Putting Metaphysics First: Essays on Metaphysics and Epistemology [1 ed.] 0199576971, 9780199576975

Citation preview

Title Pages

Putting Metaphysics First: Essays on Metaphysics and Epistemology Michael Devitt

Print publication date: 2009 Print ISBN-13: 9780199280803 Published to Oxford Scholarship Online: February 2010 DOI: 10.1093/acprof:oso/9780199280803.001.0001

Title Pages (p.i) Putting Metaphysics First (p.ii) (p.iii) Putting Metaphysics First

(p.iv) Great Clarendon Street, Oxford OX2 6DP Oxford University Press is a department of the University of Oxford. It furthers the University's objective of excellence in research, scholarship, and education by publishing worldwide in Oxford New York Auckland Cape Town Dar es Salaam Hong Kong Karachi Kuala Lumpur Madrid Melbourne Mexico City Nairobi New Delhi Shanghai Taipei Toronto With offices in Argentina Austria Brazil Chile Czech Republic France Greece Guatemala Hungary Italy Japan Poland Portugal Singapore South Korea Switzerland Thailand Turkey Ukraine Vietnam Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries Published in the United States Page 1 of 2

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Title Pages by Oxford University Press Inc., New York © in this volume Michael Devitt 2010 The moral rights of the author have been asserted Database right Oxford University Press (maker) First published 2010 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this book in any other binding or cover and you must impose this same condition on any acquirer British Library Cataloguing in Publication Data Data available Library of Congress Cataloging in Publication Data Devitt, Michael, 1938– Putting metaphysics first : essays on metaphysics and epistemology / Michael Devitt. p. cm. Includes bibliographical references and index. ISBN 978–0–19–957697–5 (pbk. : alk. paper)—ISBN 978–0–19– 928080–3 (hardback : alk. paper) 1. Metaphysics. 2. Knowledge, Theory of. I. Title. BD111.D48 2009 110—dc22 2009039125 ISBN 978–0–19–928080–3 (hbk.) 978–0–19–957697–5 (pbk.) 10 9 8 7 6 5 4 3 2 1

Access brought to you by: Page 2 of 2

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Preface

Putting Metaphysics First: Essays on Metaphysics and Epistemology Michael Devitt

Print publication date: 2009 Print ISBN-13: 9780199280803 Published to Oxford Scholarship Online: February 2010 DOI: 10.1093/acprof:oso/9780199280803.001.0001

(p.v) Preface There are four recurring themes in this collection of papers. First, I make a sharp distinction between metaphysical issues and semantic issues. Second, I argue that metaphysical issues have a certain priority over epistemological and semantic issues. Third, I defend an epistemological naturalism that I take largely from Quine. And last but not least, I defend “realism about the external world”. Aside from these themes, I take a dim view of Plato's “One over Many” problem. I examine the metaphysics of “nonfactualism” and the metaphysics of truth. I defend moral realism. I take a stance on various realism issues in biology. I defend an intrinsic biological essentialism. I urge an evidential role for intuitions in philosophy as ordinary empirical judgments. Almost all of the chapters in this volume are, or are based on, fairly recent publications: seven, on publications in the last five years; all but two of the rest, on publications in the last ten or so years. It has taken me several years to put this collection together. The major cause of delay was the effort of writing Ignorance of Language (2006d). A minor cause was the problem of selecting from many papers on realism about the external world. The easy part was deciding not to include any paper, like ones on Dummett (Devitt 1983a), Putnam (Devitt 1983b, 1983c), and Rorty (Devitt 1988b), that had already appeared, near enough, as chapters in my book, Realism and Truth (1984/1991b). It was also obvious that I should focus on papers written since the book. The difficult part arose from the overlapping nature of my recent papers on this realism topic. (Invitations to write on a topic are few and far between before one has written a book on it but then they come flooding in!) I have addressed the overlap problem by omitting some papers altogether—for example, “A Naturalistic Defense of Realism” (1999a)—and including only parts of others—for example, “Underdetermination and Realism” (2002a) and “Worldmaking Made Hard: Rejecting Global Response Page 1 of 2

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Preface Dependency” (2004). Some small areas of overlap remain but they serve the purpose of making chapters fairly self‐contained and so I hope will not irritate the reader. Where a formerly published paper is included in full, I have left it basically unchanged, simply correcting typos and adding references where that seemed appropriate. If substantive changes were called for I have confined these to new footnotes, introduced by “[2009 addition]”, or to postscripts. (p.vi) This is the first of two collections of papers that Peter Momtchilof suggested I put together for OUP. I expect to start soon on the second, A Shocking Idea about Meaning and Other Essays in the Philosophy of Language and Linguistics.

Access brought to you by:

Page 2 of 2

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Acknowledgements

Putting Metaphysics First: Essays on Metaphysics and Epistemology Michael Devitt

Print publication date: 2009 Print ISBN-13: 9780199280803 Published to Oxford Scholarship Online: February 2010 DOI: 10.1093/acprof:oso/9780199280803.001.0001

(p.vii) Acknowledgements Chapter 1 is reprinted from “ ‘Ostrich Nominalism’ or ‘Mirage Realism’?”, Pacific Philosophical Quarterly, 61 (1980): 433–9, and is republished here by kind permission of Wiley Blackwell. Chapter 2 is reprinted from “Aberrations of the Realism Debate”, Philosophical Studies, 61 (1991): 43–63, and is republished here by kind permission of Springer. Chapter 3 is reprinted from most of part I of “Underdetermination and Realism”, in Ernest Sosa and Enrique Villanueva (eds.), Realism and Relativism: Philosophical Issues, 12 (2002), 26–50, and is republished here by kind permission of Wiley Blackwell. Chapter 4 is reprinted from “Scientific Realism”, in Frank Jackson and Michael Smith, (eds.), The Oxford Handbook of Contemporary Philosophy (2005), 767–91. Chapter 5 is reprinted from “Incommensurability and the Priority of Metaphysics”, in P. Hoyningen‐Huene and H. Sankey (eds.), Incommensurability and Related Matters (2001), 143–57, and is republished here by kind permission of Springer. Chapter 6 is largely a reprint of most of II.3–11.5 of “Worldmaking Made Hard: Rejecting Global Response Dependency”, Croatian Journal of Philosophy, 6 (2006): 3–25, and is republished here by kind permission of Kruzak. Chapter 7 is reprinted from “The Metaphysics of Nonfactualism”, in James E. Tomberlin (ed.), Philosophical Perspectives, 10, Metaphysics (Cambridge, Mass.: Blackwell Publishers, 1996), 159–76, and is republished here by kind permission of Wiley Blackwell.

Page 1 of 2

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Acknowledgements Chapter 8 is reprinted from “The Metaphysics of Truth”, in Michael Lynch (ed.), The Nature of Truth (2001), 579–611, and is republished here by kind permission of the MIT Press. Chapter 9 is reprinted from “Moral Realism: A Naturalistic Perspective”, Croatian Journal of Philosophy, 4 (2004): 1–15, and is republished here by kind permission of Kruzak. Chapter 10 is reprinted from “Natural Kinds and Biological Realisms”, in Joseph Campbell, Michael O'Rourke, and Matthew Slater (eds.), Carving Nature at its Joints: Topics in Contemporary Philosophy, vol. 8 (Cambridge, MA: MIT Press, forthcoming), and is republished here by kind permission of the MIT Press. (p.viii) Chapter 11 is reprinted from “Resurrecting Biological Essentialism”, Philosophy of Science, 75 (2008): 344–82, and is republished here by kind permission of Wiley Blackwell. Chapter 12 is reprinted from “Naturalism and the A Priori”. Philosophical Studies, 92 (1998): 45–65, and is republished here by kind permission of Springer. Chapter 13 is reprinted from “No Place for the A Priori”, in Michael J. Shaffer and Michael Veber (eds.), New Perspectives on A Priori Knowledge and Naturalism (Open Court, forthcoming), and is republished here by kind permission of Open Court. Chapter 14 was first delivered at the International Ontology Congress VI, “From the Gene to Language: the State of the Art”, held in San Sebastian in September 2004, and published in its proceedings: Victor Gomez Pin, Jose Ignacio Galparaso, and Gotzon Arrizabalaga (eds.), Ontology Studies Cuadernos de Ontologia: Proceedings of VI International Ontology Congress (San Sebastian, 2004) (2006). Reprinted with kind permission from Universidad del Pais Vasco. Chapter 15 is reprinted from “On Determining What There Isn't”, in Dominic Murphy and Michael A. Bishop (eds.), Stich and his Critics (Oxford: Wiley‐ Blackwell, 2009), 46–61, and is republished here by kind permission of Wiley Blackwell.

Access brought to you by:

Page 2 of 2

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Introduction

Putting Metaphysics First: Essays on Metaphysics and Epistemology Michael Devitt

Print publication date: 2009 Print ISBN-13: 9780199280803 Published to Oxford Scholarship Online: February 2010 DOI: 10.1093/acprof:oso/9780199280803.001.0001

Introduction Michael Devitt (Contributor Webpage)

DOI:10.1093/acprof:oso/9780199280803.003.0001

Abstract and Keywords This introductory chapter starts with an overall summary of the book, and then goes on to provide a summary of each chapter. Topics covered in the book include nominalism, realism, commonsense realism, scientific realism, moral realism, naturalism, biological essentialism, and nonfactualism. Keywords:   nominalism, commonsense realism, scientific realism, nonfactualism, moral realism, truth, biological realisms, biological essentialism, intuitions, naturalism

The volume has two parts, Part I on metaphysics, and Part II on epistemology. The chapters in the metaphysical part are largely concerned with realism issues. Chapter 1 is concerned with realism about universals, dismissing Plato's notorious “one over many” problem. Chapters 2 to 6 are concerned with realism about the external physical world, including the unobservable entities posited by science. My stance on that realism is fairly uncompromising and is a major theme of the book. In a survey twenty years ago, Heil noted the dominance of “anti‐realist tracts” but pointed out that “Australia, isolated and out of the loop evolutionarily, continues as stronghold of realists and marsupials” (1989: 65). Sadly, I think that the realists may be a little endangered. The sort of “nonfactualism” exemplified by moral noncognitivism and positivistic instrumentalism is mostly characterized in semantic terms. Yet, underlying that semantics there has to be a metaphysical antirealism. That antirealism is hard to characterize. Chapter 7 is my attempt. Deflationists are similarly antirealist about truth but their discussions are dominated by attention to the semantics of the truth term. Chapter 8 attempts to characterize the deflationist metaphysics Page 1 of 9

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Introduction of truth and argues briefly for a correspondence metaphysics. Chapter 9 argues for moral realism and Chapter 10 considers various realism issues in biology. Finally, Chapter 11 argues for the unfashionable view in the philosophy of biology that biological taxa have (partly) intrinsic essences. Confidence in the a priori does not seem to have been diminished by decades of failure to make any coherent sense of it. The epistemological part of the volume begins with two chapters, 12 and 13, opposing the a priori from a Quinean naturalistic perspective. That naturalism is another major theme of the book. From the naturalistic perspective the intuitions that so dominate the methodology of “armchair philosophy” cannot be a priori. Indeed, I argue in Chapter 14 that they are ordinary empirical judgments. (p.2) Finally, Chapter 15 addresses the problem of telling what entities do not exist. There are two other major themes of the book. One of these emphasizes the distinction between metaphysical issues about what there is and what it's like as against semantic issues about meaning, truth, and reference. The other is captured in the title of the volume, “Putting Metaphysics First”. We should approach epistemology and semantics from a metaphysical perspective rather than vice versa. We should do this because we know much more about the way the world is than we do about how we know about, or refer to, that world. The epistemological turn in modern philosophy, and the linguistic turn in contemporary philosophy, were something of disasters in my view. My view here reflects, of course, my epistemological naturalism. The metaphysics I want to put first is a naturalized one. I turn now to a more detailed summary of the book.

