (and more plausible) versions of naturalism could use the same kind of response to Smith’s objection.25 Now suppose an action is morally right in virtue of being a case of helping a friend to a very high degree (and to make things easier let us suppose that there is no other available action that is also morally right). Now the externalist is not committed to the view that the virtuous agent needs to figure out all the weighted sums in order to be in a position to know that this is the morally right action. In fact, it is compatible with the externalist that an agent knows which actions are morally right by being able to reliably tell, for instance, when actions exemplify enough of one of these properties. So in our case the virtuous agent might know that: (A) This action is morally right in virtue of instantiating property X to degree d. 23 Of course the second one is not a clear example of a natural property. But for our purposes it does not matter. 24 I will ignore ties, too. 25 So, for instance, Richard Boyd’s view that, roughly, identifies the property of moral goodness with a cluster of properties could easily adapt the account of moral motivation expounded here. See Boyd (1988).
222
s e rg io t e n e n b au m
Moreover, it is compatible with the externalist view that the virtuous agent also cares, at least in the ideal case, for each of the component properties, and thus that she would have an independent motivation to relieve pain or help a friend. So in this case the agent knows that a certain action is morally right, is motivated to do what morality requires, but the de dicto desire to do what is morally right plays no essential role in her motivation. Obviously this does not yet answer Smith’s concern. It is unclear how a virtuous agent, on this picture, could change her mind about what is morally right and reliably act accordingly; after all, her motivation in this picture does not seem to dovetail changes in moral judgments. But here the desire to do what is morally right can perform a different, but also completely unobjectionable role. It is easier to see this case if we look first at fundamental changes in moral beliefs regarding the relative importance of different constituent properties; in our toy theory, this would amount to changes in beliefs regarding relative weights of the component properties. But since, ex hypothesi, the virtuous agent already has the direct motivation to act so as to bring about the instantiation of either property, all that changes now is that the desire to be moral adds motivation to give more weight to one of these component properties. For instance, suppose that Mary used to think that one should always tell the truth even when it might hurt someone’s feelings. But now, after talking to her friends, she is convinced that sometimes she should lie to protect her friend’s feelings; in particular, she’s now convinced that she should not tell Larry that his poetry is juvenile. Before having these conversations, Mary was ready to visit Larry and let him know that his poetry was juvenile, but now she instead tells him that she finds his poetry very deep and inspiring.26 Before and after her conversation Mary cared directly about honesty and the weal and woe of her friend. However, the desire to be moral first required that she gave more weight to honesty and later to not to hurt her friend’s feelings, and the extra motivational force of the desire to be moral tips the balance in favor of either desire, depending on her belief about what is morally right. But it is not clear what could be objectionable about the motivation to be honest playing that role. We can see this more clearly if we think that there are two independent desires that correspond to two different types of moral beliefs. One kind of moral belief is about which properties are components of the “morally Perhaps needless to say, I am not endorsing the moral judgment here.
26
Externalism, motivation, and moral knowledge
223
right,”27 and a corresponding desire to engage in actions that are morally right (that exemplify such properties). Now these beliefs generate the direct motivation to be honest and to care about one’s friends’ feelings. This motivation counts as direct, because the virtuous agent does not believe that such things are instrumental in bringing about what is morally right, but they are instances of the morally right. The second kind of moral belief is a belief about the various weights that are attached to the graded properties that constitute what is morally right;28 in our case, these would be the newly acquired belief that it is more important not to hurt one’s friend’s feelings. And here we can think that the moral agent has a desire to do what is morally best, a desire that earlier “sided” with the direct motivation to be honest and now with the direct motivation not to hurt her friend’s feelings. It is hard to see now why this latter kind of motivation would be objectionable. After all if asked: “Why did you lie to your friend; don’t you care about honesty and truth-telling?” Mary’s answer could appropriately be something like: “I care about both, but in this case, it was more important not to hurt my friend’s feelings.” But this fits well with what my externalist naturalist would say about this case; it seems plausible to think that this kind of statement expresses that one’s motivation was settled by the desire to do what is right when those two competing motives were in play. Accepting this point does not imply that the virtuous agent needs to eschew the direct motivation to look after friends or be honest, given that this kind of direct motivation plays an essential role in bringing about her virtuous actions. It is worth noting that one could easily modify this naturalist response to Smith’s objection even if one thought that the motivation that is derived from this kind of belief about what instantiates moral rightness is also objectionable. For as long as our naturalist does not have too long a list of component properties, it can be part of the conception of a virtuous agent that the virtuous agent must have an independent direct motivation to instantiate, or bring about the instantiation of such a property. So under this conception part of what makes someone count as a virtuous agent is that she has a direct motivation to be honest, to care for the wellbeing of her friends, etc. The rest of the solution remains the same, except now the only role the desire to be moral has is to side with the motivation
27 Something roughly along the lines of “which properties are pro tanto good.” 28 Roughly, these would be beliefs about how to move from judgments of pro tanto good to judgments about overall good.
224
s e rg io t e n e n b au m
that, according to the agent’s moral beliefs, is the one that follows that balance of moral reasons.29 Of course, this modified solution seems to fail if the change in virtuous agents’ belief is of a more radical nature. The modified solution does not seem to apply to the case in which the moral agent comes to believe for the first time that an action can be morally right in virtue of instantiating a property that he previously thought was morally irrelevant. Suppose Jane is convinced for the first time that loyalty to one’s country is morally right, something that should count in favor of our actions. Suppose she had not previously had any direct motivation to perform patriotic acts. On the account we have, this new belief about what is morally right cannot guarantee that, in so far as she is rational, Jane will now have a direct motivation to be loyal to her country. But it is not clear that it is part of the ordinary conception of morality that change in belief could produce direct motivation in this way. Here it might be the case that Jane can only perform her duty to her country from her desire to do what is morally right.30 It might be the case that a virtuous person would, in such a situation, have a second-order desire to acquire the relevant direct motivation, but it does not seem so intuitive that we need to think that a virtuous agent must, just in virtue of her rationality, care directly for patriotic acts as soon as she is persuaded that it is morally right to perform patriotic acts, even if she previously saw no value in them and had no motivation to perform patriotic acts. 29 Perhaps one of the moral beliefs of the moral agent is that she must acquire this kind of direct motivation, and thus the moral agent would have a second-order desire to directly desire certain things. On this view, one does not count as a virtuous agent until this desire is at least partly satisfied. Jamie Dreier also suggests that second-order desires can help answer Smith’s challenge to the externalist (Dreier 2000). However, the second-order motivation to acquire direct motivation to X whenever one forms the belief that X is morally right cannot, on its own, make it the case that the virtuous agent reliably acquires the direct motivation when she changes her moral beliefs in so far as she is rational. For rationality alone does not guarantee that we have the firstorder motivation that we desire to have. So, for instance, take Frankfurt’s famous case of the psychiatrist who wants to (intrinsically) desire taking drugs so that he will know how it feels to be a drug addict. The second-order desire, on its own, cannot guarantee that the psychiatrist actually desires (intrinsically) to take drugs, no matter how rational the psychiatrist is. Similarly my desire to intrinsically desire to eat Brussels sprouts (rather than just desiring it as a means to better health) unfortunately does not get satisfied just by virtue of my being rational. All that Dreier can show is that the moral agent on this picture will want to acquire the first-order motivation, not that he will acquire it. Of course, one could simply postulate a mechanism in the virtuous agent that guarantees the formation of the first-order desire. But then one might as well postulate a mechanism that works directly from the moral belief. 30 This parallels some of the cases that Olson considers (2002). See also Hume’s interesting discussion of how the sense of duty can be a secondary motive to perform virtuous actions (2000 [1739]).
Externalism, motivation, and moral knowledge
225
Smith is right that the naturalist can explain the reliable motivation of a virtuous agent only by postulating a non-derivative desire to do what is morally right, and, ultimately, there might be something problematic about thinking that such a desire plays a prominent role in moral motivation. Is it really plausible to think that a virtuous agent cares directly about morality as such, independently about the particular ends that she judges to be morally right? Perhaps other aspects of non-analytic naturalism make this picture of moral motivation even more problematic. It seems strange to think that virtuous motivation would change its direction, for instance, simply by a new discovery that certain properties play a certain role in the explanation of social facts.31 Perhaps this kind of moral motivation is incompatible with other features of moral commitment. I have not tried here to allay all these concerns.32 But whatever problems there might be with the content of a direct motivation to do what is morally right, the existence of such a motivation, or even the role it has to play in the externalist account, is not incompatible with the direct motivation to pursue specific ends that, as Smith rightly observes, must be part of the motivational makeup of the virtuous agent.33 31 Adams (1999) raises an objection along similar lines; Boyd (2003) tries to provide an answer to Adams’ objection. 32 In fact, in “Moral Faith and Moral Reason” (unpublished MS), I argue that the non-analytic naturalist version of externalism is incompatible with important features of moral commitment. 33 I would like to thank Tom Hurka, Jennifer Nagel, and Rob Shaver for very helpful comments on an earlier version of this paper.
ch apter 12
Naturalism, absolutism, relativism Michael Smith
Gilbert Harman begins his essay “Is There a Single True Morality?” by telling us: As far back as I can remember thinking about it, it has always seemed to me obvious that the dictates of morality arise from some sort of convention or understanding among people, that different people arrive at different understandings, and that there are no basic moral demands that apply to everyone. (Harman 2000a: 77)
Having said this, however, he immediately admits that this opinion is not shared by many of his philosophical colleagues, and he goes on to explore the issue that divides them. Harman’s hypothesis is that they have different attitudes towards naturalism. Naturalism, he tells us, decisively favors the view that basic moral demands apply only to some. He therefore spends the bulk of his paper trying to make out that connection. This is an extremely important conclusion, if it is correct, so my aim in what follows is to consider the arguments Harman provides to support his hypothesis. To anticipate, I will argue that the view that moral demands apply only to some gains no support from naturalism, and I attempt to identify where the real source of support for that view lies. 1 2 .1 A b s olu t i s m v e r s us r e l at i v i s m Let us begin by clarifying the two views about morality that are up for discussion. As we have already seen, one of these is the view that basic moral demands apply only to some, not to everyone. Harman calls this view Earlier versions of this paper were presented at the Conference in Moral Philosophy, University of Reykjavík, and at the Institut für Philosophie, Humboldt-Universität zu Berlin. I am grateful to all of those present on these occasions who gave me such helpful comments. I am also grateful to the editors for their advice on the penultimate version. My research was supported by a Research Award from the Alexander von Humboldt Foundation.