Part I: Metaphysics Chapter 1, “ ‘Ostrich Nominalism’ or ‘Mirage Realism’?” (1980), is concerned with the issue of realism about universals. In Nominalism and Realism (1978a), David Armstrong carefully demolishes various nominalist responses to Plato's famous “one over many” problem but simply dismisses the Quinean response as “Ostrich Nominalism”. In response I argue for three theses. (1) Plato's problem is pseudo. So to ignore it is not behave like an ostrich. Rather, to adopt realism because of this problem that isn't there is to be a “Mirage Realist”. (2) This problem is Armstrong's main reason for adopting realism but he is also motivated by some good reasons. So Armstrong is largely a Mirage Realist. (3) Quine does not ignore any real problem for nominalism and so he is not an Ostrich Nominalist. I start the postscript to this chapter by correcting a mistake in the original paper. The paper proposed a “semantic” criterion of ontological commitment that was not strictly Quine's and which invited a criticism it received in a response from Armstrong (1980) and in a response from Hugh Mellor and Alex Oliver (1997b). I go on to criticize these responses, discuss a version of the one Page 2 of 9

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Introduction over many proposed by David Lewis (1983), and, finally, diagnose the appeal of the one over many. Chapters 2 to 6 are all concerned with realism about the external physical world. Chapter 2 looks at the current realism debate. Chapter 3 argues for realism about the observable physical world of common sense and science; (p.3) Chapter 4, for realism about the unobservable world of science. Chapter 5 argues against incommensurability and constructivism; Chapter 6, against global response dependency. Chapter 2, “Aberrations of the Realism Debate” (1991a), finds much to object to in the current debate. Contrary to received opinion, I think that the issue of realism about the physical world has almost nothing to do with semantic issues about truth. This thesis of mine is perhaps what is most distinctive about my approach to the realism issue. The chapter begins by arguing for this thesis at some length. It follows from the thesis that it is an aberration to identify the two issues (Dummett 1977, 1978), to dismiss the realism issue out of hostility to correspondence truth (Rorty 1979; Fine 1986a, b), to think that the realism issue is one of interpretation, or to argue against realism by criticizing various claims about truth and reference (Putnam 1978; Laudan 1981). It is also an aberration to identify realism with nonskepticism (Margolis 1986), truth‐as‐the‐aim‐of‐ science (van Fraassen 1980), or scientific convergence (Blackburn 1980). Realism is an overarching metaphysical issue which should be settled, I argue, before any of these epistemological and semantic issues. (The latter argument is pursued in much more detail in Chapters 3 and 5.) The approach to realism exemplified in this chapter has not gone down well in some circles. I respond in the postscript. On my view, there are two dimensions to realism, independence and existence. Some think that the independence dimension must be understood in semantic terms (Williams 1993), others that the existence dimension must be (Appiah 1991). I argue against both views. Some think that my naturalistic approach is, in one way or another, “not philosophy” (Rorty 1979; Fine 1986a, b; Blackburn 1999; Dummett 2007; Putnam 2007). I respond. Chapter 3, “Underdetermination and Commonsense Realism”, based on “Underdetermination and Realism” (2002a), considers the underdetermination theses that lead to extreme skepticism. If true, these theses would count against nearly all our knowledge and hence undermine realism about the observable world of common sense and science. The traditional responses to these theses rest on a priori knowledge. I reject a priori knowledge (see Chapters 12 and 13). But even if there were such knowledge, these traditional responses tend to involve bizarre metaphysics and to be otherwise unsatisfactory. Instead, I start with a Moorean response: realism is much more firmly based than the epistemological theses that are thought to undermine it. The Moorean response Page 3 of 9

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Introduction is supported by a naturalistic one that appeals to scientific practice. Rather than proceeding from an a priori epistemology to an a priori metaphysics, we should proceed from an (p.4) empirical metaphysics to an empirical epistemology. We should put metaphysics first. Chapter 4, “Scientific Realism” (2005c), starts by arguing, in keeping with Chapter 2, that the central issue of scientific realism is a metaphysical one about the existence and nature of the unobservables of science. I go on to consider the arguments for and against this realism. The most famous argument for is from the success of science. This and other related explanationist arguments rest on abduction. Critics reject abduction but it is not clear that they are right to do so: abduction seems on a fairly equal footing with other ampliative methods of inference that opponents of scientific realism who are not total skeptics must accept. The first argument against to be considered is that from underdetermination. It starts from the claim that any theory faces many empirically equivalent rivals. An assessment of this argument depends on careful attention to “empirical equivalence”. Once we note the role of auxiliary hypotheses in testing theories, and our ability to create evidence in novel experiments, the argument collapses. The second argument against to be considered is the pessimistic meta‐induction: “Past theories have been mostly wrong; so probably present theories are too”. Realist defenses against this powerful argument are explored. To sustain the argument, the historical details have to show not only that we were nearly always wrong in our unobservable posits but that, despite methodological improvements, we have not been getting significantly righter. It seems to me most unlikely that this case can be made. The discussion of the argument from underdetermination takes a liberal view of “possible evidence”. The postscript to this chapter starts by considering the consequences of taking the more restricted view of W. V. Quine (1970) and Bas van Fraassen (1980). The underdetermination argument still fails. The Postscript goes on with some thoughts on what is required for an abduction to be good. It concludes by rejecting Kyle Stanford's interesting argument from “unconceived alternatives” (2006), a new version of the pessimistic meta‐induction. Chapter 5, “Incommensurability and the Priority of Metaphysics” (2001a) was largely prompted by Paul Hoyningen‐Huene's (1993) admirable presentation and defense of the views of Thomas Kuhn (1970). The chapter aims to reject a semantic doctrine, “Incommensurability”, commonly attributed to Kuhn and to Paul Feyerabend (1981a, b). Kuhn and Feyerabend also subscribe to the neo‐ Kantian metaphysical doctrine of “Constructivism” according to which we make “phenomenal” worlds with our theories. Different theories make different worlds; so the worlds we live in are theory‐relative. (p.5) Constructivism is Page 4 of 9

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Introduction clearly opposed to the “Realism” I have defended in earlier chapters. I argue that the Incommensurability issue rests on the Realism issue. In discussing the Realism issue I reject four arguments for Constructivism. Two Kantian arguments make the mistake criticized in Chapters 2 and 3 of not putting metaphysics first: rather they proceed from an a priori epistemology or semantics to an a priori metaphysics. Two arguments by Paul Hoyningen‐Huene (1993) and his co‐authors support relativism but do nothing to support the Kantian core of Constructivism. Without Constructivism, Incommensurability is left unsupported. I conclude by arguing against “meta‐incommensurability”. The postscript looks critically at Michael Lynch's recent constructivist doctrine, “pluralism” (1998). Lynch recognizes the implausibility of full‐blown Constructivism and urges a more moderate doctrine according to which what we all share is not a “noumenal” world but rather a minimal phenomenal world. Lynch says interestingly little in support of his doctrine and what he does say, I argue, is extremely unconvincing. The Realism argued for in Chapters 2 to 5 forms the background to Chapter 6, “Global Response Dependency and Worldmaking”, based on “Worldmaking Made Hard: Rejecting Global Response Dependency” (2004). “Worldmaking” is Constructivism without the commitment to relativism. The chapter considers the relation between Worldmaking and the global response dependency theory of concepts urged by Philip Pettit (1990, 1991). According to Pettit, all concepts are of dispositions to produce a certain sort of response in normal humans in normal conditions. Pettit denies that this theory leads to Worldmaking. I argue that he is wrong. The theory leads to the view that all properties are response‐dependent and this leads to Worldmaking. For that reason alone, Pettit's theory of concepts should be rejected. This ends my discussion of realism about the external physical world. In the remaining chapters in Part I, I consider some other metaphysical issues: nonfactualism, the nature of truth, moral realism, biological realisms, and biological essentialism. Chapter 7, “The Metaphysics of Nonfactualism” (1996b), finds the usual characterizations of nonfactualism unsatisfactory. The problem partly comes from focusing on nonfactualism's special semantics instead of on the antirealist metaphysics that must motivate that semantics. The problem also comes from the genuine difficulty in characterizing this metaphysics because nonfactualism goes along with many realist utterances, claiming to be able to interpret them in a special way. I reject the usual implicit characterizations of the metaphysics: that there are no properties or facts in the problematic area. Using the examples of instrumentalism, moral noncognitivism, and deflationary truth, (p.6) I argue for a general method for characterizing this metaphysics: make precise the idea that, in the problematic area, there is no reality with a nature to be explained Page 5 of 9

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Introduction and with a causal‐explanatory role. There should always be some uncontroversially factual language in which to state this rejection of the problematic reality. Of course, the rejection may seem implausible, but that is the price that nonfactualism must pay for its motivation. Finally, I turn to the special semantics of nonfactualism, rejecting accounts of this in terms of properties, facts, and truth conditions, but accepting ones that contrast the apparently descriptive or factual function of indicative sentences in the problematic area with their alleged function as expressive, prescriptive, or whatever. Chapter 8, “The Metaphysics of Truth” (2001b), has two parts. Part I is concerned with the difference between deflationary and correspondence truth. I argue that this difference remains unclear for a variety of reasons. In particular, there has been insufficient attention to the distinction between the metaphysics of truth and the linguistics of the truth term and hence to what deflationary theories say, or should say, about that metaphysics. Indeed, there has been a pervasive use/mention sloppiness in discussions of truth and the truth term. Emphasizing the similarity between deflationism and the sort of “nonfactualism” discussed in Chapter 7, I argue that the metaphysics of deflationism should reject the need for and possibility of explaining the nature or causal role of truth. It is largely because of this metaphysics that deflationism rejects a standard semantics and a descriptive role for the truth term. A rather shocking consequence of my discussion is that Tarski's famous theory is not really a theory of truth at all. It is a theory of truth terms. In part II, I argue, from a naturalistic perspective, that the case for correspondence truth over deflationism is strong provided we can explain reference. If we cannot explain reference then we should adopt deflationism. The very heavy price for this would be eliminativism about meaning. This price seems so heavy that surely we should be optimistic about explaining reference. Chapter 9, “Moral Realism: A Naturalistic Perspective” (2002c), starts by addressing the question, “What is moral realism?” I reject standard answers in terms of truth and meaning (Sayre‐McCord 1988c; Railton 1996). This is in keeping with my stand, in Chapter 2, on realism about the physical world. These standard answers are partly motivated by the phenomenon of noncognitivism. Noncognitivism does indeed cause trouble for a straightforwardly metaphysical answer, as noted in Chapter 7, but still such an answer can be given. The chapter goes on to consider the question, “Why believe moral (p.7) realism?” I argue that it is prima facie plausible and its alternatives are not. But what about the arguments against moral realism? I look critically at the argument from “queerness”, the argument from relativity (Mackie 1988), the argument from explanation, and epistemological arguments (Harman 1988). But there is a major

Page 6 of 9

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Introduction worry for moral realism: how can it be accommodated in a naturalistic world view? I offer some brief and inadequate remarks in response to this question. Chapter 10, “Natural Kinds and Biological Realisms” (2009e), examines some interesting “realism” issues in biology, arguing that these are best seen as being about which kinds are explanatorily significant, which ones are “natural kinds”. Seeing them as “realism” issues has caused unclarity and confusion. The main issue discussed is that between Ereshefsky's “pluralistic antirealism” about species (1998) and Kitcher's “pluralistic realism” (1984). This is a “category” issue about the significance of a kind being a species not a “taxon” issue about the significance of an organism being a member of a certain species. I argue that, so far as “realism” is concerned, the views of Ereshefsky and Kitcher are actually the same. The views differ only in that Ereshefsky's pluralism is more conservative than Kitcher's. I argue for Ereshefsky's view. The discussion then turns to the higher taxa and categories. I dismiss the view that the higher taxa “do not exist in the same sense as do species” (Eldredge and Cracraft 1980) and that “there is no such thing as a reptile” (Sterelny and Griffiths 1999: 197). And I doubt the cladistic view that only monophyletic groups are “real”, that is to say, natural kinds. However, the signs are that the higher categories are not natural kinds and hence that the Linnaean hierarchy should be abandoned. Chapter 11, “Resurrecting Biological Essentialism” (2008b), defends the doctrine that Linnaean taxa, including species, have essences that are, at least partly, underlying intrinsic, mostly genetic, properties. The consensus among philosophers of biology is that such essentialism is deeply wrong, indeed incompatible with Darwinism. I argue that biological generalizations about the morphology, physiology, and behavior of species require structural explanations that must advert to these essential properties. The chapter is mostly devoted to rejecting two common sorts of objection to intrinsic essentialism. First, I reject the objection that, according to current “species concepts” (theories of species), species are relational. The species concepts are concerned with the “category” problem of what it is for a kind to be a species and throw little light on the “taxon” problem of what it is for an organism to be a member of a particular kind. So they are largely irrelevant to the essentialism issue. I tentatively diagnose the error as coming from a conflation of the taxon and category problems. This conflation is encouraged by the mistaken idea that the (p.8) species concepts provide an answer to the conspecificity problem, the problem of saying in virtue of what organisms are in the same species. Second, the chapter argues that, contrary to what is frequently claimed, this essentialism can accommodate features of Darwinism associated with variation and change. In particular, essentialism can accept gradual change and a certain indeterminacy about species.

Page 7 of 9

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Introduction Part II: Epistemology In Coming to Our Senses (1996a) I argue briefly against a priori knowledge on two grounds: first, confirmation holism removes any strong motivation for thinking that mathematics and logic are immune from empirical revision; second, the idea of a priori knowledge is deeply obscure, as the history of failed attempts to explain it show. Chapter 12, “Naturalism and the A Priori” (1998), defends this position from Georges Rey's argument for a reliablist a priori (1993, 1997, 1998) and Hartry Field's for an a priori logic (1996, 1998). I argue that Rey has not explained a way of knowing at all, hence not an a priori one: he has not shown how the beliefs reliably generated by his “logical sub‐system” are epistemically nonaccidental. The dominant idea of Field's argument is that logic must be seen as a priori because we need logic to get evidence for anything. I give a reason for thinking that this idea is “fishy”: an evidential system can undermine itself. Chapter 13, “No Place for the A Priori” (2009c), is a detailed argument for what was briefly argued in Coming to Our Senses: all knowledge is empirical —“justified by experience”—and hence there is no place for the a priori. This stands opposed to the view that there is a nonempirical method of justifying beliefs but not to the view that there is a nonempirical source of beliefs, not to the view that some beliefs are innate. If any beliefs are innately justified, their justification must come somehow from the experiences (broadly construed) of our distant ancestors. I argue (i) that we do not need the a priori and (ii) that we cannot have it. Concerning (i) the focus is on the view, briefly considered in Chapter 12, that logic must be seen as a priori because we need logic to get evidence for or against anything. One response, inspired by the image of Neurath's boat, hopes to justify each logical principle in turn using other principles. But this seems likely to be a vain hope. Another response argues that logic can be seen as empirical if rule‐circular arguments are allowed. And apriorists cannot disallow such arguments because any justification of a priori reasoning would have to be rule‐circular. (p.9) Concerning (ii), I argue that the whole idea of the a priori is unexplained and deeply obscure. Traditional attempts at explanation that appeal to analyticity fail in two ways. They rest on an unexplained acceptance of logical truths and on the mistaken view that competence with a concept is sufficient for knowledge about it. A consideration of the contemporary views of George Bealer (1992, 1999), Laurence BonJour (1998), and Christopher Peacocke (2005) helps to bring out the obscurity of the a priori. Intuition mongering is common in the theory of reference and in philosophy generally. Why is this appropriate? And why is it appropriate for linguists to take intuitions as the main evidence for a grammar. The Chomskian answer to the Page 8 of 9

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Introduction latter question is that the intuitions are derived by a rational process from a representation of linguistic principles in the mind. Stephen Stich (1996) has suggested (although not endorsed) an analogous answer to the question about referential intuitions. In Chapter 14, “Intuitions” (2006c), I take a different view. I argue for a naturalistic and non‐Cartesian view of intuitions in general. Intuitions are empirical central‐processor responses to phenomena differing from other such responses only in being immediate and fairly unreflective. Applying this view to linguistic and referential intuitions yields an explanation of their evidential role without any appeal to the representation of rules. And the view yields a naturalistic view of the characteristic method of “armchair philosophy”. In “Deconstructing the Mind”, Stephen Stich (1996) rightly points out that typical arguments for eliminativism about the mind—including his own earlier ones—presuppose a description theory of reference. But, he notes, other theories of reference are possible; for example, a causal‐historical theory. So the issue of eliminativism seems to come down to that of the theory of reference. Stich argues for a gloomy conclusion about this theory, whether taken as concerned with folk semantics or as proto‐science. This yields a gloomy conclusion about the importance and interest of the eliminativism issue and, indeed, of realism issues in general. What has gone wrong? Stich thinks that the mistake came in the “semantic ascent” at the beginning, in looking to the theory of reference to settle metaphysical/ontological questions. In Chapter 15, “On Determining What There Isn't” (2009b) I argue that Stich is absolutely right about that, but for the wrong reasons. The folk‐semantic view of the theory of reference is a nonstarter, despite its popularity. The theory of reference should be taken as proto‐science and divided into two tasks: characterizing reference and showing that reference has a scientific role. Then we should not share Stich's gloom about it. Still, an appropriately modest view of the accomplishments of this proto‐science goes decisively against the (p.10) semantic ascent approach to ontological issues. How then are we to settle ontological issues? I think that Stich's answer is very much along the right lines. We do not have any principles adequate to help us with the difficult cases and it may be that some of these are indeterminate. But this is no excuse for constructivism.

Access brought to you by:

Page 9 of 9

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

“Ostrich Nominalism” or “Mirage Realism”? *

Putting Metaphysics First: Essays on Metaphysics and Epistemology Michael Devitt

Print publication date: 2009 Print ISBN-13: 9780199280803 Published to Oxford Scholarship Online: February 2010 DOI: 10.1093/acprof:oso/9780199280803.001.0001

“Ostrich Nominalism” or “Mirage Realism”? * Michael Devitt (Contributor Webpage)

David Armstrong's approach to “the problem of universals” has a contemporary gloss: he leaves it to “total science . . . to determine what universals there are.” Nevertheless his conception of the problem shows him to be a devotee of the “old‐time” metaphysics. The problem is the traditional one allegedly posed by the premise of “Platos One over Many argument”: “Many different particulars can all have what appears to be the same nature” (p. xiii).1 It is a pity that Armstrong takes no serious account of the “new” metaphysics of W. V. Quine and others according to which there is no such problem as Armstrong seeks to solve.2 In my view this Quinean position is a much stronger rival to Armstrong's Realism about universals than the many others he carefully demolishes. The universals we are concerned with here are properties (what Quine calls “attributes”) and relations. “Realists” believe in them, “Nominalists” don't. After outlining five versions of Nominalism, Armstrong mentions the Quinean position as a possible sixth under the title “Ostrich or Cloak‐and‐dagger Nominalism”: I have in mind those philosophers who refuse to countenance universals but who at the same time see no need for any reductive analyses of the sorts just outlined. There are no universals but the proposition that a is F is perfectly all right as it is. Quine's refusal to take predicates with any ontological seriousness seems to make him a Nominalist of this kind.(p. 16) Worse, these philosophers are guilty of trying to have it both ways: denying universals whilst, prima facie, unashamedly making use of them. They commit the (p.14) sin of failing to answer “a compulsory question in the examination paper” (p. 17). In Quinean language, they fail to face up to their ontological commitments.

Page 1 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

“Ostrich Nominalism” or “Mirage Realism”? * Ostriches are reputed to ignore problems by putting their heads in the sand. Mirages are another feature of desert life: people see things that aren't there. An “Ostrich Nominalist” is a person who maintains Nominalism whilst ignoring a problem. A “Mirage Realist” is a person who adopts Realism because he sees a problem that isn't there. My major thesis is as follows: 1. To maintain Nominalism whilst ignoring the One over Many argument is not to be an Ostrich Nominalist; rather to adopt Realism because of that argument is to be a Mirage Realist. Establishing this thesis would not, of course, show Realism to be unjustified (let alone false): there might be problems independent of the One over Many argument for which Realism is a possible solution. Armstrong thinks there are. I agree. To the extent that he is responding to those problems he is not a Mirage Realist. My thesis about him is as follows: 2. Armstrong is largely though not entirely a Mirage Realist. Correspondingly, a Nominalist could be an Ostrich by putting his head in the sand as real problems loom. However correct his stand on the One over Many argument he could otherwise commit the sin that Armstrong complains of. I don't know whether there are any Ostrich Nominalists, but the only philosopher Armstrong alleges (tentatively) to be one, Quine, is not:

3. Quine is not an Ostrich Nominalist.

Argument for Thesis 1 According to Armstrong, the problem posed by the One over Many argument is that of explaining “how numerically different particulars can nevertheless be identical in nature, all be of the same ‘type’ ” (p. 41). What phenomena are supposed to need explaining here? I take it that what Armstrong is alluding to is the common habit of expressing, assenting to, and believing, statements of the following form: (1) a and b have the same property (are of the same type), F‐ness. To settle ontological questions we need a criterion of ontological commitment. Perhaps Quine's criterion has difficulties, but something along that line is mandatory. The key idea is that a person is committed to the existence of (p.15) those things that must exist for the sentences he accepts to be true. What must exist for a given sentence to be true is a semantic question to which our best theory may give no answer in which we have confidence. Furthermore the sentence may, by its use of quantifiers or singular terms, suggest an answer which the person would want to resist. Hence, in my view, the importance of Quine's mention of paraphrase in this context. Suppose the given sentence seems to require for its truth the existence of G's yet the person can offer another sentence, which serves his purposes well enough, and which is known not to have that requirement. This is known because our semantic theory can be Page 2 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

“Ostrich Nominalism” or “Mirage Realism”? * applied to this other sentence, in a way that it cannot to the given sentence, to show that the sentence can be true even though G's do not exist. We can then say that the person's apparent commitment to G's in the given sentence arises from “a mere manner of speaking”; he is not really committed to them. Now in the ordinary course of conversation a Quinean is prepared to express or assent to the likes of (1). (1) seems to require the existence of an F‐ness for it to be true. So he appears committed to that existence. To this extent the One over Many argument does pose a problem to the Quinean Nominalist, but it is a negligible extent. He has a suitable paraphrase readily to hand: (2) a and b are both F. When the ontological chips are down, he can drop (1). There is no problem about identities in nature beyond a trivial one of paraphrase.