226
Naturalism, absolutism, relativism
227
“moral relativism.” The alternative to relativism is a view that Harman calls “moral absolutism”: [L]et me stipulate that I will take moral absolutism to be a view about the moral reasons people have to do things and to want or hope for things. I will understand a belief about absolute value to be a belief that there are things that everyone has a reason to hope or wish for. To say that there is a moral law that ‘applies to everyone’ is, I hereby stipulate, to say that everyone has sufficient reasons to follow that law. (Harman 2000a: 84)
Harman’s stipulative definition of “absolutism” helpfully clarifies what he finds appealing in relativism. When Harman says that it seems obvious to him that there are no basic moral demands that apply to everyone, what he means is that it seems obvious to him there is nothing that everyone has sufficient reason to hope or wish for (no absolute moral values) and that there is nothing that everyone has sufficient reason to do (no absolute reasons to follow the moral law). The appeal of relativism over absolutism, for Harman, therefore lies in its more plausible theory of what we have reasons to do and hope and wish for. According to relativism, though moral values do entail sufficient reasons to hope or wish for the things that are of moral value, and though the moral law does entail sufficient reasons to do the things that the moral law requires us to do, what is entailed is that only some people have sufficient reason to hope or wish for or do these things. Though the contrast between relativism and absolutism may seem stark when it is put like this, we should note an ambiguity. Does the moral absolutist hold that there are things that every human agent that we know about has sufficient reason to do and hope and wish for? Or, more strongly, does he hold that there are things that every possible human agent has sufficient reason to do and hope and wish for? Or, more strongly still, does he hold that there are things that every possible rational agent – here we remove the restriction to humans – has sufficient reason to do and hope and wish for? Since “absolutism” has been defined by stipulation, and since Harman doesn’t say which of these interpretations of “everyone” he has in mind, we will need to decide for ourselves. My proposal is that we read Harman’s stipulative definition of absolutism as charitably as we can, given the argument that he is hoping to put forward. We should therefore suppose that he has in mind absolutism of the very strongest kind. This is because if absolutism is simply the weak claim that there are things that every agent we know about has sufficient reason to do and hope and wish for, and if naturalism tells even against this weak claim, then it follows that it also tells against absolutism when
228
m ic h a e l s m i t h
it is interpreted in the very strongest way. The opposite, however, is not true. Interpreting absolutism in the strongest way thus makes it easier for Harman to confirm his hypothesis that naturalism tells against absolutism and in favor of relativism, not harder. Moreover, though absolutism interpreted in this strong way is a radical view, it is also a very familiar view in metaethics. Absolutism, so interpreted, is a version of the view, endorsed by moral rationalists, that moral requirements are requirements of rationality or reason, and hence are binding on rational agents as such. Basic moral demands, according to the rationalists, aren’t just necessary, but are also knowable a priori, as knowledge of them follows from knowledge of what it is to be a rational agent. This is why they apply to all possible rational agents. Since I take it that this is the view that at least many of Harman’s philosophical colleagues hold – I have in mind especially those inspired by Kant, theorists like Nagel (1970), Darwall (1983), and Korsgaard (1996), and those inspired by the Brentano/Ewing suggestion that there are fittingness relations between facts and attitudes, theorists like Scanlon (1998) and Parfit (2011) – it is worthwhile keeping both of these two kinds of moral rationalism, and the contemporary moral rationalists who hold these views, firmly in mind when we consider the bearing of naturalism on absolutism. These, I take it, are Harman’s real stalking horses. At this point, let me lay my own cards on the table. Though it seems to me quite plausible to suppose that there are moral demands to do or hope or wish for things just in case there are reasons for every possible rational agent to do or hope or wish for them, I am not sure whether we should accept the radical kind of absolutism that Harman describes (see especially Smith 1994: 182–84). I will explain my misgivings presently when I say something about his insistence on connecting absolutism to the view that there are sufficient reasons to do or hope or wish for things, rather than just pro tanto reasons. Moreover, though this means that I agree with Harman that there is a connection between demands and reasons, it is important to remember that both he and I therefore rule out of court a certain kind of view about the nature of moral demands, a view that is thoroughly naturalistic but not relativistic. I have in mind the account of moral demands that Peter Railton outlines and defends in his “Moral Realism” (1986). Railton begins that paper by providing a naturalistic account of what it is for something to be good for an individual, and he then makes that naturalistic account of the good for an individual do work in his account of what a moral demand is. According to Railton, the claim that morality
Naturalism, absolutism, relativism
229
demands that we do, or hope, or wish for something is simply a claim to the effect that it would maximize individual good, impartially considered, if we were to do or hope or wish for that thing. Since this account is unrestricted in its scope, it is reasonable to suppose that it applies to everyone, where “everyone” is interpreted as broadly as possible. All possible rational agents fail to comply with a moral demand when they fail to maximize the individual good, impartially considered, or so Railton seems to suggest. Importantly, however, it is also an account of moral demands which leaves it open that rational agents have no reason at all to do or hope or wish for the things that morality demands of them. To repeat, Harman’s argument for relativism simply ignores accounts of moral demands like this. I will return to this point at the end. To sum up, Harman assumes that there is a connection between moral demands and sufficient reasons – moral demands are claims about the things that people have sufficient reasons to do or hope or wish for – and, he stipulates, the issue that divides relativists from absolutists is the domain of people who have such reasons. According to Harman’s stipulations, absolutists hold that everyone has to have such reasons, and, as we are interpreting “everyone,” this is the view that every possible rational agent has to have sufficient reason to do or hope or wish for the things that morality demands. Moral absolutism is therefore a version of moral rationalism, the view that basic moral demands are claims about the reasons that people have, where these claims are both necessary and a priori. Moral relativism, by contrast, is the claim that though some possible rational agents have sufficient reasons to do or hope or wish for the things that morality demands, there are some possible rational agents who do not have such reasons. The existence of moral demands does not require that everyone has sufficient reasons to do or hope or wish for the things that morality demands. 1 2 .2 H a r m a n’s a rgu m e n t ag a i ns t mor a l a b s ol u t i s m a n d t h e f l at-f o o t e d r e s p ons e As I pointed out at the beginning, Harman’s hypothesis is that naturalism tells in favor of moral relativism. But what exactly is naturalism? Naturalism, he tells us, is the familiar methodological principle that, in doing moral philosophy, “we must concentrate on finding the place of value and obligation in the world of facts as revealed by science” (Harman 2000a: 79). The question we must ask is why he thinks that this familiar methodological principle supports the truth of relativism.
230
m ic h a e l s m i t h
Let us begin with reasons to do things. Harman begins with the assumption that, if there are any absolutist moral reasons for action at all, then there is a sufficient reason for everyone not to harm or injure others. This assumption is at least somewhat plausible because fairly minimal, and it is also a claim about moral reasons that many moral rationalists have in fact endorsed. We will therefore go along with it for the time being. With this assumption in place, he then gives the following argument for the conclusion that, at least according to naturalism, not everyone has such a reason. First Premise: if a person does not intend to do something and that is not because he or she has failed in some empirically discoverable way to reason to a decision to do that thing – inattention, lack of time, failure to consider or appreciate certain arguments, ignorance of certain available evidence, an error in reasoning, some sort of irrationality or unreasonableness, or weakness of will – then according to the naturalist the person cannot have a sufficient reason to do that thing. Second Premise: there are people, such as certain professional criminals, who do not act in accordance with an alleged requirement not to harm or injure others, where this is not due to any of these failings. (Harman 2000a: 86–87)
Since it follows from these two premises that there are people, such as certain professional criminals, who do not have sufficient reason not to harm or injure others, the upshot is that, if the premises are true, then absolutism about reasons for action is false. A similar argument could also be constructed to show that absolutism about values is false, too. This simplifies matters, as it suggests that we can focus exclusively on the plausibility of the premises of the argument just given, for if those premises turn out to be false, then the premises of the analogous argument in the case of absolute values will presumably be false, too. Consider First Premise. This premise spells out more fully both what it is to have sufficient reason for action and what a naturalist must say, given that this is what it is to have such a reason. When someone has sufficient reason for action, First Premise tells us, they are in a position to reason themselves to a corresponding intention to act, absent ignorance, irrationality, and the like. There are, in other words, no unreduced reasons for action – or more cautiously, if deciding and intending are themselves actions, then apart from reasons for mental actions like reasons for deciding and intending, there are no unreduced reasons for action (from now on I will forgo the caution) – as all reasons for action reduce to reasons for deciding or intending.
Naturalism, absolutism, relativism
231
This premise is one that we should go along with as well, because it is common ground with many of Harman’s opponents. Scanlon, for example, would certainly accept it. He thinks that ‘reason for action’ is not to be contrasted with ‘reason for intending’. The connection to action, which is essential to intentions, determines the kinds of reasons that are appropriate for them, but it is the connection with judgment-sensitive attitudes that makes events actions, and hence the kind of things for which reasons can sensibly be asked for and offered at all. (Scanlon 1998: 21)
But since someone who has a sufficient reason for intending is in a position to reason themselves to having that intention, assuming they have a sensitivity to reasons, and since an insensitivity to reasons is itself plainly a kind or irrationality, the upshot is that Scanlon would happily go along with the idea that someone who has a sufficient reason for action is in a position to reason himself to an intention so to act, absent ignorance, irrationality, and the like. First Premise also tells us that if someone who is in a position to reason themselves to an intention to act in a certain way fails to reason themselves to that intention, then this failure is empirically tractable. This is one place where naturalism comes into the picture. For what’s important about the list of possible explanations that Harman provides – “inattention, lack of time, failure to consider or appreciate certain arguments, ignorance of certain available evidence, an error in reasoning, some sort of irrationality or unreasonableness, or weakness of will” – is that it indicates the sorts of explanations that might be proffered when someone fails to reason themselves to an intention, and that, in each case, the explanation appeals to some empirical fact about the agent, a fact that is amenable to further scientific understanding. A psychologist could in principle provide us with a more fine-grained account of what each of these explanations amounts to. Whereas First Premise insists on the naturalistic respectability of both reasons for action and the explanations that we give of people’s failure to intend to do what they have sufficient reason to do, Second Premise tells us that when someone like “the professional criminal” intends to harm and injure others, no such explanation is in the offing. Two points are worth noting about this premise. The first is that, if Second Premise is true, then absolutism is false on even the weakest of the interpretations we have considered. Not only is it false that every possible rational agent has sufficient reason not to harm and injure others, but there are also rational agents we know about – drug lords, human traffickers, and others we read about
232
m ic h a e l s m i t h
in the newspapers on a daily basis – who lack such reasons. The second point is that, even if Second Premise is true, and hence even if absolutism is false on even the weakest of the interpretations we have so far considered, absolutists might retreat to a slightly different claim about the connection between moral demands and the reasons possessed by every possible rational agent. The alternative form of absolutism I have in mind claims not that if there is a moral demand not to harm and injure others, then there is sufficient reason for every possible rational agent not to harm and injure others. It claims instead that if there is a moral demand not to harm and injure others, then there is a pro tanto reason for every possible rational agent not to harm and injure others. The difference is that whereas even the weakest form of absolutism we considered earlier holds that moral demands entail reasons to intend not to harm or injure others that cannot be outweighed by other reasons, the alternative form of absolutism allows for the possibility of their being outweighed, or their being equally weighty, or their being incommensurable. Second Premise, even if true, is consistent with the truth of this alternative form of absolutism. Even the professional criminal may have a pro tanto reason not to harm or injure others, it is just that his reason not to do so is either outweighed by, or equally weighty as, or incommensurable with, other reasons to harm or injure them. I mention this alternative form of absolutism not just because it is a position in logical space that is overlooked in Harman’s argument, but also because there are rationalists who accept a form of absolutism like this. Susan Wolf thinks that moral reasons come into conflict with, and so must be weighed against, reasons of “personal perfection,” and she further thinks that, sometimes at least, it is the moral reasons that get outweighed in such cases (Wolf 1982). Reasons of personal perfection that might outweigh moral reasons include reasons to act in ways that are funny, or that display an agent’s personal style or cool. Of course, Wolf might not think that there are many, or perhaps even any, professional criminals whose criminal activities are so humorous or stylish or cool that their reasons to engage in these actions outweigh their reasons not to harm or injure others, but the logical point remains. Though Wolf agrees that moral demands provide us with reasons, she denies that moral demands entail sufficient reasons. What this alternative form of absolutism shows is that Harman’s initial assumption was not as minimal as it could have been. Harman began with the assumption that if there are any absolutist moral reasons for action at
Naturalism, absolutism, relativism
233
all, then there is a sufficient reason for everyone not to harm or injure others. This is a fairly minimal assumption, and many moral rationalists do accept it, but an even more minimal assumption that he might have made instead, one which is accepted by even more rationalists, is that if there are any absolutist moral reasons for action at all, then there is at least a pro tanto reason for everyone not to harm or injure others. Having identified this alternative form of absolutism, however, I will say no more about it. The reason, as we will see, is that if Harman’s argument succeeds in showing that naturalism counts against the stronger form of absolutism he describes, then a slight variation on that same argument will plausibly show that it counts against the alternative form of absolutism just identified, too. Let us now ask whether the premises of Harman’s argument, clarified in the way we have just clarified them, are true. In Second Premise Harman claims that the successful criminal’s intention to harm his victim cannot be explained by “inattention, lack of time, failure to consider or appreciate certain arguments, ignorance of certain available evidence, an error in reasoning, some sort of irrationality or unreasonableness, or weakness of will.” It might be thought that Harman is here relying on the more or less common-sense observation that when we look around us, we see lots of rational people, some of whom have nice intentions and some of whom don’t, and that the successful criminal is just a particularly striking example of the latter. But note that this observation – that the people we see around us, including the professional criminal, meet the minimal standards of rationality required for us to see them as agents – is far too weak to support Second Premise. What Harman needs to suppose, for Second Premise to be true, is that certain professional criminals aren’t just minimally rational, but that they are neither ignorant of nor insensitive to anything relevant to such arguments as there might be for or against their having the intentions that they have. In other words, he needs to suppose that, at least in relevant respects, they are maximally rational. But whether or not we will agree with him about this turns not on our accepting the commonsense observation, and it doesn’t turn on our commitment to naturalism, either. It turns rather on the specific assumptions that we make about what an argument for having a certain intention would look like. For example, one assumption we might make is that an agent’s intentions are principally governed by a norm of coherence: an agent’s intentions are rationally required to cohere with his intrinsic desires and his beliefs about how those intrinsic desires can be satisfied. Let’s call this
234
m ic h a e l s m i t h
MEANS–ENDS. Of course, even if an agent’s intentions are principally governed by MEANS–ENDS, we might also think that there are additional principles that govern an agent’s intrinsic desires and means–ends beliefs as well. For example, we might think that his beliefs are required to be both evidentially well supported and true, and that his intrinsic desires are required to support preferences that are both transitive and complete. Call this expanded set of principles MEANS–ENDS+. Crucially, however, if MEANS–ENDS+ exhausts the principles that govern intentions – or, to be more precise, if the principles that govern intentions are like MEANS–ENDS+ in not including any principles that allow us to directly criticize the contents of an agent’s intrinsic desires, and hence the contents of his intentions – then the upshot will be that such arguments as we can give to agents for having certain intentions rather than others will depend on which intrinsic desires those agents have to begin with. Note that a commitment to MEANS–ENDS+ need not commit us to the implausible sounding view that an agent’s intrinsic desires and beliefs – or facts about his intrinsic desires and beliefs, or facts about the intrinsic desires and beliefs he would have if his intrinsic desires supported preferences that were transitive and complete and his beliefs were epistemically justified and true (from hereon I will mostly ignore these complications) – are his reasons for intending to act in a certain way (the implausible-sounding view is discussed by Jonathan Dancy under the name “psychologism” [see Dancy 2000]). We might instead suppose that what makes non-desiderative facts, such as the fact that by hitting someone you will cause him harm and injury, a reason not to intend to hit him, is the fact that the person for whom that reason is a reason has certain intrinsic desires that would cohere best with his not intending to hit someone. An agent’s intrinsic desires, together with the beliefs he would have if he had beliefs that were epistemically justified and true, might in this way be a condition of the non-desiderative facts that are reasons being reasons, not themselves reasons (see also Schroeder 2007). But while a commitment to MEANS–ENDS+ need not commit us to the implausible-sounding view that an agent’s intrinsic desires and beliefs are his reasons for intending to act in a certain way, note that reasons for intending, even if those reasons are non-desiderative facts, would still turn out to be very different from reasons for believing. This is because, as we have just seen, the considerations that are reasons for intending would turn out to be reasons only conditionally on the agent’s having certain desires. Whether some consideration is a reason for believing, by contrast, depends on whether it supports the truth of the thing believed, something that holds or doesn’t hold quite independently of what an agent
Naturalism, absolutism, relativism
235
happens to desire. The considerations that are reasons for believing are thus not reasons only conditionally on the agent’s having certain desires. We will return to this point below. If something like MEANS–ENDS+ exhausts the principles that govern the formation of an agent’s intentions then Second Premise turns out to be true. Certain professional criminals’ intrinsic desires are so unusual that the non-desiderative fact that by hitting people they will cause them harm and injury is not a reason not to intend to hit them, because a crucial condition of that non-desiderative fact’s being a reason is absent. Such professional criminals’ possession of the intention to harm and injure others is thus not the result of their insensitivity to any argument. Assuming that MEANS–ENDS+ exhausts the principles that govern the formation of an agent’s intentions would therefore lead us to conclude that only some people have sufficient reasons not to harm or injure others. (Indeed, it would lead us to conclude that only some people have even a pro tanto reason not to harm or injure others as well. This is why I said earlier that a variation on Harman’s argument would work even against those absolutists who hold that moral reasons, even though they are possessed by all possible rational agents, are merely pro tanto reasons.) What’s striking about this argument, however, is that naturalism plays no role at all in it. The argument is entirely driven by the specific assumptions we have made about the principles that govern intentions, the assumption that MEANS–ENDS+ exhausts these principles. If we make different assumptions, things look very different. Suppose, for example, we assume that an agent’s intentions aren’t just governed by MEANS–ENDS+, but that in order to be fully rational they must also meet a requirement of universality. If we further think, as some Kantians do, that we can derive (say) Kant’s Formula of Humanity from his Formula of Universal Law, perhaps together with some additional premises – let’s call this whole package UNIVERSALIZATION+ (compare Korsgaard 1996) – then we will conclude that the professional criminal does have a reason not to harm or injure others. For in having that intention he will display an insensitivity to the arguments that show that the only intentions that are universalizable are intentions to treat people never merely as a means, but also always as an end. If these assumptions are correct, Second Premise is false. But supposing that these assumptions are correct shows nothing whatsoever about our commitment to naturalism. There is nothing anti-science in supposing that reasons are universalizable. Or suppose we assume that there are certain considerations – that is, certain ways the world could be – that count in favor of everyone’s having
236
m ic h a e l s m i t h
certain intrinsic desires rather than others quite independently of what they desire. Perhaps the facts that harm and injury have the intrinsic natures that they do provide everyone with a reason to intrinsically desire that they not be harmed or injured, where these reasons are unconditional in just the same way that reasons for believing are unconditional. Let’s call this view REASONS+ (Scanlon 1998). In that case, too, we will conclude that the professional criminal has a reason not to harm or injure others. For in having the intention to do so he will display an insensitivity to arguments that would, if he weren’t so insensitive, move him from premises about the intrinsic natures of harm and injury to having intrinsic desires that people not be harmed or injured, intrinsic desires that best cohere with his having the intention not to harm or injure others. If this assumption is correct, then Second Premise is once again false. But supposing that reasons for desires and aversions are just like reasons for belief in being unconditional doesn’t seem to be in any way contrary to the impulse to find the “place of value and obligation in the world of facts as revealed by science.” What all of this suggests is a rather flat-footed response to Harman’s argument. First Premise insists on the naturalistic respectability of both the reasons for doing or hoping or wishing for things and the explanations that we give of people’s failure to be sensitive to such reasons. Second Premise then claims, in effect, that absolutist reasons are not naturalistically respectable. But Second Premise is true only if we make certain assumptions about the principles of reason that govern an agent’s intentions, and Harman hasn’t said anything at all about how naturalism bears on which assumptions we should make. True enough, if MEANS– ENDS+ exhausts the principles that govern an agent’s intentions, then Second Premise is true. But if intentions are governed as well by either UNIVERSALIZATION+ or REASONS+, then Second Premise is false. The flat-footed response to Harman is thus that, absent some reason for thinking that naturalism tells especially in favor of the principles that govern intentions being more like MEANS–ENDS+, and less like UNIVERSALIZATION+ and REASONS+, naturalism has no bearing at all on whether Second Premise is true, and hence no bearing at all on the disagreement between relativists and absolutists. 1 2 .3 F i r s t r e ply t o t h e f l at-f o o t e d r e s p ons e Harman considers a version of this response. The absolutist might agree that the criminal must be irrational, or at least unreasonable. Seeing that a proposed course of action will probably cause serious
Naturalism, absolutism, relativism
237
injury to some outsider, the criminal does not treat this as a reason not to undertake that course of action. This must be irrational, or unreasonable, because such a consideration simply is such a reason and indeed is an obvious reason, a basic reason, not one that has to be derived in some complex way through arcane reasoning. But then it must be irrational, or at least unreasonable, for the criminal not to care sufficiently about others. The criminal’s lack of concern is what’s responsible for the criminal’s not taking the likelihood of harm to an outsider to be a reason against the proposed course of action. This is one way an absolutist might argue. The relativist’s reply to such an argument is that, on any plausible characterization of reasonableness or unreasonableness (or rationality or irrationality) as notions that can be part of the scientific conception of the world, the absolutist’s claim is just false. Someone can be completely rational without feeling respect and concern for outsiders. But, of course, this reply appeals to naturalism. The absolutist who rejects naturalism in favour of autonomous ethics relies on an unreduced normative characterization of rationality and irrationality (or reasonableness and unreasonableness). (Harman 2000a: 90)
Harman here argues that the relativist’s conception of rationality or reasonableness is itself more naturalistically respectable than the absolutist’s. This is because the absolutist, unlike the naturalist, relies “on an unreduced normative characterization of rationality … or reasonableness.” But is this really true? Remember that, according to Harman, there are no unreduced sufficient reasons for action. All sufficient reasons for action reduce to sufficient reasons for intending. As Harman imagines the absolutist’s view in this passage, however, the absolutist denies this, claiming instead that the fact that a course of action would probably cause serious injury to some outsider is a “basic reason” not to undertake that course of action, “not one that has to be derived in some complex way through arcane reasoning.” I take it what he means by this is that, according to the absolutist, the criminal’s reason not to harm and injure others isn’t explained by his having reasons to intend not to harm and injure others, but that the reverse is the case. According to the absolutist he imagines, it is “irrational, or at least unreasonable, for the criminal not to care sufficiently about others” because his “lack of concern is what’s responsible for [his] not taking the likelihood of harm to an outsider to be a reason against the proposed course of action.” In other words, we appeal to unreduced reasons for action in explaining the rationality or reasonableness of an agent’s intentions, not vice versa. So this is, in effect, to suppose that the absolutist rejects First Premise. Harman carefully notes that this is “one way an absolutist might argue.” This is true. But it is not the only way he might argue, and it is
238
m ic h a e l s m i t h
not the best way for him to argue, either. In order to see that this is so, consider the two absolutist responses to Harman’s argument I described earlier. Both the absolutists I described earlier agree with First Premise. They think that sufficient reasons for action reduce to sufficient reasons for intending. It is simply that they also hold, as Harman must too, that in explaining what a sufficient reason for intending is, we need to appeal to the rational principles that govern intentions, whatever those rational principles turn out to be, because it is these rational principles that fix what the reasons for intending are. The difference between the absolutists that I described and their relativist opponent is simply that they appeal to different rational principles. Whereas the relativist supposes that rational principles like MEANS–ENDS+ exhaust the principles in play, the absolutists think that a role is also played either by principles like UNIVERSALIZATION+ or REASONS+. In order to explain the irrationality or unreasonableness of the professional criminal’s lack of concern for others, these absolutists therefore do not need to appeal to unreduced reasons for action. Instead they try to explain why the professional criminal has a sufficient reason not to harm or injure others by appealing to the fact that he has a sufficient reason to intend not to harm or injure them. Naturalism’s insistence that there are no unreduced reasons for action thus cuts no ice against them. 1 2 . 4 S e c on d r e ply t o t h e f l at-f o o t e d r e s p ons e Perhaps Harman has in mind a different objection. In that same passage he tells us that “on any plausible characterization of reasonableness or unreasonableness (or rationality or irrationality) as notions that can be part of the scientific conception of the world … [s]omeone can be completely rational without feeling respect and concern for outsiders.” As he sees things, the requirement that we come up with a characterization of rationality or reasonableness that squares with our scientific conception of the world thus tells against there being principles governing intending like UNIVERSALIZATION+ or REASONS+. But what is the connection supposed to be between coming up with a characterization of rationality or reasonableness and having a scientific conception of the world? In order to answer this question, we need to answer a prior question first. What are we doing when we come up with a characterization of rationality or reasonableness? One answer, inspired by the thought that the norms of rationality or reasonableness are inter-defined with our concepts of belief and desire,
Naturalism, absolutism, relativism
239
is that a characterization of rationality and reasonableness will follow from a spelling out of everything that we can know a priori about belief and desire. For example, those relativists who think that we can exhaustively characterize rationality or reasonableness in terms of something like MEANS–ENDS+ might suggest that this follows from the fact that all we can know a priori about belief and desire is that they are the states they are in virtue of their characteristic, but complementary, directions of fit: belief is a state that aims to represent things as being a certain way when and because they are that way; desire is a state that aims at its own satisfaction. These claims about the aims of belief and desire might equally well have been put in more explicitly normative terms: beliefs are states that are supposed to be true and well justified (beliefs are supposed to be knowledge); desires are states that are supposed to be satisfied. Or they might equally well have been put in counterfactual terms: in agents who are fully rational and reasonable, beliefs are states that are true and well justified and desires are states that are satisfied. The relativist might then argue that, if we were to spell out these claims about the different and complementary directions of fit of belief and desire more fully, we would have no alternative but to conclude that principles something like MEANS–ENDS+ exhaust the principles of rationality and reasonableness governing the interactions of beliefs with beliefs and beliefs with desires. For though the idea that belief aims at the truth and desire aims at its own satisfaction provides us with ample reason to suppose that there are principles like MEANS–ENDS+ – this is because the satisfaction of intrinsic desires will be conditional on their combining with true beliefs about what can be done to satisfy them by giving rise to corresponding intentions (Smith 2004b) – it provides us with no reason at all to suppose that there are principles like UNIVERSALIZATION+ or REASONS+. In this way, the relativist might argue, the proper characterization of rationality or reasonableness in terms of MEANS–ENDS+ supports relativism about reasons for action. Unfortunately, however, if this is what we are doing when we come up with a characterization of rationality or reasonableness, then even though it is clear why we might end up thinking that the proper characterization of these notions supports relativism about reasons for action, the requirement that we come up with a characterization that is part of, or squares with, our scientific conception of the world doesn’t seem to play any role at all in the argument. Everything turns on what we can say a priori about belief and desire and what this entails. Moreover, though we have just seen how such a priori reflections might lead us to suppose that
240
m ic h a e l s m i t h
relativism about reasons for action is correct, the argument that was given for this conclusion turned crucially on the claim that all we can say about belief and desire a priori is that belief aims at the truth and desire aims at its own satisfaction, a claim with which absolutists disagree. So not only does the argument from the proper characterization of rationality or reasonableness to relativism about reasons for action have nothing to do with naturalism, it is also highly controversial. For example, some absolutists will disagree on the grounds that we can know a priori that desire is a state that aims at its own satisfaction in a lawlike way. When an agent has some intrinsic desire, and that intrinsic desire is satisfied, but the satisfaction of that intrinsic desire would be inconsistent with the satisfaction of other desires that are themselves suitably universal in their content, then that formal feature of the agent’s intrinsic desire is grounds for rational criticism of the intentions with which it coheres (compare Korsgaard 1996). They will therefore think that, if we were to fully spell out the claims that belief aims at the truth and desire aims at its own satisfaction in a lawlike way, then we would be forced to conclude that principles like MEANS–ENDS+ do not exhaust the principles of rationality or reasonableness, because there are additional principles like UNIVERSALIZATION+. Other absolutists will disagree on the grounds that we can know a priori that desire is like belief in being a “judgment-sensitive attitude,” that is, a state for which reasons can be given, where the considerations that are reasons for desiring are just like the considerations that are reasons for believing in having their status as reasons unconditionally, not conditionally on the presence of some desire (compare Scanlon 1998). These absolutists will think that, if we were to fully spell out the claims that belief aims at the truth and that desire aims at its own satisfaction while bearing in mind that both states are judgment-sensitive attitudes, then we would be forced to conclude that principles like MEANS–ENDS+ do not exhaust the principles of rationality, because there are additional principles like REASONS+. If either of these absolutists is right, then the proper characterization of rationality and reasonableness gives support to absolutism about reasons for action, not to relativism. What’s much more important for present purposes, however, is not the fact that absolutists could in this way claim to find support from the proper characterization of rationality and reasonableness, but that naturalism is thereby shown to be completely irrelevant to the issue that divides them from their relativist opponents. What divides them is what we can say a priori about belief and desire. This is what relativists and absolutists really disagree about.