Armstrong will not be satisfied by this, of course: “You have simply shifted the problem. In virtue of what are a and b both F?” The Quinean sees only a trivial problem here too. It is in virtue of the following: (3) a is F; (4) b is F. Armstrong will still be dissatisfied: “In virtue of what is a (or b) F?” If the One over Many argument poses a problem it is this. That was historically the case and, though Armstrong always states the problem in terms of identities in nature, it is the case for him too.3 If there is no problem for the Nominalist in (3) and (4) as they stand then he has an easy explanation of identities in nature.

The Realist who accepts the One over Many problem attempts to solve it here by claiming the existence of a universal, F‐ness, which both a and b have. The Nominalist who accepts the problem attempts to solve it without that claim. The Quinean rejects the problem. (p.16) The Quinean sees no problem for Nominalism in the likes of (3) because there is a well‐known semantic theory which shows that (3) can be true without there being any universals: (3) is true if and only if there exists an x such that ‘a’ designates x and ‘F’ applies to x. So (3) can be true without the existence of F‐ness. There is no refusal here “to take predicates with any ontological seriousness”. The Quinean thinks that there really must exist something (said as firmly as you like) that the predicate ‘F’ applies to. However that thing is not a universal but simply an object. Further, in denying that this object need have properties, the Quinean is not denying that it really is F (or G, or whatever). He is not claiming that it is “a bare particular”. He sees no need to play that game.

Page 3 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

“Ostrich Nominalism” or “Mirage Realism”? * The Realist may reply that this is a mistaken statement of the truth conditions of (3) and that the correct one does require the existence of F‐ness for (3)’s truth. Until a good argument for this reply is produced the Quinean is entitled to go on thinking he has no problem. All of this is not to say that there is nothing further about (3), or about a being F, that might need explanation. I can think of four possible problems here. None of them pose any special difficulty for the Nominalist: they are irrelevant to “the problem of universals”. (i) We might need to explain what caused a to be F. (ii) We might need to explain what was the purpose of a being F. Nobody interested in “the problem of universals” is likely to confuse their problem with (i) or (ii) and so I shall set them aside immediately. It is not so easy to keep the next two problems distinct from “the problem of universals”. (iii) If ‘F’ is not a fundamental predicate then as reductivists we might need to explain what constitutes a being F: perhaps we will want to be told that it is in virtue of being G, where ‘G’ is some physical predicate (a is a gene in virtue of being a DNA molecule). (iv) We might need to explain the semantics of ‘F’: we might want to know what makes it the case that ‘F’ applies to a. The traditional “problem of universals” has often appeared in a misleading semantic guise: how can ‘F’ “be applied to an indefinite multiplicity of particulars” (p. xiii; Armstrong does not approve of this way of putting the problem)? The strictly semantic problem of multiplicity does not have anything to do with universals. We need to explain the link between ‘F’ and all F things in virtue of which the former applies to the latter. This is not different in principle from explaining the link between ‘a’ and one object, a, in virtue of which the former designates the latter. The explanation of ‘F’’s application (p.17) depends on a theory of one semantic relation, application, the explanation of ‘a’’s designation depends on a theory of another, designation. A feature of the explanations will be that it is F things that are linked to ‘F’, and a that is linked to ‘a’. The F‐ness of F things and the a‐ness of a need not go unexplained in the semantics. Thus I think it is part of a good explanation of the link between ‘tiger’ and the many objects that it applies to that those objects are genetically of a certain sort. So the semantic problem may require some answer to the question: in virtue of what is a F? But the answer required is of type (iii), a reductivist answer. In denying that there is any problem for the Nominalist about (3) it is important to see that we are not denying the reductivist problem (iii), nor the semanticist problem (iv), nor some combination of (iii) and (iv). What we are denying can be brought out vividly by taking ‘F’ to be a fundamental predicate, say a physical predicate. Then there is no problem (iii): we have nothing to say about what Page 4 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

“Ostrich Nominalism” or “Mirage Realism”? * makes a F, it just is F; that is a basic and inexplicable fact4 about the universe. Problem (iv) remains: it is the problem of explaining the link between the predicate ‘F’ and that basic fact. Nothing else remains to be explained. Why be dissatisfied with this? Explanation must stop somewhere. What better place than with a fundamental physical fact of our world? Armstrong feels that we need to go further. How can we tell who is right? There is one sure sign that explanation has not gone far enough: an explanation that goes further. Thus if Armstrong's Realist response to the One over Many argument is a genuine explanation then there must be a genuine problem here to be explained. My final remarks in support of thesis I will consider Armstrong's response. One Realist response, but not Armstrong's, to the One over Many argument runs as follows: a is F in virtue of having the property F‐ness. We explain (3) by (5) a has F‐ness. An obvious question arises: how is (5) to be explained? The Realist feels that the one‐ place predication (3) left something unexplained, yet all he has done to explain it is offer a two‐place predication (a relational statement). If there is a problem about a being F then there is at least an equal problem about a having F‐ness. Furthermore, the point of this manoeuvre for the Realist is to commit us to universals. In ontology, the less the better. Therefore this (p.18) sort of Realist makes us ontologically worse off without explanatory gain. Any attempt by him to achieve explanatory power by explaining (5) seems doomed before it starts: it will simply raise the same problem as (5); he is in a vicious regress. If there is a problem about (3) this sort of Realist cannot solve it.

Armstrong calls the doctrine we have just considered “relational Immanent Realism”, and rejects it for reasons not unconnected to mine (pp. 104–7). In its place he offers us “non‐relational Immanent Realism.” This doctrine is obscure. Armstrong offers us (5), or the similar, ‘F‐ness is in a,’ and simply declares it to be non‐relational and inexplicable: particulars are not related to universals but bonded to them in a metaphysical unity (pp. 108–11). We have just seen that (5), taken at face value, cannot explain any problem about (3): it is a relational statement and so any problem for (3) is a problem for it. Armstrong avoids this grievous difficulty for Realism by fiat: (5) is not to be taken at face value. How then is it to be taken? Do we have even the remotest idea of what the words ‘in’ and ‘have’ mean here if they are not construed as relational predicates? Armstrong's Realism replaces the explanatory failings of relational Realism with a complete mystery. I suspect that Armstrong views sentences like (5) as attempts to speak the unspeakable: to talk about “the link” between particulars and universals without saying they are related. (Note the scare‐quotes around ‘in’ on p. 108 and the use of a special hyphenating device on p. 111.)

Page 5 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

“Ostrich Nominalism” or “Mirage Realism”? * Talk of “particulars” and “universals” clutters the landscape without adding to our understanding. We should rest with the basic fact that a is F. Even the alleged unity of particular and universal can be captured without mystery: a predication must involve both a singular term and a predicate; drop either partner and you say nothing. For the Nominalist the unity of predication is an unexciting linguistic fact. The move to relational Realism loses the unity. Armstrong's non‐relational Realism attempts to bring it back with metaphysical glue. These are “degenerating problem shifts” (Lakatos). Armstrong sees the One over Many argument as posing a problem for Nominalism and offers a Realist solution. If his solution were real then the problem would be real. The solution is not real. So it throws no doubt on my earlier argument that the problem is not real. Indeed the Quinean can gain much comfort from Armstrong's book: it is a powerful argument for thesis 1. We have just demonstrated the failings of Armstrong's response to the One over Many argument. Armstrong himself carefully, and convincingly, demolishes every other known response to it. This (p.19) chronicle of two thousand years of failure makes the task seem hopeless. The alternative view that there is no problem to solve becomes very attractive. I take my major thesis to be established: 1. To maintain Nominalism whilst ignoring the One over Many argument is not to be an Ostrich Nominalist; rather to adopt Realism because of that argument is to be a Mirage Realist. Even if there are universals they cannot form part of a solution to the One over Many problem, because that problem is a mirage.

Argument for Thesis 2 The arguments for theses 2 and 3 will be brief. It follows from thesis 1 that in so far as Armstrong adopts Realism because of the One over Many argument, he is a Mirage Realist. At the beginning of his book he indicates that he sees that argument as the main one for universals (p. xiii). When he talks of “the problem of universals” it is the problem allegedly posed by that argument that he is referring to (e.g. p. 41). Almost the whole book is taken up with the consideration of responses to that argument. Armstrong is largely a Mirage Realist. In one chapter, drawing on the ideas of Arthur Pap and Frank Jackson, Armstrong offers quite independent reasons for Realism (pp. 58–63).5 We all assent to, express, believe, statements like the following: (6) Red resembles orange more than it resembles blue; (7) Red is a colour; Page 6 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

“Ostrich Nominalism” or “Mirage Realism”? * (8) He has the same virtues as his father; (9) The dresses were of the same colour. Unlike (3) these seem to require the existence of properties for them to be true. Whether or not they are sufficient for Realism depends on whether or not we can find acceptable paraphrases without that commitment. There is nothing illusory about this problem for a Nominalist. Armstrong is not entirely a Mirage Realist. So,

2. Armstrong is largely though not entirely a Mirage Realist.

(p.20) Argument for Thesis 3 For Quine to be an Ostrich Nominalist would be for him to ignore the ontological problem posed by his acceptance of statements like (6) to (9). A priori it is unlikely that this would be so. Quine, more than any other philosopher, has pointed out what constitutes an ontological commitment and has preached against ignoring such. Philosophers, like others, can fail to practise what they preach, but I suggest that it is unlikely that Quine would fail here, about as unlikely as that he would confuse use and mention. A quick glance through Word and Object6 shows that he does not fail. In a section on abstract terms he considers, for example, the sentence, (10) Humility is a virtue. a sentence that raises much the same problem as Armstrong's (8), and sees it as committing him to the existence of “an abstract object” (p. 119), in fact to “an attribute”, what Armstrong would call “a property”. He goes on to “deplore that facile line of thought” that allows us to ignore this (pp. 119–20). He considers ways to paraphrase away this apparent commitment to attributes and admits the difficulties (pp. 121–3). The issues are postponed until chapter 7. He does not there discuss sentences like (6) to (10) directly, so far as I can see, but his strategy for them is clear enough: all talk of attributes is to be dispensed with in favour of talk of eternal open sentences or talk of classes (p. 209). Whatever the merits of this approach it is not the behaviour of an Ostrich. So,

3. Quine is not an Ostrich Nominalist. Postscript to “ ‘Ostrich Nominalism’ or ‘Mirage Realism’?”

1. Criteria of Ontological Commitment 1.1 Semantic and Quinean criteria

My presentation of the Quinean criterion of ontological commitment begins with the claim that “a person is committed to the existence of those things that (p. 21) must exist for the sentences he accepts to be true”. I go on: “What must exist for a given sentence to be true is a semantic question” requiring a “semantic theory” to answer it. I propose part of a theory for that purpose: ‘a is F’ is true iff there exists an x such that ‘a’ designates x and ‘F’ applies to x. This was a mistake7 and is at odds with “Putting Metaphysics First”, the title and guiding idea of the present volume.

Page 7 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

“Ostrich Nominalism” or “Mirage Realism”? * The real Quinean criterion of ontological commitment is simpler than the semantic criterion I presented. The Quinean criterion is that a person is committed to the existence of a or Fs if he says that a or Fs exist. As Quine points out: We commit ourselves to an ontology containing numbers when we say there are prime numbers larger than a million; we commit ourselves to an ontology containing centaurs when we say there are centaurs; and we commit ourselves to an ontology containing Pegasus when we say Pegasus is. (1961: 8) Having emphasized that mere talk of red houses, roses, and sunsets does not commit us to the existence of redness, Quine goes on:

We can very easily involve ourselves in ontological commitments by saying, for example, that there is something (bound variable) which red houses and sunsets have in common. (p. 12)8 If we said this we would commit ourselves to the property redness, a universal.9 But no semantic theory is needed to show this commitment. For someone to show this, the person must of course understand what we say but this understanding does not require the application of a semantic theory. Our ordinary understanding of a language is a skill quite unlike any theoretical knowledge of language. Or so I have argued at painful length (1981: 95–110; 1991b: 270–5; 2006d).10 (p.22) 1.2 The Priority of the Quinean Criterion

This is not to say that there is anything wrong with using the semantic criterion to establish commitment. However, it has the disadvantage of requiring a semantic theory that may be controversial. Furthermore, it is important to note that the Quinean criterion is explanatorily prior. Consider Tom's assertion: (1) Lulu is a cat. Applying the semantic criterion, and using our semantic theory, we see this assertion as ontologically committing Tom to Lulu and cats, because, first,

(a) (1) is true iff there exists something such that ‘Lulu’ designates it and ‘cat’ applies to it; and second,

(b) If ‘Lulu’ designates, it designates Lulu; (c) If ‘cat’ applies, it applies to cats. What is it about (a), (b), and (c) that shows that Tom is ontologically committed to Lulu and cats in asserting (1)? Clearly it is our talk of existence. Tom is ontologically committed to cats if we are right in saying, as a consequence of (a) and (c), that cats must exist for (1) to be true. So the semantic criterion presupposes that saying that cats exist establishes an ontological commitment to cats; i.e. it presupposes the Quinean criterion.11

Page 8 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

“Ostrich Nominalism” or “Mirage Realism”? * 1.3 Consequences of Using the Semantic Criterion

In using the semantic criterion I had to apply a semantic theory. This invited an objection made by David Armstrong in his response to my paper: “there may be alternative, and perhaps more satisfying, ways of giving the semantics (p.23) for ‘Fa’. Devitt offers no argument against this possibility” (1980: 445). Hugh Mellor and Alex Oliver object similarly in their helpful introduction to a collection of papers that includes mine and Armstrong's: “Since [Devitt] and Quine link ontological commitment to a semantic theory, we need some way of telling when such a theory is correct, and an argument for its supposed link with ontological commitment” (1997b: 14–15). Now, whatever the merits of these objections they do nothing to support Armstrong's traditional claim that “the main argument for the existence of universals is Plato's ‘One over Many’ ” (1980: 440). And that claim is the main target of my paper and, of course, of Quine's “On What There Is”. To support the traditional claim, we need an argument to show that the semantic theory I use is not correct. For, if that theory is correct, the One over Many poses no problem for the Nominalist. To sustain the traditional claim it is not sufficient to suggest, rightly, that the semantic theory might not be correct. The onus is on those who make the traditional claim to show that the theory is not correct.12 As noted, my semantic proposal for the truth conditions of ‘a is F’ was: ‘a is F’ is true iff there exists an x such that ‘a’ designates x and ‘F’ applies to x. Mellor and Oliver continue their objection as follows:

suppose the following clause appears in the correct semantic theory: ‘a is F’ is true iff there is a ϕ such that ‘is F’ designates ϕ and ‘a’ falls under ϕ. This suggests that ‘a is F’ is only committed to F‐ness, not to an entity designated by ‘a’. (1997b: 15) According to my Nominalistic semantic proposal, the semantic criterion takes ‘a is F’ to be ontologically committed to an F object; according to the present Realist proposal, the criterion takes ‘a is F’ to be committed to the universal F‐ness. Which semantic proposal should we prefer? That is a good question. But whatever the answer, it will do nothing to sustain the traditional claim about the importance of the One over Many. Clearly, if my Nominalistic proposal should be preferred, the semantic criterion shows that the One over Many is no problem for Nominalism, just as I argued. If, on the other hand, (p.24) the Realist proposal should be preferred, then the One over Many is irrelevant: the argument that justified the preference for a Realist semantics would have already established Realism.13

Page 9 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

“Ostrich Nominalism” or “Mirage Realism”? * So why should we prefer one semantic proposal over the other? There can be no purely semantic answer. Rather, we to look to our best scientific theories of the world and apply the Quinean criterion. According to that criterion, our best theories clearly and unequivocally commit us to objects but the jury is still out on whether they commit us to universals. So we should prefer a Nominalistic semantics that is committed only to objects. In this way we put metaphysics first.14 1.4 Revisiting the One over Many

Putting all the weight on the semantic criterion was a mistake. Set that criterion aside and apply the Quinean one. Then the basic case against Armstrong and the One over Many is that the “identity in nature” of red houses and sunsets need not commit us to the universal redness. We accept that the houses and sunsets exist and explain away their “identity in nature” as simply a matter of their all being red. No more need be said. The One over Many is a pseudo problem and to adopt Realism about universals because of it is, indeed, to be a Mirage Realist.

2. Paraphrasing Mellor and Oliver have another problem with the Quinean approach to ontological commitment, arising from the role that the approach gives to paraphrasing. They describe this role as follows: Suppose we have a sentence S, apparently committed to some entity e, and an equivalent sentence S’ which is said to be uncommitted to e. This, it is said, shows that S is only apparently committed to e. Calling on Alston (1958), they object:

Why should we think it is S and not S’ that deceives us? Why not say the S’ is really committed to e because its equivalent S is? What we need here is what we do not have, namely a test for when a sentence is only apparently, and when really, committed or uncommitted to some entity. (1997b: 15) (p.25) This objection fails because it rests on several misunderstandings.15 First, we need to distinguish the ontological commitments of sentences and of people. Quine's criterion “applies in the first instance to discourse and not to men” (1961: 103). When it is applied to S and S’, their commitments will be real not simply apparent: either they do or do not “quantify over” e by saying that e exists; neither “deceives us”. And there is no role for paraphrasing here. Second, paraphrasing has its role in determining the commitments of people. Suppose that S is committed to e but that S’ is not. Suppose that a person asserts S thereby apparently committing himself to e. Yet suppose, further, that he is prepared to withdraw that assertion in favor of S’ because S’ lacks the commitment and yet will serve his purposes well enough. He thus “frees himself from ontological commitments of his discourse” (ibid.). His commitment to e is only apparent not real: it arose from “an avoidable manner of speaking” (p. 13). Third, in these circumstances, we might say that S and S’ are “equivalent” in some sense— they serve the same purpose—but we should not say that they are equivalent in the

Page 10 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

“Ostrich Nominalism” or “Mirage Realism”? * sense of being synonymous. For, if they were synonymous, it is hard to see how they could differ in their ontological commitments.16

3. Response to Armstrong 3.1 The Genuine Problem of Universals

In my paper I followed Quine in allowing that there is a genuine problem of universals arising from our apparent quantification over properties; see Arguments for Thesis 2 and for Thesis 3.17 Armstrong has quite a bit more to say about this problem in his response (1980: sec. II). In assessing this response, it is important to note that I made no attempt to solve this genuine problem in my paper. In particular, I presented no argument against Realism about universals. And I shall not do so here. 3.2 The One Over Many

I wrote my paper largely because I felt that Armstrong, in his book (1978a), had taken “no serious account” of Quine's argument that the One over Many (p.26) is not a genuine problem for Nominalism. This Quinean point is the concern of my Argument for Thesis 1, occupying almost the entire paper. It seems to me that Armstrong, in his response (1980), still takes no serious account of the Quinean argument. Rather, his attention is elsewhere, looking for other possible reasons for Realism. We have already seen evidence of this in section 1.3 above. Here is more. Armstrong again insists that the One over Many “shows that there is a strong preliminary case for accepting universals” (p. 440). But the Quinean point is, of course, that the One over Many provides no case at all. Armstrong draws attention to ubiquitous talk of sameness of type. He claims that this sameness is “a Moorean fact” that needs an account. He then simply repeats the charge that Quine is an ostrich for not giving such an account, for “refusing to answer a compulsory question” (p. 441). But, of course, my Argument for Thesis 1 is precisely that Quine has answered this question. To illustrate, a and b might be “of the same type” in virtue of their both being red; no talk of the universal redness is called for. Armstrong has largely ignored the Quinean argument. Armstrong does add a new wrinkle: he raises the sameness‐of‐type problem at the level of predicates. He claims that the Quinean answer I have just illustrated rests on a rule along the following lines: Suppose that we are given sentences that of the form ‘a is—and b is—.’ If but only if the two blanks are filled by the same predicate, it is permitted to rewrite the sentence as ‘a and b are both—,’ with the same predicate in the new blank. But ‘same predicate’ here is a type‐notion. . . . Some account must then be given, reductive or otherwise, of what sameness of type is. (p. 442) But the Quinean treats talk of the sameness of type of predicates in just the same way that he treats talk of the sameness of type of stones, trees, cats, or whatever. Just as Page 11 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

“Ostrich Nominalism” or “Mirage Realism”? * we paraphrase away talk of objects sharing the property redness by saying that they are all red, we paraphrase away talk of objects sharing the property of being of the word type ‘red’ by saying that they are all ‘red’ tokens. The Quinean treatment is quite general. The One over Many is a pseudo problem at all levels. 3.3 “Solutions” to the Pseudo Problem