Naturalism, absolutism, relativism
241
1 2 .5 T h i r d r e ply t o t h e f l at-f o o t e d r e s p ons e A good question to ask at this point is whether the answer just given to the question of what we are doing when we provide a characterization of rationality or reasonableness is one that a naturalist could give at all. After all, according to the answer just given, knowledge of the specific normative claims about what is rational or reasonable follow a priori from a proper understanding of belief and desire. But isn’t the idea that we can have such a priori knowledge itself a violation of the methodological principle that, in doing moral philosophy, “we must concentrate on finding the place of value and obligation in the world of facts as revealed by science?” Doesn’t it require us to suppose that there are irreducible normative relations that are discoverable a priori, facts about the ways in which belief and desire ought to interact (compare Scanlon 2009)? This is a fair question, but it is important to note that it is one to which the relativist, just as much as the absolutist, must have an answer. Anyone with naturalistic leanings will want to resist populating the world with unanalyzed normative relations. But I do not think that the account just given of what we are doing when we provide a characterization of rationality and reasonableness does require us to suppose that there are irreducibly normative relations. Instead I think that it requires us to suppose, plausibly, that belief and desire are best understood in functionalist terms. In order to illustrate how this helps answer the question, let’s fix on the relativist’s preferred normative characterization of belief and desire, the idea that belief is supposed to be true and desires are supposed to be satisfied. Robert Stalnaker spells out one way in which we might conceive of states with these aims in functionalist terms as follows. Belief and desire … are correlative dispositional states of a potentially rational agent. To desire that P is to be disposed to act in ways that would tend to bring it about that P in a world in which one’s beliefs, whatever they are, were true. To believe that P is to be disposed to act in ways that would tend to satisfy one’s desires, whatever they are, in a world in which P (together with one’s other beliefs) were true. (Stalnaker 1984: 15)
If some such functionalist account of belief and desire is along the right lines, then the differences between people with different beliefs and desires will amount to a difference in the functional roles that are played by the states that they are in when they are fully rational and reasonable. People may still have beliefs and desires, of course, even though they are not in states that are in fact playing these roles. It is just that in that case they will not count as being fully rational or reasonable.
242
m ic h a e l s m i t h
The attraction of this functionalist way of understanding belief and desire, and the correlative ideas of what it is to be fully rational and reasonable, is that it allows us to naturalize the quite general normative claims that we made initially in characterizing these psychological states. For it suggests that the proper way to understand these normative claims is in much the same way as we understand similar normative claims in the case of other functional kinds, claims such as “Clocks are supposed to tell the time,” or “Thieves aren’t supposed to get caught.” In each case what we have is a functional kind where things may be of that kind even though they don’t function optimally, and where, as a result, functioning optimally provides a standard against which the actual function of something of that kind can be measured. A slow clock is supposed to tell the time because clocks are objects of a functional kind such that, when an object is a perfectly functioning member of that kind, it does tell the time. The normative characterization of a clock – the claim that a clock is an object that is supposed to tell the time – thus isn’t an alternative to the functional characterization, but is rather just another way of stating the functional characterization. Similarly, when we say that beliefs are supposed to be true, this is just another way of saying that a belief is a state of a kind such that, when a state is a perfectly functioning member of that kind, it is true (contrast Zangwill 1998). The same is true of desire, mutatis mutandis. This way of interpreting the normative characterizations of belief and desire is thus maximally friendly to the methodological principle, for it suggests that nothing but states governed by causal relations in complex and interrelated ways – perfectly functioning beliefs and desires are still just beliefs and desires, after all – is required for there to be beliefs and desires that stand in just the normative relations that they do stand in to each other. A fully rational and reasonable agent is just someone in whom belief and desire play their functional roles optimally. Irreducible normative relations are therefore no more a part of the story of what it is to be an optimally functioning believer and desirer than they are a part of the story of what it is to be an optimally functioning clock or thief. The illustration just given has, of course, assumed that the relativist’s preferred normative characterization of the natures of belief and desire is correct. But the illustration is useful because it suggests that we could provide a similar functionalist interpretation of each of the absolutist’s preferred normative characterizations of the natures of belief and desire that we described earlier as well. This provides us with another way of getting at what the relativists and absolutists really disagree about, and
Naturalism, absolutism, relativism
243
what the two absolutists we described earlier disagree about as well. They disagree with each other about which beliefs and desires an agent would have, and the way that these states would interact with each other, if those agents were fully rational and reasonable: that is, optimally functioning. The relativist thinks that someone who is fully rational and reasonable may have an intention to harm and injure others, much as the professional criminal does. The absolutist thinks that no such intention would be possessed by such an agent. So here we have yet another reason to think that naturalism is irrelevant to the debate between relativists and absolutists. For the disagreement just described is a disagreement within the naturalist camp, a disagreement about what an optimally functioning believer and desirer is like, not a disagreement between naturalists and those who reject naturalism. 1 2 .6 F ou r t h r e ply t o t h e f l at-f o o t e d r e s p ons e This suggests a fourth line of reply to the flat-footed response to Harman’s argument that I have been running on behalf of the absolutists, and the last that I will consider here. Whether we are in fact in the sorts of functional states that are described by the relativist, or the sorts of functional states that are described by the absolutists, might be thought to be an empirical question, not a question whose answer is settled a priori. Harman might therefore suggest that the a priori questions on which we have focused are a side-show. The important questions, he might say, are all empirical. There are two points to make about this reply. The first is that, though it is evidently a contingent fact that we are believers and desirers – we might have suffered irreversible brain damage and been incapable of either believing or desiring – it seems to me that our knowledge that we are believers and desirers, if indeed we are, may well be a priori knowledge for all that. Perhaps a careful working through of what it is to have any knowledge at all shows that we are the sorts of beings who are in psychological states with the functional profiles of belief and desire. The second is that it is hardly credible to suppose, as the imagined reply does, that the answers to the a priori questions on which we have been focusing are a side-show. For we don’t want to know just any old thing about the professional criminal that explains why he is the cause of harm and injury to others, we want to know quite specific things, such as whether his harming and injuring others is itself the product of his intention to harm and injure them; whether he has that intention because of (say) ignorance or
244
m ic h a e l s m i t h
insensitivity to arguments; and, if it is, whether that ignorance or insensitivity is itself culpable because he had the capacity to be otherwise. The questions to which we seek answers are therefore questions that can only be framed in the terms whose nature we have been trying to clarify. (For the record, this is why it seems to me that we can safely ignore analyses of moral demands like Railton’s. Railton’s analyses in effect change the subject.) The upshot is that, even if it were ultimately an empirical question whether or not we are believers or desirers, the answer to this empirical question matters a great deal to us, and figuring out the answer to this empirical question would require both a priori reflection and empirical investigation. Indeed, even if the evidence were already in, and that evidence favored the relativist, rather than the absolutist – though, for the record, I haven’t seen or heard such evidence, and nor does Harman provide any – we would still need to know what to make of that evidence. Would the evidence show that only some humans have sufficient reasons to do or hope or wish for what morality demands? This would be the conclusion if the relativist had the correct a priori story. Or would it show that no humans have sufficient reasons to do anything at all? This would be the conclusion if the absolutist had the correct a priori story. To repeat, these are questions that can only be answered a priori by answering the sorts of questions that the relativists and the absolutists described earlier are attempting to answer, questions about the natures of belief and desire, rationality and reasonableness. Naturalism is irrelevant to the issue that divides them because these are questions about which naturalists can and do disagree.
Bibliography
Adams, R. 1999. Finite and Infinite Goods: A Framework for Ethics. New York: Oxford University Press. Aristotle. 2000. Nicomachean Ethics, trans. R. Crisp. Cambridge University Press. Armstrong, D. M. 1968. A Materialist Theory of the Mind. London: Routledge. Audi, P. (Unpublished MS). “Property Identity and Non-Causal Determination.” Audi, R. 1989. Practical Reasoning. London and New York: Routledge. 1993. “Ethical Naturalism and the Explanatory Power of Moral Concepts.” In S. Wagner and R. Warner (eds.), Naturalism: A Critical Appraisal. University of Notre Dame Press, 95–115. 1997. “Moral Judgment and Reasons for Action.” In G. Cullity and B. Gaut (eds.), Ethics and Practical Reason. Oxford University Press, 125–59. 1999. “Self-Evidence.” Philosophical Perspectives 13: 205–28. 2000. “Philosophical Naturalism at the Turn of the Century.” Journal of Philosophical Research 25: 27–45. 2001. The Architecture of Reason: The Structure and Substance of Rationality. Oxford University Press. 2002. “An Internalist Theory of Normative Grounds.” Philosophical Topics 29, 1/ 2: 19–46. 2004. The Good in the Right: A Theory of Intuition and Intrinsic Value. Princeton University Press. 2006. Practical Reasoning and Ethical Decision. London: Routledge. 2008. “Intuition, Inference, and Rational Disagreement in Ethics.” Ethical Theory and Moral Practice 11: 475–92. 2010. “Moral Perception and Moral Knowledge.” Proceedings of the Aristotelian Society, Supplementary Volumes 84: 79–97. Austin, J. L. 1956–57. “A Plea for Excuses.” Proceedings of the Aristotelian Society 57: 1–30. 1962. How to Do Things with Words. Oxford University Press. Ayer, A. J. 1956. The Problem of Knowledge. Harmondsworth, UK: Penguin Books. 1965. “On the Analysis of Moral Judgements.” Repr. in Philosophical Essays. London: Macmillan, 231–49. Bagnoli, C. 2011. “Constructivism.” The Stanford Encyclopedia of Philosophy. www. plato.stanford.edu/entries/constructivism-ethics (last accessed September 29, 2011). 245
246
Bibliography
Baker, M. 2001. Atoms of Language. New York: Basic Books. Ball, S. W. 1988. “Reductionism in Ethics and Science: A Contemporary Look at G. E. Moore’s Open Question Argument.” American Philosophical Quarterly 25: 197–213. Baron, J. 1992. “The Effect of Normative Beliefs on Anticipated Emotions.” Journal of Personality and Social Psychology 63: 320–30. 1994. “Nonconsequentialist Decisions.” Behavioral and Brain Sciences 17: 1–10. Bartels, D., and Medin, D. 2007. “Are Morally Motivated Decision Makers Insensitive to the Consequences of Their Choices?” Psychological Science 19: 24–28. Beardsley, M. 1981. Aesthetics: Problems in the Philosophy of Criticism, 2nd edn. Indianapolis: Hackett Publishing Company. Bechtel, W., and Richardson, R. 1998. “Vitalism.” In E. Craig (ed.), Routledge Encyclopedia of Philosophy, vol. ix. London: Routledge, 639–43. Bennett, K., and McLaughlin, B. 2005. “Supervenience.” In E. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Fall 2005 edn.), www.seop.leeds.ac.uk/ entries/supervenience/, accessed July 1, 2011. Bentham, J. 1988 [1789]. An Introduction to the Principles of Morals and Legislation. Buffalo, NY: Prometheus Books. Blackburn, S. 1984. Spreading the Word. Oxford: Clarendon Press. 1985. “Supervenience Revisited.” In Ian Hacking (ed.), Exercises in Analysis: Essays by Students of Casimir Lewy. Cambridge University Press, 47–67. 1993. Essays in Quasi-Realism. New York: Oxford University Press. 1998. Ruling Passions. Oxford: Clarendon Press. Bloomfield, P. 2001. Moral Reality. New York: Oxford University Press. Boghossian, P. 1994. “The Transparency of Mental Content.” Philosophical Perspectives 8: 33–50. Boyd, R. N. 1988. “How to Be a Moral Realist.” In G. Sayre-McCord (ed.), Essays in Moral Realism. Ithaca, NY: Cornell University Press, 182–217. 1993. “Metaphor and Theory Change: What Is ‘Metaphor’ a Metaphor For?” In A. Orotney (ed.), Metaphor and Thought, 2nd edn. Cambridge University Press, 481–533. 2003. “Finite Beings, Finite Goods: The Semantics, Metaphysics and Ethics of Naturalist Consequentialism, Part ii.” Philosophy and Phenomenological Research 66, 3: 505–53. Brandom, R. 1994. Making It Explicit. Cambridge, MA: Harvard University Press. Brandt, R. B. 1954. Hopi Ethics: A Theoretical Analysis. University of Chicago Press. 1979. A Theory of the Good and the Right. Oxford University Press. Brink, D. O. 1989. Moral Realism and the Foundations of Ethics. Cambridge University Press. 1992. “A Puzzle about the Authority of Morality.” In J. Tomberlin (ed.), Philosophical Perspectives: Ethics. Atascadero, CA: Ridgeview Publishing Company, 1–26.
Bibliography
247
1997. “Moral Motivation.” Ethics 108, 1: 4–32. Broad, C. D. 1971. Broad’s Critical Essays in Moral Philosophy, ed. David Cheney. London: George Allen & Unwin. Burge, T. 1996. “Our Entitlement to Self-Knowledge.” Proceedings of the Aristotelian Society 96: 91–116. 1998. “Reason and the First Person.” In C. Wright, B. C. Smith, and C. Macdonald (eds.), Knowing Our Own Minds. Oxford University Press, 243–70. Capitan, H., and Merrill, D. (eds.) 1967. Art, Mind, and Religion. Pittsburgh University Press. Carnap, R. 1935. Philosophy and Logical Syntax. London: Kegan Paul, Trench, Trubner & Co. 1950a. Logical Foundations of Probability. University of Chicago Press. 1950b. “Empiricism, Semantics, and Ontology.” Repr. in Meaning and Necessity. University of Chicago Press, 205–21. 1967 [1928]. The Logical Structure of the World. London: Routledge & Kegan Paul. Carritt, E. F. 1928. The Theory of Morals. London: Oxford University Press. 1937. “An Ambiguity of the Word ‘Good.’” Proceedings of the British Academy 23: 51–80. Chan, D. (ed.) 2008. Moral Psychology Today: Essays on Values, Rational Choice, and the Will. Berlin: Springer. Chomsky, N. 1972. Language and Mind, enlarged edn. New York: Harcourt Brace Jovanovich. Copp, D. 1995. Morality, Normativity, and Society. Oxford University Press. 2001. “Rationality, Autonomy, and Basic Needs.” In N. Roughley (ed.), Being Humans. Berlin: de Gruyter, 334–55. (ed.) 2006a. Oxford Handbook of Ethical Theory. New York: Oxford University Press. 2006b. “Introduction: Metaethics and Normative Ethics.” In Copp (ed.) 2006a, 3–35. 2007. Morality in a Natural World. Cambridge University Press. 2008. “Realist-Expressivism and Conventional Implicature.” In R. ShaferLandau (ed.), Oxford Studies in Metaethics, vol. iv. Oxford University Press, 167–202. 2009. “Toward a Pluralist and Teleological Theory of Normativity.” Philosophical Issues 19: 21–37. 2010a. “Normativity, Deliberation, and Queerness.” In R. Joyce and S. Kirchin (eds.), A World without Values: Essays on John Mackie’s Error Theory. Berlin: Springer, 141–65. 2010b. “The Wrong Answer to an Improper Question?” In S. Black and E. Tiffany (eds.), Reasons to Be Moral Revisited. Canadian Journal of Philosophy, Supplementary Volume 33: 97–130. Cullen, S. 2010. “Survey-Driven Romanticism.” Review of Philosophy and Psychology 1: 275–96.
248
Bibliography
Cuneo, T. 2007. The Normative Web. Oxford University Press. Cushman, F., Young, L. and Hauser, M. 2006. “The Role of Conscious Reasoning and Intuition in Moral Judgments: Testing Three Principles of Harm.” Psychological Science 17: 1082–89. Dancy, J. 2000. Practical Reality. Oxford University Press. 2004. Ethics without Principles. Oxford University Press. 2006. “Nonnaturalism.” In Copp (ed.) 2006a, 122–45. Danto, A. 1997. After the End of Art. Princeton University Press. Darley, J., and Schultz, T. 1990. “Moral Rules: Their Content and Acquisition.” Annual Review of Psychology 41: 523–56. Darwall, S. 1983. Impartial Reason. Ithaca, NY: Cornell University Press. 1992. “Internalism and Agency.” In J. Tomberlin (ed.), Philosophical Perspectives: Ethics. Atascadero, CA: Ridgeview Publishing Company, 155–74. Darwall, S. Gibbard, A., and Railton, P. 1992. “Toward a Fin de Siècle Ethics: Some Trends.” Philosophical Review 101: 115–89. Davidson, D. 1991. “What Is Present to the Mind?” Philosophical Issues 1: 197–213. Davidson, D., and Harman, G. (eds.) 1972. Semantics of Natural Language. Dordrecht, The Netherlands: D. Reidel. Doris, J. 2002. Lack of Character. Cambridge University Press. (ed.) 2010. The Moral Psychology Handbook. Oxford University Press. Doris, J., and Stich, S. 2005. “As a Matter of Fact: Empirical Perspectives on Ethics.” In F. Jackson and M. Smith (eds.), The Oxford Handbook of Contemporary Analytic Philosophy. Oxford University Press, 114–52. 2008. “Moral Psychology: Empirical Approaches.” In E. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Fall 2008 edn.), http://plato.stanford.edu/archives/win2008/entries/moral-psych-emp/, accessed July 1, 2011. Dreier, J. 2000. “Dispositions and Fetishes: Externalist Models of Moral Motivation.” Philosophy and Phenomenological Research 61, 3: 619–38. Druckman, J. 2001. “Evaluating Framing Effects.” Journal of Economic Psychology 22: 91–101. Dummett, M. 1959. “Truth.” Proceedings of the Aristotelian Society 59: 141–62. 1981. Frege: Philosophy of Language. Cambridge, MA: Harvard University Press. Dupré, J. 1993. The Disorder of Things: Metaphysical Foundations of the Disunity of Science. Cambridge, MA: Harvard University Press. Fagley, N., and Miller, P. 1987. “The Effects of Decision Framing on Choice of Risky versus Certain Options.” Organizational Behavior and Human Decision Processes 39: 264–77. Fine, K. 1982. “First-Order Modal Theories III – Facts.” Synthese 53: 43–122. Finlay, S. 2010. “Recent Work on Normativity.” Analysis 70: 331–46. Foot, P. 1972. “Morality as a System of Hypothetical Imperatives.” Philosophical Review 81, 3: 305–16. 2001. Natural Goodness. Oxford: Clarendon Press.
Bibliography
249
2002. “Morality as a System of Hypothetical Imperatives.” Reprint in Virtues and Vices. Oxford University Press, 157–73. Frankena, W. K. 1939. “The Naturalistic Fallacy.” Mind 48, 192: 464–77. Frisch, D. 1993. “Reasons for Framing Effects.” Organizational Behavior and Human Decision Processes 54: 399–429. Fumerton, R. 1990. Reason and Morality. Ithaca, NY: Cornell University Press. Gallie, W. 1955–56. “Essentially Contested Concepts.” Proceedings of the Aristotelian Society 56: 167–98. Garner, R. 2010. “Abolishing Morality.” In R. Joyce and S. Kirchin (eds.), A World without Values. Berlin: Springer, 217–33. Gibbard, A. 1990. Wise Choices, Apt Feelings. Cambridge, MA: Harvard University Press. Goldman, A. 1986. Epistemology and Cognition. Cambridge, MA: Harvard University Press. Greene, J. 2008a. “The Secret Joke of Kant’s Soul.” In Sinnott-Armstrong (ed.) 2008a, vol. iii, 35–79. 2008b. “Reply to Mikhail and Timmons.” In Sinnott-Armstrong (ed.) 2008a, vol. iii, 105–17. Haack, S. 1976. “The Pragmatist Theory of Truth.” British Journal for the Philosophy of Science 27: 231–49. Haidt, J. 2001. “The Emotional Dog and Its Rational Tail: A Social Intuitionist Approach to Moral Judgment.” Psychological Review 108: 814–34. Haidt, J., and Bjorklund, F. 2008. “Social Intuitionists Answer Six Questions about Moral Psychology.” In Sinnott-Armstrong (ed.) 2008a, vol. ii, 181–217. Haidt, J., and Hersh, M. 2001. “Sexual Morality: The Cultures and Emotions of Conservatives and Liberals.” Journal of Applied Social Psychology 31: 191–221. Hall, E. 1950–51. “The ‘Proof’ of Utility in Bentham and Mill.” Ethics 61: 66–68. Hare, R. D. 1993. Without Conscience: The Disturbing World of the Psychopaths among Us. New York: Pocket Books. Hare, R. M. 1952. The Language of Morals. Oxford University Press. Harman, G. 1963. “Generative Grammars without Transformation Rules: A Defense of Phrase Structure.” Language 39: 597–616. 1977. The Nature of Morality. New York: Oxford University Press. 1996. “Moral Relativism.” In G. Harman and J. J. Thomson, Moral Relativism and Moral Objectivity. Malden, MA: Blackwell, 1–64. 1999. Reasoning, Meaning, and Mind. Oxford: Clarendon Press. 2000a. “Is There a Single True Morality?” In Explaining Value and Other Essays in Moral Philosophy. Oxford: Clarendon Press, 77–99. 2000b. “Moral Agent and Impartial Spectator.” In Harman (2000a), 181–95. 2009. “Guilt-Free Morality.” In R. Shafer-Landau (ed.), Oxford Studies in Metaethics, vol. iv. Oxford University Press, 203–14. Hartshorne C. and Weiss, P. (eds.) 1934. Collected Papers of Charles Sanders Peirce, vol. v. Cambridge, MA: Harvard University Press.
250
Bibliography
Hauser, M. 2006. Moral Minds. New York: HarperCollins. Hauser, M., Cushman, F., Young, L. Jin, R., and Mikhail, J. 2006. “A Dissociation between Moral Judgment and Justification.” Mind and Language 22: 1–21. Hauser, M., Young, L., and Cushman, F. 2008. “Reviving Rawls’s Linguistic Analogy: Operative Principles and the Causal Structure of Moral Actions.” In Sinnott-Armstrong (ed.) 2008a, vol. ii, 107–43. Heller, Z. 1999. “Talk of the Town: The Bench.” The New Yorker, May 24: 29. Hinckfuss, I. 1987. The Moral Society: Its Structure and Effects. Discussion Papers in Environmental Philosophy 16. Canberra: Philosophy Program (RSSS), Australian National University. Hirstein, W. 2005. Brain Fiction: Self-Deception and the Riddle of Confabulation. Cambridge, MA: MIT Press. Horgan, T. 1993. “From Supervenience to Superdupervenience: Meeting the Demands of a Material World.” Mind 102: 555–86. Horgan, T., and Timmons, M. 1992. “Troubles for New Wave Moral Semantics: The ‘Open Question Argument’ Revived.” Philosophical Papers 21: 153–75. 2008. “What Does Phenomenology Tell Us about Moral Objectivity?” Social Theory and Policy 25: 267–300. 2009. “Analytical Moral Functionalism Meets Moral Twin Earth.” In Ravenscroft (ed.) 2009, 221–36. Hubin, D. 2001. “The Groundless Normativity of Instrumental Rationality.” Journal of Philosophy 98, 9: 445–68. Huemer, M. 2005. Ethical Intuitionism. Houndmills, UK: Palgrave Macmillan. Hume, D. 1885 [1748] . An Enquiry Concerning Human Understanding, ed. L. A. Selby-Bigge. Oxford University Press. 1955 [1748]. An Enquiry Concerning Human Understanding, ed. C. W. Hendel. Indianapolis: Bobbs-Merrill. 2000 [1739]. A Treatise of Human Nature, ed. D. Norton and M. Norton. Oxford University Press. Hursthouse, R. 1999. On Virtue Ethics. Oxford University Press. Jackson, F. 1992. “Critical Notice of S. Hurley, Natural Reasons.” Australasian Journal of Philosophy 70: 475–87. 1998. From Metaphysics to Ethics: A Defence of Conceptual Analysis. Oxford: Clarendon Press. 2003. “Cognitivism, A Priori Deduction, and Moore.” Ethics 3: 557–75. 2004. “Why We Need A-Intensions.” Philosophical Studies 118, 1/2: 257–77. 2005. “What Are Cognitivists Doing When They Do Normative Ethics?” Normativity, Philosophical Issues. Supplement to Noûs 15: 94–106. 2009. “Reply to My Critics.” In Ravenscroft (ed.) 2009, 387–474. Jackson, F., and Pettit, P. 1995. “Moral Functionalism and Moral Motivation.” Philosophical Quarterly 45, 178: 20–40. Johnston, M. 2010. Surviving Death. Princeton University Press. Jou, J., Shanteau, J., and Harris, R. 1996. “An Information Processing View of Framing Effects: The Role of Causal Schemas in Decision Making.” Memory and Cognition 24: 1–15.