In my paper I emphasized that the Quinean does not deny that an object really is red or whatever. Armstrong is unmoved, insisting that, on the Quinean view, “particulars are a sort of structureless blob . . . they lack real internal structure” (p. 446). This is a caricature. It foists on the Quinean an ontological framework that is motivated by the One over Many problem, (p.27) just the problem that the Quinean rejects. So the problem does not lead the Quinean to traffic in “bare particulars”, “mere thisnesses”, and the like; as I remarked, “he sees no need to play that game”. Suppose that, according to the Realist, an object has an internal structure F‐ness. Then, according to the Quinean, it really is F, said as firmly as you like. Nothing more need be said. The Quinean gains comfort from the dismal failure of attempted solutions to the One over Many problem through more than two millennia. Armstrong poses the issue vividly: Realists hold that “a particular involves a factor of particularity (haeccitas, thisness) together with properties which are universals. The question is then this: how are the two components of a particular to be put together?” (p. 446) Relational answers always generate a vicious regress, as Armstrong has demonstrated convincingly. Armstrong's non‐relational answer is that “the thisness and the nature are . . . not related”. Armstrong admits, with admirable frankness, that this is “profoundly puzzling” and that my claim that it is an “inexplicable mystery” is not implausible (p. 447). Indeed, as I say, non‐relational answers seem to involve “attempts to speak the unspeakable: to talk about the ‘the link’ between particulars and universals without saying they are related”. 3.4 Armstrong's ad Hominem

I agree that there is a genuine problem of universals but have made no attempt to solve it. I think, as Armstrong points out, “that it may be necessary to postulate universals”. Suppose that it is. Armstrong asks how I will “solve the problem of how universals stand to particulars?” He thinks that I will “end up saying something similar” to what he and other Realists have had to say (p. 448). I might well end up saying something similar to what the Relational Realists have had to say. But, if I did, I would not be open to the vicious‐regress objection that Armstrong has leveled at these Realists. For, that objection is effective only against those who are Realists as a solution to the One over Many problem. Those Realists must accept a further universal every time they relate a universal to a particular. My Realism would not be a solution to that problem and so I would not have to accept the further universal.18 A Relational Realism that is a response to the pseudo problem of universals must indeed generate a vicious regress; a Relational Realism that is a response to the genuine problem need not. If the One over Many were a genuine problem, the vicious‐regress objection Page 12 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

“Ostrich Nominalism” or “Mirage Realism”? * would show that it (p.28) was insoluble. So, we have very good reason to think that it is not a genuine problem.

4. Lewis's One over Many David Lewis also finds Armstrong's discussion of the One over Many problem “unconvincing” (1983: 351) and agrees with much of what I say about it (p. 354). However, he thinks that this problem, concerned with a and b have the same property (are of the same type), F‐ness, is “too easy”. There is another One over Many problem concerned with the less definite

a and b have some common property (are somehow of the same type). that is not so easy to deal with (p. 355). I agree, but I think that it is still not that difficult.

First, what are we to make of the less definite claim? Lewis has remarked earlier: “Because properties are so abundant, they are undiscriminating. Any two things share infinitely many properties, and fail to share infinitely many others” (p. 346). So, if we are prepared to talk of properties it seems trivial to say that two objects share one. In any case, it is easy for the Quinean to paraphrase such an apparently trivial claim with the similarly apparently trivial: a and b resemble each other. But doubtless someone making the less definite claim has in mind that a and b share a significant property, “one of an elite minority of special properties” which Lewis calls “natural properties” (p. 346). If so, we can paraphrase the less definite claim well enough with:

a and b significantly resemble each other. In sum, in the paper we paraphrased away an apparent commitment to a and b sharing a particular property F‐ness by talking of their both being F. We now paraphrase away an apparent commitment to a and b sharing some property or other by talking of their resembling, or significantly resembling, each other.

(p.29) 5. A Diagnosis The One over Many is a pseudo problem. Why, then, are philosophers so beguiled by it?19 I suspect that the reason is an implicit commitment to the “ ‘Fido’‐Fido” theory of meaning. This theory starts from the appealing idea that the meaning of a proper name like ‘Fido’ is the object it names, Fido.20 The theory generalizes this view of meaning to all terms. The theory has had a persistent hold over the minds of philosophers and many others. Consider, for example, what Armstrong has to say about Quine's “extraordinary doctrine” of ontological commitment. He claims that, according to this doctrine, “predicates involve no ontological commitment”; a predicate “need not be taken with ontological seriousness” it has “what has been said to be the privilege of Page 13 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

“Ostrich Nominalism” or “Mirage Realism”? * the harlot: power without responsibility” (1980: 443). Yet, as we have seen, the Quinean takes predicates with great ontological seriousness: ‘There are Fs’ is committed to the existence of Fs because of the role of the predicate ‘F’. Why is that not serious enough for Armstrong? It looks as if he thinks that to be really serious we must take ‘F’ to name something: we must adopt the ‘Fido’‐Fido theory. It is easy to see how this theory leads us to universals. Consider ‘That rose is red’. This sentence, like all others, has a certain complexity. It has two terms, the singular term ‘that rose’ and the general term ‘red’, of different grammatical categories and playing quite different roles. How can the ‘Fido’‐Fido theory cope with this complexity? It has to see the two types of term naming two types of entities: the different roles of the terms require different types of entities. The entity named by ‘that rose’ is a particular rose; that named by ‘red’ is the universal, redness, which can be shared by many particulars. The One over Many begins to look like a real problem. In my view, the ‘Fido’‐Fido theory is false. The present chapter illustrates a quite different way of coping with the complexity of sentences. It is not that each term stands in the one semantic relation of naming to different kinds of entities. Rather, the terms stand in different semantic relations to the same kind of entities, neither “particulars” nor “universals” but just plain objects. Thus ‘that rose’ designates a certain object, a rose, while ‘red’ applies to many objects, including many roses. Where the ‘Fido’‐Fido theory catches the complexity (p. 30) with different sorts of entity, we catch it with different sorts of relation. The only entities we need are objects of the familiar sort. If this is the right diagnosis, the One over Many problem shows the cost of not putting metaphysics first: our semantics should be driven by our metaphysics not vice versa. Perhaps that metaphysics cannot be Nominalist but we have seen that the One over Many provides no reason for thinking that it cannot.21 Notes:

(*) First published in Pacific Philosophical Quarterly, 61 (Devitt 1980). Reprinted with kind permission from Wiley‐Blackwell. I am indebted to Elizabeth Prior for help with the first draft of this paper and to David Armstrong and Frank Jackson for helpful comments on that draft. (1) Such references are to Armstrong 1978a: i. (2) See particularly Quine's discussion in “On What There Is” (1961: 9–14), which Quine's discussion is largely aimed at a position like Armstrong's (“For ‘McX’ read ‘McArmstrong’ ”: Elizabeth Prior). (3) See e.g. his remarks on Ostrich Nominalism (quoted above) and his discussion of the varieties of Nominalism, pp. 12–16. Page 14 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

“Ostrich Nominalism” or “Mirage Realism”? * (4) Lest an uncharitable reader should take this talk as committing me to the existence of facts, let me hasten to add that such talk is a mere manner of speaking, eliminable at the cost of style and emphasis. (5) Given the importance Armstrong attaches to the One over Many argument for Realism, this chapter's title, “Arguments for Realism”, is misleading. (6) Quine 1960. (7) As I soon came to realize (1984: 40–3; 1991b: 50–3). This subsection and the next draws on that discussion. (8) So, note, the Quinean criterion is not “biased against universals”. The criterion makes it easy for us to commit ourselves to absolutely anything. (9) Jody Azzouni (2007) has offered an ingenious argument against the Quinean criterion. I am not convinced. Azzouni is on strongest ground in denying the ontological commitment of ‘there is . . . ’ in ordinary language. If he were right about that (which I don't think he is), the Quinean would simply fall back on ‘there exists . . . ’, or even ‘there really exists . . . ’, as the locus of commitment. But Azzouni denies that these expressions, indeed any expression in the vernacular, have a meaning that entails ontological commitment. How then do we convey that commitment? For Azzouni this is a pragmatic rather than semantic matter: we use rhetorical indicators, stage‐whisperings, and the like. This is prima facie very implausible. Given our ordinary and scientific interest in ontological matters it would be strange indeed if we had no expression in our language that entailed ontological commitment. It would be extraordinary e.g. that there could be no words in a theory that committed it to a certain ontology. And ‘exists’ seems perfectly designed for the job. (10) Quine sometimes uses ‘true’ and ‘refer’ in stating his criterion: “a theory is committed to those and only those entities to which the bound variables of the theory must be capable of referring in order that the affirmations made in the theory be true” (pp. 13–14). But this use of ‘true’ simply exploits the “denominalizing” (“disquotational”) property of ‘true’ and does not make the criterion semantic; similarly, ‘refer’. So these uses do not make Quine's criterion dependent on the application of a semantic theory. Indeed, Quine has a deflationary view of truth. For more on the deflationary view, and the denominalizing property of the truth term, see ch. 2, part I, and ch. 8, particularly sec. 3, in the present volume. (11) And just the same goes for the “truthmaker” criterion proposed recently by David Armstrong (2004: 23–4). For, on this criterion, a theory is ontologically committed to certain entities only because when we apply the criterion to the theory we conclude that those entities must exist in order to make the theory true. The criterion is parasitic on the Quinean criterion, as Jonathan Schaffer Page 15 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

“Ostrich Nominalism” or “Mirage Realism”? * points out (2008: 16). In 7.3, of the present volume, I argue against the view that we need a metalinguistic statement of the truth conditions of a statement in order to establish the commitments of the statement. (12) A similar response is appropriate to Armstrong's claim about the technical term ‘applies’ that features in my semantic theory: “The Realist may well argue, correctly I believe, that a convincing account of the semantics of ‘applies’ cannot be given without appeal to the properties and/or relations of the object a” (p. 445). Of course the Realist might argue this but he hasn't. So the problematic nature of the One over Many has still not been established. In any case, ‘applies’ is surely no more problematic for the Nominalist than any other two‐place predicate. (13) The One over Many would be similarly irrelevant if Armstrong (2004) were right in supposing that the entities that must exist for our theories to be true are “states‐of‐affairs”. Whatever justified that supposition would establish Realism about states‐of‐affairs without appeal to the One over Many argument. (14) I draw here on Devitt 1984: 45–6; 1991b: 57–8. (15) My own discussion of paraphrasing in the present chapter is not as clear as it should have been. (16) Despite all this I do accept that the role of paraphrasing raises a difficulty for Quine's criterion (1984: 44; 1991b: 54). (17) I cited Word and Object (1960) as evidence of how seriously Quine takes the problem of universals. I might also have cited “Logic and the Reification of Universals” (1961: 102–38). (18) As Quine says in his response to Armstrong on the regress problem, “the use of a two‐place predicate is not itself a reference to the relation” (1980: 451). (19) My discussion in this subsection draws on Devitt and Sterelny 1999: 279– 80. (20) Although the idea is appealing I think it is not right (1981, 1996a, 2009 d). (21) My thanks to David Armstrong and Jonathan Schaffer for comments on this postscript.