Bibliography
251
Joyce, R. 2001. The Myth of Morality. Cambridge University Press. 2006. The Evolution of Morality. Cambridge, MA: MIT Press. 2007a. “Morality, Schmorality.” In P. Bloomfield (ed.), Morality and SelfInterest. Oxford University Press, 51–75. 2007b. “Moral Anti-Realism.” In E. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Fall 2007 edn.), http://plato.stanford.edu/entries/moral-antirealism/, accessed July 1, 2011. 2011a. “The Accidental Error Theorist.” In R. Shafer-Landau (ed.), Oxford Studies in Metaethics, vol. vi. Oxford University Press. 2011b. “The Error in ‘The Error in the Error Theory.’” Australasian Journal of Philosophy. Kahane, G. 2011. “Evolutionary Debunking Arguments.” 45 Nous 103–125. Kahneman, D. 1994. “The Cognitive Psychology of Consequences and Moral Intuition.” The Tanner Lecture in Human Values, University of Michigan, Ann Arbor, November. Kamm, F. M. 1996. Morality, Mortality, vol. ii. Oxford University Press. 1998. “Moral Intuitions, Cognitive Psychology, and the Harming-versus-NotAiding Distinction.” Ethics 108: 463–88. 2008. “Responses to Commentators on Intricate Ethics.” Utilitas 20: 111–42. Kant, I. 1948 [1785]. Groundwork of the Metaphysics of Morals, trans. H. J. Paton. London: Hutchinson. Kauppinen, Antti. 2007. “The Rise and Fall of Experimental Philosophy.” Philosophical Explorations 10: 95–118. Kim, J. 1984. “Concepts of Supervenience”. Philosophy and Phenomenological Research 45, 2: 153–76. Repr. in Supervenience and Mind. Cambridge University Press (1993), 53–78. King, J. 1998. “What Is a Philosophical Analysis?” Philosophical Studies 90: 155–79. Kohlberg, L. 1981. Essays in Moral Development, vol. i, The Philosophy of Moral Development. New York: Harper & Row. 1984. Essays in Moral Development, vol. ii, The Psychology of Moral Development. New York: Harper & Row. Korsgaard, C. 1996. The Sources of Normativity. Cambridge University Press. Kühberger, A. 1995. “The Framing of Decisions: A New Look at Old Problems.” Organizational Behavior and Human Decision Processes 62: 230–40. 1998. “The Influence of Framing on Risky Decisions: A Meta-Analysis.” Organizational Behavior and Human Decision Processes 75: 23–55. Kuhn, D. 1991. The Skills of Argument. Cambridge University Press. Kunda, Z. 1999. Social Cognition: Making Sense of People. Cambridge, MA: MIT Press. Ladd, J. 1957. The Structure of a Moral Code: A Philosophical Analysis of Ethical Discourse Applied to the Ethics of the Navaho Indians. Cambridge, MA: Harvard University Press. LeBoeuf, R., and Shafir, E. 2003. “Deep Thoughts and Shallow Frames: On the Susceptibility to Framing Effects.” Journal of Behavioral Decision Making 16: 77–92.
252
Bibliography
Levin, I., and Chapman, D. 1990. “Risk Taking, Frame of Reference, and Characterization of Victim Groups in AIDS Treatment Decisions.” Journal of Experimental Social Psychology 26: 421–34. Levin, I., Johnson, R., and Davies, M. 1987. “How Information Frame Influences Risky Decisions.” Journal of Economic Psychology 8: 43–54. Levine, J. 1983. “Materialism and Qualia: The Explanatory Gap.” Pacific Philosophical Quarterly 64: 354–61. Lewis, D. K. 1970. “How to Define Theoretical Terms.” Journal of Philosophy 67: 427–46. Repr. in Philosophical Papers, vol. i. Oxford University Press (1983), 78–95. 1989. “Dispositional Theories of Value. Part ii.” Proceedings of the Aristotelian Society, Supplementary Volumes 63: 113–37. Livengood, J., Sytsma, J., Feltz, A., Scheines, R., and Machery, E. 2010. “Philosophical Temperament.” Philosophical Psychology 23: 313–30. Loeb, D. 1998. “Moral Realism and the Argument from Disagreement.” Philosophical Studies 90, 3: 281–303. Loke, W. 1989. “The Effects of Framing and Incomplete Information on Judgments.” Journal of Economic Psychology 10: 392–41. Loke, W., and Lau, S. 1992. “Effects of Framing and Mathematical Experience on Judgments.” Bulletin of the Psychonomic Society 30: 393–95. Loke, W., and Tan, K. 1992. “Effects of Framing and Missing Information in Expert and Novice Judgment.” Bulletin of the Psychonomic Society 30: 187–90. Lowe, E. 2010. “Ontological Dependence.” In E. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Spring 2010 edn.), http://plato.stanford.edu/ archives/spr2010/entries/dependence-ontological/, accessed July 1, 2011. Ludwig, K. 2007. “The Epistemology of Thought Experiments: First Person versus Third Person Approaches.” Midwest Studies in Philosophy: Philosophy and the Empirical 31: 128–59. Macdonald, C., and Macdonald, G. 2010. “Introduction.” In Emergence in Mind. Oxford University Press, 1–21. McElroy, T., and Seta, J. 2003. “Framing Effects: An Analytic-Holistic Perspective.” Journal of Experimental Social Psychology 39: 610–17. Machery, E., Mallon, S., Nichols, S., and Stich, S. 2004. “Semantics, CrossCultural Style.” Cognition 92: B1–B12. MacIntyre, A. 1981. After Virtue. University of Notre Dame Press. Mackie, J. L. 1977. Ethics: Inventing Right and Wrong. Harmondsworth, UK: Penguin Books. Maley, C. Forthcoming. “Toward Clarifying Guilt and Shame.” Southern Journal of Philosophy. McLaughlin, B. 1997. “Emergence and Supervenience.” Intellectica 2: 25–43. McNaughton, D., and Rawling, P. 2003. “Naturalism and Normativity.” Supplement to the Proceedings of the Aristotelian Society 77: 23–45. McNeil, B., Pauker, S., and Tversky, A. 1988. “On the Framing of Medical Decision.” In D. Bell, H. Raiffa, and A. Tversky (eds.), Decision Making: Descriptive, Normative and Prescriptive Interactions. Cambridge University Press, 562–68.
Bibliography
253
Mikhail, J. 2007. “Universal Moral Grammar: Theory, Evidence and the Future.” Trends in Cognitive Sciences 11, 4: 143–52. 2008. “Moral Cognition and Computational Theory.” In Sinnott-Armstrong (ed.) 2008a, vol. iii, 81–91. Milgram, S. 1974. Obedience to Authority. New York: Harper & Row. Mill, J. S. 1979 [1861]. Utilitarianism. Indianapolis: Hackett Publishing Company. Miller, P., and Fagley, N. 1991. “The Effects of Framing, Problem Variations, and Providing Rationale on Choice.” Personality and Social Psychology Bulletin 17: 517–22. Miller, R. 1985. “Ways of Moral Learning.” Philosophical Review 94, 4: 507–56. Moore, G. E. 1903. Principia Ethica. Cambridge University Press. 1922a. “The Conception of Intrinsic Value.” In Philosophical Studies. London: Routledge & Kegan Paul, 253–75. 1922b. “The Nature of Moral Philosophy.” In Philosophical Studies. London: Routledge & Kegan Paul, 310–39. 1953. Some Main Problems of Philosophy. London: George Allen & Unwin. 1968. “Reply to My Critics.” In P. A. Schilpp (ed.), The Philosophy of G. E. Moore, 3rd edn. La Salle, IL: Open Court, 535–677. 1993a [1903]. “Preface to the Second Edition.” In Principia Ethica, rev. edn., ed. T. Baldwin. Cambridge University Press, 1–27. 1993b [1903]. Principia Ethica, rev. edn., ed. T. Baldwin. Cambridge University Press. Mormann, T. 2007. “Carnap’s Logical Empiricism, Values, and American Pragmatism.” Journal for the General Philosophy of Science 38: 127–46. Nagel, T. 1970. The Possibility of Altruism. Princeton University Press. 1974. “What Is It Like to Be a Bat?” Philosophical Review 83: 435–50. Nehamas, A. 2008. Only a Promise of Happiness: The Place of Beauty in a World of Art. Princeton University Press. Nichols, S. 2008. “Moral Rationalism and Empirical Immunity.” In SinnottArmstrong (ed.) 2008a, vol. iii, 395–407. Nisbett, R., and Wilson, T. 1977. “Telling More Than We Can Know: Verbal Reports on Mental Processes.” Psychological Review 84: 231–59. Norcross, A. 2008. “Off Her Trolley? Frances Kamm and the Metaphysics of Morality.” Utilitas 20: 65–80. Nozick, R. 1968. “Moral Complications and Moral Structures.” Natural Law Forum 13: 150. 1969. “Newcomb’s Problem and Two Principles of Choice.” In Nicholas Rescher (ed.), Essays in Honor of Carl G. Hempel. Dordrecht, The Netherlands: D. Reidel, 114–46. Nuccetelli, S., and Seay, G. (eds.) 2007a. Themes from G. E. Moore: New Essays in Epistemology and Ethics. Oxford University Press. 2007b. “What’s Right with the Open Question Argument?” In Nuccetelli and Seay (eds.), 261–82. O’Connor, T., and Wong, H. 2005. “The Metaphysics of Emergence.” Noûs 39: 658–78.
254
Bibliography
Olson, J. 2002. “Are Desires De Dicto Fetishistic?” Inquiry 45, 1: 89–96. Otsuka, M. 2008. “Double Effect, Triple Effect and the Trolley Problem: Squaring the Circle in Looping Cases.” Utilitas 20: 92–110. Paese, P., Bieser, M., and Tubbs, M. 1993. “Framing Effects and Choice Shifts in Group Decision Making.” Organizational Behavior and Human Decision Processes 56: 149–65. Parfit, D. 2011. On What Matters. Oxford University Press. Peirce, C. S. 1934a. “Practical and Theoretical Beliefs.” In Hartshorne and Weiss (eds.) 1934, 376–85. 1934b. “Judgment and Assertion.” In Hartshorne and Weiss (eds.) 1934, 385–87. Perry, R. B. 1970 [1926]. “Value as Any Object of Interest.” Excerpted in W. Sellars and J. Hospers (eds.), Readings in Ethical Theory. New York: Appleton-Century-Crofts, 138–51. Originally published in General Theory of Value. Cambridge, MA: Harvard University Press. Petrinovich, L., and O’Neill, P. 1996. “Influence of Wording and Framing Effects on Moral Intuitions.” Ethology and Sociobiology 17: 145–71. Pigden, C. R. 2007. “Desiring to Desire: Russell, Lewis and G. E. Moore.” In Nuccetelli and Seay (eds.) 2007, 244–60. Pinillos, N., Smith N., Nair, G., Marchetto, P., and Mun, C. 2011. “Philosophy’s New Challenge: Experiments and Intentional Action.” Mind and Language 26: 115–39. Pitcher, G. 1970. “Pain Perception.” Philosophical Review 74: 368–93. Pollock, J. 1995. Cognitive Carpentry: A Blueprint for How to Build a Person. Cambridge, MA: Bradford/MIT Press. Prichard, H. A. 2002. Moral Writings, ed. J. MacAdam. Oxford: Clarendon Press. Prinz, J. 2007. The Emotional Construction of Morals. Oxford University Press. 2008a. “Empirical Philosophy and Experimental Philosophy.” In J. Knobe and S. Nichols (eds.), Experimental Philosophy. Oxford University Press, 189–208. 2008b. “Resisting the Linguistic Analogy: A Commentary on Hauser, Young, and Cushman.” In Sinnott-Armstrong (ed.) 2008a, vol. ii, 157–70. 2008c. “Reply to Dwyer and Tiberius.” In Sinnott-Armstrong (ed.) 2008a, vol. i, 427–39. Putnam, H. 1967. “Psychological Predicates.” In Capitan and Merrill (eds.) 1967, 37–48. 1975. “The Meaning of ‘Meaning.’” Repr. in Mind, Language and Reality: Philosophical Papers, Volume 2. Cambridge University Press, 215–71. Quine, W. V. 1960. Word and Object. Cambridge, MA: MIT Press. 1975. “On Empirically Equivalent Systems of the World.” Erkenntnis 9: 313–28. 1981. Theories and Things. Cambridge, MA: Harvard University Press. 1986. “Reply to Roger F. Gibson, Jr.” In L. Hahn and P. Schilpp (eds.), The Philosophy of W. V. Quine. La Salle, IL: Open Court, 684–85. 1989. “Three Indeterminacies.” In R. Barrett and R. Gibson (eds.), Perspectives on Quine. Oxford: Basil Blackwell, 1–16. 1990. The Pursuit of Truth. Cambridge, MA: Harvard University Press.
Bibliography
255
Railton, P. 1986. “Moral Realism.” Philosophical Review 95: 163–207. 1992. “Some Questions about the Justification of Morality.” In James Tomberlin (ed.), Philosophical Perspectives: Ethics. Atascadero, CA: Ridgeview Publishing Company, 27–53. Ravenscroft, I. (ed.), 2009. Minds, Ethics, and Conditionals: Themes from the Philosophy of Frank Jackson. Oxford: Clarendon Press. Rawls, J. 1971. A Theory of Justice. Cambridge, MA: Harvard University Press. 1993. Political Liberalism. New York: Columbia University Press. Ridge, M. 2007. “Anti-Reductionism and Supervenience.” Journal of Moral Philosophy 4, 3: 330–48. 2008. “Moral Non-Naturalism.” In E. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Spring 2010 edn.), http://plato.stanford.edu/archives/spr2010/ entries/moral-non-naturalism/, accessed July 1, 2011. Roedder, E., and Harman, G. 2010. “Linguistics and Moral Theory.” In Doris (ed.) 2010, 272–95. Roemer, J. 1996. Theories of Distributive Justice. Cambridge, MA: Harvard University Press. Roorda, J. 1997. “Fallibilism, Ambivalence, and Belief.” Journal of Philosophy 94: 126–55. Rosenbaum, S. P. (ed.). 1995. The Bloomsbury Group. University of Toronto Press. Ross, L., and Nisbett, R. 1991. The Person and the Situation. Philadelphia, PA: Temple University Press. Ross, W. D. 1930. The Right and the Good. Oxford: Clarendon Press. 1939. Foundations of Ethics. Oxford University Press. 1954. Kant’s Ethical Theory. Oxford University Press. Rowe, W. 1993. Philosophy of Religion: An Introduction. Belmont, CA: Wadsworth. Russell, B. 1910. “Pragmatism.” In Philosophical Essays. London: Longman, Green, and Co., 87–126. Scanlon, T. 1998. What We Owe to Each Other. Cambridge, MA: Harvard University Press. 2009. “Being Realistic about Reasons.” John Locke Lectures, Oxford University, www.philosophy.ox.ac.uk/lectures/john_locke_lectures/past_ lectures, accessed November 20, 2010. Schilpp, P. (ed.) 1963. The Philosophy of Rudolf Carnap. La Salle, IL: Open Court. Schnider, A. 2008. The Confabulating Mind: How the Brain Creates Reality. Oxford University Press. Schroeder, M. 2005. “Realism and Reduction: The Quest for Robustness.” Philosophers’ Imprint 5,1 (February), http://quod.lib.umich.edu/cgi/t/text/ text-idx?c=phimp;rgn=main;idno=3521354.0005.001, accessed July 1, 2011. 2007. Slaves of the Passions. Oxford University Press. Schroeter, F., and Schroeter, L. 2003. “A Slim Semantics for Thin Moral Terms?” Australasian Journal of Philosophy 81: 191–207.
256
Bibliography
Searle, J. 1969. Speech Acts: An Essay in the Philosophy of Language. Cambridge University Press. Shafer-Landau, R. 2003. Moral Realism: A Defence. Oxford: Clarendon Press. 2005. “Error Theory and the Possibility of Normative Ethics.” In John Hawthorne (ed.), Philosophical Issues 15: Normativity. Oxford: Blackwell, 107–20. Shafir, E. 1998. “Philosophical Intuitions and Cognitive Mechanisms.” In M. Depaul and W. Ramsey (eds.), Rethinking Intuition. Lanham, MD: Rowman and Littlefield, 59–74. Sharp, F. C. 1928. Ethics. New York: The Century. Shaver, R. 2007. “Non-Naturalism.” In Nuccetelli and Seay (eds.) 2007, 283–306. Shiloh, S., Salton E., and Sharabi, D. 2002. “Individual Differences in Rational and Intuitive Thinking Styles as Predictors of Heuristic Responses and Framing Effects.” Personality and Individual Differences 32: 415–29. Sidgwick, H. 1907. The Methods of Ethics, 7th edn. London: Macmillan. 1998. Practical Ethics. New York: Oxford University Press. 2000. “The Distinction between ‘Is’ and ‘Ought.’” Repr. in Essays on Ethics and Method, ed. M. G. Singer. Oxford: Clarendon Press, 59–62. Sieck, W., and Yates, J. 1997. “Exposition Effects on Decision Making: Choice and Confidence in Choice.” Organizational Behavior and Human Decision Processes 70: 207–19. Singer, P. 2005. “Ethics and Intuitions.” Journal of Ethics 9: 331–52. Sinnott-Armstrong, W. 2006a. Moral Skepticisms. Oxford University Press. 2006b. “Moral Intuitionism Meets Empirical Psychology.” In T. Horgan and M. Timmons (eds.), Metaethics after Moore. New York: Oxford University Press, 339–66. (ed.) 2008a. Moral Psychology, 3 vols. Cambridge, MA: MIT Press. 2008b. “Framing Moral Intuitions.” In Sinnott-Armstrong (ed.), vol. ii, 47–76. 2008c. “How to Apply Generalities: Reply to Tolhurst and Shafer-Landau.” In Sinnott-Armstrong (ed.) 2008a, Vol. 2, 97–105. Skinner, B. F. 1971. Beyond Freedom and Dignity. New York: Alfred A. Knopf. Slote, M. 1979. “Assertion and Belief.” In J. Dancy (ed.), Papers on Language and Logic. Keele University Library, 177–90. Smart, J. J. C. 1959. “Sensations and Brain Processes.” Philosophical Review 68: 141–56. Smith, M. 1994. The Moral Problem. Oxford: Blackwell. 2000. “Moral Realism.” In H. LaFollette (ed.), The Blackwell Guide to Ethical Theory. Oxford: Blackwell, 15–37. 2004a. “Objectivity and Moral Realism: On the Significance of the Phenomenology of Moral Experience.” In Smith, M. (ed.) 2004 Ethics and the A Priori: Selected Essays on Moral Psychology and Meta-Ethics. Cambridge University Press, 234–58. 2004b. “Instrumental Desires, Instrumental Rationality.” Supplement to the Proceedings of the Aristotelian Society 78: 93–109.
Bibliography
257
2004c. “In Defense of ‘The Moral Problem’: A Reply to Brink, Copp, and SayreMcCord.” In Smith, M. (ed.) 2004 Ethics and the A Priori: Selected Essays on Moral Psychology and Meta-Ethics. Cambridge University Press, 259–96. Smith, S., and Levin, I. 1996. “Need for Cognition and Choice Framing Effects.” Journal of Behavioral Decision Making 9: 283–90. Sosa, E. 2007. “Experimental Philosophy and Philosophical Intuition.” Philosophical Studies 132: 99–107. Sperber, D. 1973. Le Structuralisme en anthropologie. Paris: Éditions du Seuil. Sperber, D., and Wilson, D. 1986. Relevance: Communication and Cognition. Oxford: Blackwell. Stalnaker, R. 1984. Inquiry. Cambridge, MA: MIT Press. Stanovich, K., and West, R. 1998. “Individual Differences in Framing and Conjunction Efforts.” Thinking and Reasoning 4: 289–317. 2008. “On the Relative Independence of Thinking Biases and Cognitive Ability.” Journal of Personality and Social Psychology 94: 672–95. Street, S. 2006. “A Darwinian Dilemma for Realist Theories of Value.” Philosophical Studies 127: 109–66. Sturgeon, N. 1985. “Moral Explanations.” In D. Copp and D. Zimmerman (eds.), Morality, Reason, and Truth. Totowa, NJ: Rowman and Allanheld, 49–78. 2003. “Moore on Ethical Naturalism.” Ethics 113: 528–56. 2006. “Ethical Naturalism.” In Copp (ed.) 2006a, 91–121. Swain, S., Alexander, J., and Weinberg, J. 2008. “The Instability of Philosophical Intuitions: Running Hot and Cold on Truetemp.” Philosophy and Phenomenological Research 76: 138–55. Takemura, K. 1994. “Influence of Elaboration on the Framing of Decision.” Journal of Psychology 128: 33–39. Thomson, J. J. 1976. “Killing, Letting Die, and the Trolley Problem.” The Monist 59: 204–17. 1986. Rights, Restitution, and Risk. Cambridge, MA: Harvard University Press. 1990. The Realm of Rights. Cambridge, MA: Harvard University Press. 2008a. Normativity. Chicago, IL: Open Court. 2008b. “Turning the Trolley.” Philosophy and Public Affairs 36: 359–74. Timmons, M. 1999. Morality without Foundations. Oxford University Press. Tversky, A., and Kahneman, D. 1986. “Rational Choice and the Framing of Decisions.” Journal of Business 59: 251–78. Unger, P. 1996. Living High and Letting Die. New York: Oxford University Press. Waldmann, M., and Dieterich, J. 2007. “Throwing a Bomb on a Person versus Throwing a Person on a Bomb.” Psychological Science 18: 247–53. Warnock, M. 1960. Ethics since 1900. London: Oxford University Press. Wedgwood, R. 2007. The Nature of Normativity. Oxford University Press. Weinberg, J., Nichols, S., and Stich, S. 2001. “Normativity and Epistemic Intuitions.” Philosophical Topics 29: 429–60.
258
Bibliography
West, H. 1997. “Mill’s ‘Proof’ of the Principle of Utility.” In D. Lyons (ed.), Mill’s Utilitarianism: Critical Essays. Lanham, MD: Rowman and Littlefield, 85–98. Wheatley, T. 2009. “Everyday Confabulation.” In W. Hirstein (ed.), Confabulation: Views from Neuroscience, Psychiatry, Psychology, and Philosophy. Oxford University Press, 203–22. White, A. 1958. G. E. Moore: A Critical Exposition. Oxford: Basil Blackwell. Willett, J. 1957. Brecht on Theatre. New York: Hill and Wang. Williams, B. 1966. “Consistency and Realism.” Proceedings of the Aristotelian Society, Supplementary Volumes 40: 1–22. 1985. Ethics and the Limits of Philosophy. Cambridge, MA: Harvard University Press. Williamson, T. 2000. Knowledge and Its Limits. Oxford University Press. Wilson, T. 2002. Strangers to Ourselves. Cambridge, MA: Belknap Press. Wolf, S. 1982. “Moral Saints.” Journal of Philosophy 79: 419–39. Wollheim, R. 1965. “Minimal Art.” Arts Magazine (January): 26–32. Wong, D. 2006. Natural Moralities: A Defense of Pluralistic Relativism. Oxford University Press. Wood, A. 2008. Kantian Ethics. Cambridge University Press. Woodward, J., and Allman. J. 2007. “Moral Intuition: Its Neural Substrates and Normative Significance.” Journal of Physiology – Paris 101: 179–202. Wright, C. 2005. “Intuition, Entitlement and the Epistemology of Logical Laws.” Dialectica 58: 155–75. Wynne, M. 2005. Emotional Experience and Religious Understanding. Cambridge University Press. Zangwill, N. 1998. “Direction of Fit and Normative Functionalism.” Philosophical Studies 91: 173–203. Zimmerman, M. 2001. The Nature of Intrinsic Value. Lanham, MD: Rowman and Littlefield.