Access brought to you by:

Page 16 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate *

Putting Metaphysics First: Essays on Metaphysics and Epistemology Michael Devitt

Print publication date: 2009 Print ISBN-13: 9780199280803 Published to Oxford Scholarship Online: February 2010 DOI: 10.1093/acprof:oso/9780199280803.001.0001

Aberrations of the Realism Debate * Michael Devitt (Contributor Webpage)

Antirealism about the physical world is an occupational hazard of philosophy. Most of the great philosophers have been antirealists in one way or another. Many of the cleverest contemporary philosophers are also: Michael Dummett, Nelson Goodman, Hilary Putnam, and Bas van Fraassen. Yet antirealism is enormously implausible on the face of it. In this chapter, I shall be concerned with several aberrations of the contemporary debate over realism. These aberrations have many sources: those who oppose realism; some of those who support realism; those who stand above the battle and dismiss it; and some more neutral figures. Some aberrations have a long history, going back to the positivists and beyond. The defense of realism depends on distinguishing it from other doctrines and on choosing the right place to start the argument. In this respect, realism has not been well‐served by some of its friends, particularly an earlier Putnam (1978: 1– 77, 97–119). In Realism and Truth (1984), I argued for four maxims which formed the basis of my defense of realism.1 Many of the aberrations that concern me here infringe the maxims. To some extent, the argument of this paper depends on that earlier argument. In part I of the chapter, the focus is on the connection between realism and semantics. In part II, I consider the bearing of epistemology and naturalism on the issue. Aside from identifying aberrations, I shall occasionally offer diagnoses and respond to objections. The first aberration comes late in part I.

(p.32) I: Realism and Semantics

Page 1 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * A striking aspect of the realism debate is that it contains almost as many doctrines under the name ‘realism’ as it contains participants.2 However, some common features can be discerned in this chaos. First, nearly all the doctrines are, or seem to be, partly semantic. Consider, for example, Jarrett Leplin's editorial introduction to a recent collection of papers on scientific realism. He lists ten “characteristic realist claims” (1984b: 1–2). Nearly all of these are about the truth and reference of theories. Not one is straightforwardly metaphysical.3 However, second, amongst all the semantic talk, it is usually possible to discern a metaphysical doctrine, a doctrine about what there is and what it's like. Thus ‘realism’ is now usually taken to refer to some combination of a metaphysical doctrine with a doctrine about truth, particularly with a correspondence doctrine.4 The metaphysical doctrine, which is what I shall call “realism”, has two dimensions, an existence dimension and an independence dimension. The existence dimension commits the realist to the existence of such commonsense entities as stones, trees and cats, and such scientific entities as electrons, muons and curved space‐time. Typically, idealists, the traditional opponent of realists, have not denied this dimension; or, at least, have not straightforwardly denied it. What they have denied is the independence dimension. According to some idealists, the entities identified by the first dimension are made up of mental items: “ideas” or “sense data”. In recent times another sort of idealist has been much more common. According to these idealists, the entities are not in a certain respect “objective”: they depend for their existence and nature on the cognitive activities and capacities of our minds. Realists reject all such mind dependencies. Relations between minds and those entities are limited to familiar causal interactions long noted by folk theory: we throw stones, plant trees, see cats, and so on. Though the focus of the debate has mostly been on the independence dimension, the existence dimension is important. First, it identifies the entities that are the subject of the dispute over independence. In particular, it distinguishes a realism worth fighting for from what I call “Fig‐Leaf Realism” (1984: (p.33) 22; 1991b: 23): a commitment merely to there being something independent of us. Second, in the discussion of unobservables—the debate about scientific realism—the main controversy has been over existence. I capture the two dimensions in the following doctrine: Realism: Tokens of most commonsense, and scientific, physical types objectively exist independently of the mental. This doctrine covers both the observable and the unobservable worlds. Some philosophers, like van Fraassen, have adopted a different attitude to these two worlds. So, for the purpose of argument, we can split the doctrine in two: Commonsense

Page 2 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * Realism concerned with observables, and Scientific Realism concerned with unobservables.

In insisting on the objectivity of the world, realists are not saying that it is unknowable. They are saying that it is not constituted by our knowledge, by our epistemic values, by the synthesizing power of the mind, nor by our imposition of concepts, theories, or languages; it is not limited by what we can believe or discover. Many worlds lack this sort of objectivity and independence: Kant's “phenomenal” world; Dummett's verifiable world; the stars made by a Goodman “version”; the constructed world of Putnam's “internal realism”; Kuhn's world of theoretical ontologies;5 the many worlds created by the “discourses” of structuralists and post‐structuralists. Realism takes both the ontology of science and common sense, and the folk epistemological view that this ontology is objective and independent, pretty much for granted. Science and common sense are not, for the most part, to be “reinterpreted”.6 It is not just that our experiences are as if there are cats, there are cats. It is not just that the observable world is as if there are atoms, there are atoms. As Putnam once put it, Realism takes science at “face value” (1978: 37). Realism is the minimal realist doctrine worth fighting for. Once it is established, the battle against antirealism is won; all that remain are skirmishes. Furthermore, Realism provides the place to stand to solve the many other difficult problems that have become entangled with it. Any semantic problem needs to be disentangled from Realism.7 In particular, the correspondence theory of truth is in no way constitutive of Realism, nor of any similarly metaphysical doctrine. The first step in arguing for (p.34) this is to distinguish correspondence truth from deflationary (prosentential, disquotational) truth. According to a deflationary theory, the truth term is a linguistic device that is important to the expressive powers of language; it enables us to talk about reality by referring to sentences and thoughts. However, the truth term does not describe a sentence or thought and hence does not explain anything about its nature. The “equivalence thesis” holds for deflationary truth, as for any notion worthy of the name ‘truth’: appropriate instances of s is true if and only if p hold.8 But the thesis holds simply because to refer to a sentence and assert that it is true is just to assert the sentence. According to the correspondence theory, in contrast, the thesis holds in virtue of a genuine pairing of sentences with situations, perhaps one explained in terms of the referential relations of parts of the sentence. Most important of all, correspondence truth is an explanatory notion in a theory of the meaning or content of sentences and thoughts: it is because of this correspondence relation that sentences and thoughts have their roles.9 Page 3 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * Suppose that one has in mind the metaphysical doctrine Realism. A natural way to state it is as follows: Realism*: Most commonsense, and scientific, physical existence statements are objectively and mind‐independently true. And this is a completely appropriate way to state the doctrine, provided that ‘true’ refers to deflationary truth: it is an illustration of the sort of “semantic ascent” that deflationary truth makes possible, and often convenient, whatever the subject matter. Such ascent does not make a doctrine semantic, in the sense of being part of a theory of meaning; it does not change the subject matter at all.10 Indeed, if it did, any doctrine about anything would become part of the theory of meaning (the Oxford Dream!).

Consider now the relation between Realism and truth. On the one hand, Realism does not entail the correspondence theory nor any other theory of (p.35) truth or meaning. One can even be a realist and yet eliminativist about the semantic properties of thoughts and language, as has been nicely demonstrated by Stephen Leeds (1978) drawing on the views of Quine. On the other hand, the correspondence theory does not entail Realism. The correspondence theory claims that representations do or do not have the property truth according as they do or do not correspond in some way to reality. This is compatible with absolutely any metaphysics. The theory is often taken to require the objective independent existence of the reality which makes representations true or false. This addition of Realism's independence dimension does, of course, bring us closer to Realism. However, the addition seems like a gratuitous intrusion of metaphysics into semantics.11 And even with the addition, the correspondence theory is still distant from Realism, because it does not require that any particular sentences be true and so could not require the existence of any particular entities. The correspondence theorist could make another addition to deal with this: most commonsense, and scientific, physical existence statements are true. This addition to a semantic theory is totally gratuitous. Furthermore, if its talk of truth is minimally interpreted as merely deflationary, it is simply a statement of the existence dimension. The two additions, minimally interpreted, are Realism*; i.e. are Realism. It is hardly surprising, then, that with the help of these additions we can derive Realism from the correspondence theory! That theory is entirely irrelevant to the derivation.12 Realism is about the nature of reality in general; it is about the largely inanimate impersonal world. If correspondence truth has a place, it is in our theory of only a small part of that reality: it is in our theory of people and their language.

Page 4 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * Objection 1. “You are Simply Insisting that the Word ‘Realism’ be used Metaphysically not Semantically. That is a Merely Verbal Point.”

My main point is not verbal at all. I am insisting on a distinction between two doctrines, whatever they are called. I am insisting on carving theory at reality's joints. (p.36) On the verbal point, I claim that metaphysical doctrines like mine capture the only thing that is distinctive about views that have been called “realist” about the external world throughout the long debate. I have already indicated that correspondence truth is not distinctive. In part II we shall see that other views—for example, about the aim of science—are not either. Finally, it would be rather perverse to use ‘realism’ to refer to a doctrine that had no metaphysical intent.13 Objection 2. “Your Realist Doctrine is not What all the Argument is about.”

All the argument is not about this doctrine, but part of it certainly is. It is precisely because the doctrine is so often denied in philosophy that it is worth asserting. For examples of its denial one need only look to the history of idealism. And it is still being denied; see above. Either familiar ontic commitments are, explicitly or implicitly, paraphrased away; or, more frequently, the world is made strangely mind‐dependent. It is because of these somewhat scandalous facts that the realist goes in for the “desk‐thumping, foot‐stamping shout of ‘Really!’ ” that Arthur Fine likes to mock (1986a: 129). The realism dispute arises from the age‐old metaphysical question, “What ultimately is there, and what is it like?” I am sympathetic to the complaint that Realism, as part of an answer to this question, is rather boring. Certainly it brings no mystical glow. Nevertheless, it needs to be kept firmly at the front of the mind to avoid mistakes in theorizing about other, more interesting, matters where it makes a difference. Why has metaphysics been conflated with semantics? I suspect that the conflation is very much part of the “linguistic turn” in twentieth‐century philosophy. However, I have a more precise diagnosis of the role given to correspondence truth. Diagnosis 1

Giving Realism* as an example, I have pointed out that we can use the deflationary notion of truth to state a metaphysical view. If the distinction between deflationary and correspondence truth is overlooked, then a doctrine like Realism* may seem to be making the partly semantic claim that certain statements are correspondence‐true; that they “picture” reality. And it seems that the distinction is an easy one to overlook.

Page 5 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * My view that realism does not involve correspondence truth flies so much in the face of entrenched opinion that I shall labor the point. I shall do so (p.37) in the context of the debate over scientific realism. My minimal doctrine, Scientific Realism, is an example of what is sometimes called “entity realism”. The metaphysical doctrine underlying the debate is usually a stronger one, “theory realism”: science is mostly right not only about which unobservables exist, but also about their properties. The following is a fairly typical example of a semantic version of this stronger doctrine: Contemporary Realism: Most scientific statements about unobservables are (approximately) correspondence‐true. Why would people believe this? I suggest only because they believed something like the following two doctrines:

Strong Scientific Realism: Tokens of most unobservable scientific types objectively exist independently of the mental and (approximately) obey the laws of science. Correspondence Truth: Sentences have correspondence‐truth conditions. These two doctrines, together with the equivalence thesis, imply Contemporary Realism. Yet the two doctrines have almost nothing to do with each other. Contemporary Realism is an unfortunate hybrid.