Index
absolutism, 226, 228 Adams, R., 225n31 Allman, J., 205n19 analytical descriptivism, 132, 134n2, 141n6 analytical moral naturalism, 131 analytical naturalism, 2, 131, 132n1, 133–34, 138, 139n5, 140–41 anti-reductionism, 155, 157, 164, 167 argument from because, 43 Aristotle, 20 Armstrong, D. M., 86n9 Audi, R., 60n4, 169, 179n9, 186n17 Austin, J. L., 9, 100 autonomy of ethics, 73 Ayer, A. J., 71, 73n2
Carritt, E. F., 194n1, 195, 204–05, 208–10 Cartesian skepticism, 89 causal theory of reference, 79–80, 81n6, 82 Chan, D., 16n9 character traits, 9, 16, 22–23 Chomsky, N., 9, 16–18 cognitivism, 75–77, 80, 98n4, 99–100, 102–03 Copp, D., 3, 24n1, 29n12, 32n15, 37, 38n22, 39n23, 43n27, 51n36, 52n38, 56n42, 57n43, 110n1 Cornell realism, 171n2 Crisp, R., 3 Cullen, S., 198n8, 202n16 Cuneo, T., 4–5, 110, 111n6 Cushman, F., 196, 197n7, 209n23
Bagnoli, C., 177n8 Baker, M., 19 Ball, S., 139n5 Baron, J., 196n4, 209 Bartels, D., 196n4 Beardsley, M., 8 Bechtel, W., 26n3 Bellarmine, Robert, Cardinal, 14 Bentham, J., 132n1 Blackburn, S., 44n28, 64n15, 101, 147n8 Bloomfield, P., 79n5 Bloomsbury Group, 102–03 Boghossian, P., 138n4 Boyd, R., 79n5, 110n1, 123n25, 171n2, 175, 211n1, 221n25, 225n31 Brandom, R., 100 Brandt, R., 8n1, 179 Brink, D., 60, 79n5, 110n3, 171n2, 175, 211n2, 212n4, 217n18 Broad, C. D., 194n1, 195 Broome, J., 8 Burge, T., 138n4 Butler, Joseph, Bishop, 147
Dancy, J., 3, 25, 32n16, 33, 34n19, 35–36, 48n32, 49, 51n35, 52n39, 63n12, 146n6, 149n10, 153n16, 234 Danto, A., 8 Darley, J., 9 Darwall, S., 42n25, 142n8, 228 Davidson, D., 9, 10n6, 138n4 Dennett, D., 9 Dieterich, J., 196, 198 Doris, J., 9n4, 16, 23, 195n3, 200, 210 Dreier, J., 224n29 Druckman, J., 201 Dummett, M., 100 Dupré, J., 63n10
Carnap, R., 95–97, 98n4, 99, 101, 107n11
eliminativism, 1 emergentism, 4, 65–66 emotivism. See Ayer error theory, 1, 4, 42, 77, 89n1, 90n2, 91–92, 94–95, 97, 99, 100n7, 103, 105–06, 107n10, 111, 117n19, 122, 123–26 ethical naturalism and empirical test, 11, 13 evolutionary debunking, 15 experimental philosophy, 6, 88, 194–95, 204
259
260
Index
expressivism, 71, 75, 117, 128n28, 130, 147–48, 152, 154, See quasi-realism externalism, 211, 213 Fagley, N., 201 fetishism objection, 212, 214n9, 217n18 fictionalism, 128n28, 129–30 Fine, K., 59 Finlay, S., 63n9, 68n20 Fodor, J., 9 Foot, P., 12n7, 110n2, 112n8 Frankena, W., 142n8 Frisch, D., 201 Fumerton, R., 183n13 Galilei, G., 14 Gallie, W. B., 152n14 Garner, R., 106 Gauthier, D., 8 generative grammar, 9, 16–17 generative moral grammar, 17 Gibbard, A., 8, 142n8 Goldman, A., 183n14 Greene, J., 15, 196, 198n9 guilt, 16, 19–21, 23 Haack, S., 102 Haidt, J., 197, 199n10, 200, 207n21, 209n24, 210n25 Hall, E., 132n1 Halvorson, H., 10 Hare, R. M., 34 Harman, G., 3, 7, 10n6, 14, 19n11, 21n12, 62, 226–33, 235–38, 243–44 Hauser, M., 196n4, 197n7, 198n9, 202n16, 207, 210 Heller, Z., 214n8 Hinckfuss, I., 106 Hirstein, W., 104 Hopi, 8 Horgan, T., 63, 139n5, 141n6, 145n3, 146n6, 189n21 Hubin, D., 182n11 Huemer, M., 176n7 Hume, D., 13, 14n8, 173, 179, 184, 192, 212, 224n30 Humean moral naturalism, 111, 124 Hursthouse, R., 12n7 indeterminacy, 95–96, 99–100, 103, 107–08 internal grammar, 16 internalism, 211 Jackson, F., 4–5, 51n36, 52, 60, 70, 74n3, 81, 82n7, 86n10, 87, 131–32, 134n2, 135, 136n3, 138–39, 141n6, 142, 175n6
Johnston, M., 123n26 Jou, J., 201 Joyce, R., 4–5, 89n1, 90n1, 94, 106, 107n10, 111n5, 113n10, 114n12, 115n15, 116n17, 117n19, 118n20, 119, 121n22, 122–27, 128n28, 129n29, 130 Kahane, G., 15 Kahneman, D., 9, 199, 202 Kamm, F., 196n4, 197n5, 199n11, 201 Kant, I., 177–80 Kantianism, 87, 195 Katz, J., 9 Kauppinen, A., 198n8, 204n17 Keynes, J. M., 102 Kim, J., 64 King, J., 52n38, 53n40 Kohlberg, L., 9 Korsgaard, C., 228, 235, 240 Kuhberger, A., 201 Kuhn, D., 200 Kunda, Z., 23 Lau, S., 202 LeBoeuf, R., 201n15, 202 Leibniz’s law, 72–73 Levin, I., 201, 202 Levine, J., 63n12 Lewis, D., 86, 91–95, 97, 99, 103, 107n11, 108–09 Livengood, J., 200n13, 203 Loeb, D., 79n5 Loke, W., 201–02 Lowe, E., 67n19 Ludwig, K., 198n8, 204n17 Macdonald, C., 66 McElroy, T., 202 Machery, E., 195, 199 MacIntyre, A., 102–03 Mackie, J. L., 44n28, 74, 145n4, 147n8, 162 McLaughlin, B., 66 Maley, C., 21n12 McNaughton, D., 3, 25, 44–46 McNeil, B., 202 Medin, D., 196n4 Mikhail, J., 198n9 Milgram experiment, 22 Milgram, S., 22–23 Mill, J. S., 132n1, 142, 174, 176 Miller, G., 10 Miller, P., 201 Miller, R., 211n1 Molière [Jean-Baptiste Poquelin], 160
Index Moore, G. E., 1, 5–6, 27n7, 28, 36–37, 59–60, 70, 102, 120–21, 131, 132n1, 133–35, 138, 140, 141n7, 142, 145n1, 146–47, 171n2, 175n5, 176, 181 moral absolutism, 15 moral motivation, 20–21 moral realism, 24n1 moral relativism, 14–15, 226 moral supervenience, 2, 44n28, 63n11, 64–67, 69 Mormann, T., 97 motivation, 211, 216 Nagel, T., 26n4, 217n17, 228 natural properties, 25n2, 27n7, 29, 30n13, 37, 44, 45n31, 47–48, 50–53, 55, 70–72, 78–79, 82, 86, 169, 170, 174–77, 186n17, 189–90 naturalism and Cornell realism, 133 “hard”, 55 and moral realism, 1, 71 non-analytic, 3, 60, 133 and non-naturalism, 1, 24, 33 non-reductive, 74 in philosophy, 8, 10, 15 and relativism, 226, 229, 238, 244 society-centered, 38, 46, 51–52, 56 “soft”, 37, 55–56 utilitarian, 44 Naturalistic Reduction, 10 Nehamas, A., 8 network account, 4, 70, 78, 82–88 Nichols, S., 199 Nietzsche, F., 20 Nisbett, R., 22, 207n20 noncognitivism, 90n2, 96–97, 98n4, 99–103, 107, 175, 176n7, 177, 179 non-naturalism, 70, 133, 144–48, 150–53, 156, 163, 165–66 non-reductive realism. See anti-reductionism Norcross, A., 196n4 normative fact, 32–33, 37, 39, 44, 50n34, 54 normative functionalism, 11–12 normative naturalism, 24, 32, 37, 39, 44, 46, 49, 50–51, 53, 55–57 normative property, 25n2, 33, 37, 40, 44, 46–48, 50n34, 51, 53–54 Nozick, R., 8, 9n3 Nuccetelli, S., 131, 139n5 O’Connor, T., 66 O’Neill, P., 195, 196n4 Olson, J., 215n12, 224n30 open question argument, 1, 5, 60, 131–33, 134n2, 135–42, 145–46, 147n6 Otsuka, M., 196n4
261
Paese, P., 201 Parfit, D., 3, 24–25, 26n3, 27n8, 29n12, 35–37, 39, 40n24, 41, 42n25, 43–44, 45n31, 46–51, 52n39, 54–57, 61–62, 64, 65n17, 67, 69n21, 137, 146n6, 228 Parfit’s triviality objection, 49 Peirce, C. S., 100 Perry, R. B., 132 Petrinovich, L., 195, 196n4 Pettit, P., 8, 51n36, 82n7 philosophical naturalism, 1, 7 Piaget, J., 9 Pigden, C., 139n5 Pinillos, N., 198n8, 200 Pitcher, G., 164 Pollock, J., 9 Prichard, H. A., 194n1, 195, 204–06, 207n21, 208–10 Principle of Double Effect, 18 Prinz, J., 195n2, 196–97, 209n23, 210 projectibility, 73 Putnam, H., 9, 78, 79n5, 80 quasi-realism, 1 Quine, W. V., 107n11, 108–09 Railton, P., 110n3, 111n4, 112n8, 142n8, 211n1, 212n4, 228, 244 Ramsey sentences, 93 Rawling, P., 3, 25, 44–46 Rawls, J., 8, 9, 177–78 reductionism, 152n13, 158, 161–62 representationalism, 70, 73, 76, 78–84, 87 response dependent theories, 11, 13–14 Richardson, R., 26n3 Ridge, M., 2, 5, 59, 63n11, 142n8 Roemer, J., 28n10 Roorda, J., 108n12 Rosenbaum, S. P., 102 Ross, L., 22 Ross, W. D., 176, 194n1, 195, 198, 200, 203–06, 208–09, 210n26 Rowe, W., 120n21 Russell, B., 102 Scanlon, T., 228, 231, 236, 240–41 Schilpp, P., 98n4 Schnider, N., 104 Schroeder, M., 62n7, 110n3, 234 Schroeter, F., 154n17 Scruton, R., 214n8 Searle, J., 100 Seay, G., 131, 139n5 Sen, A., 8 Seta, J., 202
262 Shafer-Landau, R., 27n9, 29n12, 63n11, 69n21, 112n7, 147n7, 148n9, 149, 152, 153n16, 179n9 Shafir, E., 199, 201n15, 202 shame, 20 Sharp, F. C., 132 Shaver, R., 194n1 Shiloh, S., 202 Sidgwick, H., 60n3, 68, 194n1, 195, 205n19, 210n26 Sieck, W., 201–02 Singer, P., 15 Sinnott-Armstrong, W., 9n4, 16, 90n2 Skinner, B. F., 183 Slote, M., 100 Smart, J. J. C., 64n16, 86 Smith, A., 13, 14n8 Smith, M., 5–7, 43n26, 131–35, 136n3, 138–39, 142, 190n23, 212n5, 213, 214n9, 215–17, 218n21, 219–23, 224n29, 225–26, 228, 239 Smith, S., 202 social convention theories, 11, 13 Soft Naturalist’s Dilemma, 37, 55 Sosa, E., 198n8 Spencer, H., 132n1 Sperber, D., 8n1 Stalnaker, R., 241 Stanovich, K., 199n11, 201–02 Stich, S., 9, 16, 195n3, 200, 210 Street, S., 15 Sturgeon, N., 28n11, 29–30, 36, 62n8, 68–69, 145n2, 171n2, 175n5, 189n22 subjectivism, 41–43, 47, 54–56 supervenience, 11, 70, 72, 74, 86, 144–63, 166–67, 176 Swain, S., 195, 198, 200, 202n16 Takemura, K., 201 Tenenbaum, S., 6, 211, 213n6, 225n32
Index theory of structured propositions, 52n39 Thomson, J., 12n7, 18, 197n5, 202 Timmons, M., 121n22, 139n5, 141n6 trope nominalism, 149 Tversky, A., 9, 202 Unger, P., 196n4 utilitarian naturalism, 44–46, 48–51 utilitarianism, 80, 82, 87–88 virtue ethics, 12 Waldmann, M., 196, 198 Warnock, M., 132n1 Wedgwood, R., 5, 26n2, 27n6, 144, 153–67 Weinberg, J., 194n1, 195, 199–200 West, H., 132n1 West, R., 199n11, 201–02 Westermarck, E., 132n1 Wheatley, T., 104 Willett, J., 107 Williams, B., 100, 142n8 Williamson, T., 100 Wilson, T., 207n20 Wolf, S., 232 Wollheim, R., 8 Wong, D., 108n12 Wong, H., 66 Wood, A., 179n9 Woodward, J., 205n19 Wright, C., 68 Wynne, M., 187n19 Yates, J., 201, 202 Zangwill, N., 242 Ziff, P., 9 Zimmerman, M., 45n29