Strong Scientific Realism is a metaphysical doctrine about the underlying nature of the world in general. To accept this doctrine we have to be confident that science is discovering things about the unobservable world. Does the success of science show that we can be confident about this? Is inference to the best explanation appropriate here? Should we take skeptical worries seriously? These are just the sort of epistemological questions that have been, and still largely are, at the centre of the realism debate. Their home is with Strong Scientific Realism not with Correspondence Truth. Correspondence Truth is a semantic doctrine about the pretensions of one small part of the world to represent the rest. The doctrine is the subject of lively debate in the philosophy of language, the philosophy of mind and cognitive science. Do we need to ascribe truth conditions to sentences and thoughts to account for their roles in the explanation of behavior and as guides to reality? Do we need reference to explain truth conditions? Should we prefer a conceptual‐ role semantics? Or should we, perhaps, near enough eliminate meaning altogether? These are interesting and difficult questions,14 but they have no immediate bearing on scientific realism. (p.38) Semantic questions are not particularly concerned with the language of science. Even less are they particularly concerned with “theoretical” language “about unobservables”. Insofar as the questions are concerned with that language, they have no direct relevance to the metaphysical concerns of Strong Page 6 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * Scientific Realism. They bear directly on the sciences of language and mind and, via that, on the other human sciences. They do not bear directly on science in general.15 Many philosophers concerned with semantics and not in any way tainted by antirealism are dubious of the need for a correspondence notion of truth.16 Are there atoms? Are there molecules? If there are, what are they like? How are they related to each other? Strong Scientific Realism says that we should take science's answers pretty much at face value. So there really are atoms and they really do make up molecules. That is one issue. Another issue altogether is about meaning. Do statements have correspondence‐truth conditions? Correspondence Truth says that they do. This applies as much to ‘Cats make up atoms' as to ‘Atoms make up molecules'; indeed it applies as much to ‘The Moon is made of green cheese’. Put the first issue together with the second and we get a third: is ‘Atoms make up molecules' correspondence‐true? My point is that this issue is completely derivative from the other two. It arises only if we are wondering about, first, the meanings of sentences ranging from the scientific to the silly; and about, second, the nature of the unobservable world. Suppose that we had established that Correspondence Truth was right for the familiar everyday language. Suppose further that we believe that atoms do make up molecules, and the like. Then, of course, we would conclude that Correspondence Truth applies to ‘Atoms make up molecules', and the like, and so conclude that such sentences are correspondence‐true. What possible motive could there be for not concluding this? Scientific theories raise special metaphysical questions not semantic ones. Strong Scientific Realism and Correspondence Truth have very different subject matters and should be supported by very different evidence. Underlying Contemporary Realism is a conflation of these two doctrines that has been detrimental to both. Not only are semantic doctrines not constitutive of the metaphysical issue of realism, they are, with one exception to be explored in the next part, almost entirely irrelevant to the assessment of realism.17 (p.39) A lot more needs to be said in describing, let alone defending, my approach to realism. In particular, the bearing of epistemology on realism needs to be considered. Nevertheless, if I am right so far, we can start listing aberrations and naming names. Aberration 1

The way that the realism issue is posed by the British School founded by Dummett is mistaken. The School starts with a properly metaphysical statement of the issue. This is immediately replaced by a formulation in terms of truth, which is then taken not as deflationary but as part of a theory of meaning.18 Page 7 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * Whatever the merits of the various theories of meaning then proposed, the theories are irrelevant (above‐mentioned exception aside) to the metaphysical issue which they are alleged to settle. For the metaphysical issue is not one about meaning. Aberration 2

The fulminations of Richard Rorty (1979) and Arthur Fine (1986a, 1986b) against correspondence truth, taken by them as reasons for putting the realism issue behind us, are in fact irrelevant to realism. For the issue is not about correspondence truth.19 Aberration 3

A great deal of Larry Laudan's “confutation of convergent realism” (1981) is beside the realist point because it is aimed at “assertions about the interrelations between truth, reference and success” (p. 22); in particular, at the idea that we need to see a scientific theory as (nondeflationary‐) true, or its terms as referring, to explain the success of the theory. I agree that there is little to be said for this idea.20 However, Realism does not depend on such arguments in favor of truth and reference. (p.40) Aberration 4

Putnam has produced a model‐theoretic argument (1978: 125–7; 1983: 1–25) against “metaphysical realism” and in favor of “internal realism”. Putnam starts by arguing against the possibility of determinate reference relations to a mind‐ independent reality. As a result, there is no way in which the “ideal” theory—one meeting all operational and theoretical constraints—could be false. So metaphysical realism is “incoherent”. Putnam anticipates and dismisses a response that appeals to the causal theory of reference. All the causal theory does, he claims, is add more theory to the ideal theory: “How ‘causes' can uniquely refer is as much of a puzzle as how ‘cat’ can” (1978: 126). The argument has generated a storm of responses.21 Now whatever the rights and wrongs of this debate, the issue has nothing to do with Realism.22 It has to do with reference, with the theory of representation. Metaphysical realism is a hybrid of something like Realism with something like Correspondence Truth (Putnam 1985: 78; 1987: 15–16). The only part of this hybrid that may be affected by Putnam's argument is Correspondence Truth.23 Indeed, the challenge of Putnam's argument can be posed, and often seems to be posed, in a way that presupposes Realism: a representation is related causally to one mind‐independent entity but causally* to another; which relation determines reference? If it proves very difficult to naturalize reference, then perhaps we should seek a nonreferential theory of mind and language. If we were completely desperate, perhaps we might contemplate giving up naturalism. What we should never countenance for a moment is the idea that “we cut up the world into objects Page 8 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * when we introduce one or another scheme of description” (Putnam 1981: 51). To accept that idea is not to rebuild the boat whilst staying afloat, it is to jump overboard.24 (p.41) Aberration 5

Realism is not about the semantic interpretation of existence claims. It is sometimes claimed that a person's realism about certain entities cannot be simply exemplified in a belief or claim that those entities exist. We must know how such a belief or claim is to be interpreted; realism about cats must come from a certain view of the reference of ‘cat’, for example.25 I have argued against this at length elsewhere.26 Briefly, we do not need to move into a metalanguage discussion of our object language claims to establish ontic commitment. Indeed, if commitment could never be established at the level of the object language it could never be established at all. The idea that talk about the world is unclear and in need of interpretation, yet talk about language and its relation to the world is straightforward on the face of it, reflects the damage of years of living under the linguistic turn. Objection 3. “The Realist's Talk of ‘Mind‐Independent’ Reality is Obscure and Metaphorical. The Only Way to Make Sense of it is to Talk of Meaning and Truth.”

Dummett is a famous example of someone who takes this sort of view. He thinks that any metaphysical view is “a picture which has in itself no substance otherwise than as a representation of the given conception of meaning” (1977: 383); it adds only a “metaphor” (1978: pp. xxviii–xxix). He subscribes to what I call “the Metaphor Thesis” (1984: 201; 1991b: 264): metaphysics beyond meaning is mere metaphor. His support for this sweeping thesis is remarkably inadequate, consisting largely in a discussion of the realism dispute about numbers.27 Though philosophers who make this sort of objection do little to demonstrate the alleged obscurity, I think that we can discern what bothers them. It is indeed hard to know what to make of the usual statements of antirealism. Is it really claimed, for example, that there could not have been stars and (p.42) dinosaurs if there had not been people? If not, in what way, precisely, are dinosaurs and stars supposed to be dependent on us and our minds? No clear answer emerges. However, this obscurity is a problem about antirealism not realism. A first stab at a clear statement of realism denies any dependencies of the world on our minds. An intelligible claim of the mind dependency of the world stands clearly opposed to this realism. An intelligible claim about meaning or truth is beside the metaphysical point. Unintelligible claims can be ignored. Realism does not become unclear because it is sometimes opposed by nonsense. A qualification to this statement of realism is immediately called for and was made: realism allows for the familiar causal interactions between minds and the world. Realism disallows all other dependencies. That is sufficient to Page 9 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * characterize realism about “natural” objects like stones, trees, and cats, but not artifacts like chairs, pens, and cars. These objects owe their nature to the purposes for which we built them or to the way we habitually use them. So realism must be further qualified to allow for that sort of mind dependency too (1991b: 246–58; Ch. 6, sec. 3, in the present volume). The center of the realism debate has always appeared to be over the nature of reality (see discussion of Objections 1 and 2). The realist position on reality is perfectly clear. The antirealist position is often not, but what it is unclear about is reality not something else altogether. The fashionable assumption that antirealism is not about what it appears to be about is highly implausible. It requires argument not casual assertion. We shall return to this issue soon, after considering the bearing of epistemology on Realism.

II: Realism, Epistemology, and Naturalism I have spoken of Realism as a metaphysical doctrine. However, it is also a little bit epistemological. The independence dimension denies that the world is dependent for its existence and nature (except in the familiar ways) on what we believe. This denial is the full extent to which Realism is epistemological. To say this is not to overlook the enormous role that epistemology has played in the realism debate. Realism, like anything else, must be argued for by giving evidence. Historically this has immediately raised an epistemic question: is that evidence good enough? This question has dominated the debate. Skeptical doubts about the evidence have been the main motivation (p.43) for antirealism. However, arguments for and against Realism are one thing, Realism itself is another.28 Aberration 6

It is a mistake to identify realism with the view that we have knowledge of the physical world. The realist cannot, of course, be a thoroughgoing Cartesian skeptic. Nor can anyone who claims to have substantial knowledge about the world, be she paleontologist, plumber, or your ordinary common or garden antirealist. Antiskepticism is almost universal, not distinctive of realism.29 A realist's antiskepticism must, of course, take a special shape: she claims to know about a mind‐independent world. This fact about a realist epistemology is responsible for the only exception to the claim (part I) that semantic doctrines are almost entirely irrelevant to the assessment of Realism (1984: 39–40; 1991b: 44–6). From the equivalence thesis it follows that, if a certain worldly situation obtained, then the sentence, ‘Caesar had five moles', is true. Now suppose that it could be established that the correct semantics should be built around an Page 10 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * epistemic notion of truth; for example, around the notion verifiable. Then it would follow that only if ‘Caesar had five moles' were verifiable did Caesar have five moles. This would threaten Realism. What must the world be like for it to be linked this closely to our cognitive capacities? The most plausible explanation of such a close link might be that the world is in some way mind‐dependent. Thus, a verificationist theory of meaning might lead by inference to the best explanation —by abduction—to antirealism. So the realist should resist verificationism. Note, however, that this does not show that a semantic doctrine is constitutive of Realism; the relation between the doctrines is abduction not entailment. Putnam likes “the metaphor” that “the mind and the world jointly make up the mind and the world” (1981: p. xi). Ernest Lepore and Barry Loewer claim (1988: 470) that the metaphor is cashed in Putnam's identification, following Peirce, of truth with ideal justification.30 The present discussion enables us (p.44) to add to the case already made against this claim, in effect, in discussing Objection 3. The metaphor is about the nature of reality. The closest link we can get between Peircean truth and reality is exemplified in statements like, Caesar had five moles only if ‘Caesar had five moles' is ideally justified. Statements like this have no immediate bearing on the metaphysical dispute about Realism; they entail no position on that dispute. They are brought to bear only when we seek an explanation of why they hold. Furthermore, it is not obvious in advance that the best explanation will be antirealist.

Consider, for example, the analogous claim that follows from a really bizarre theory of truth: Caesar had five moles only if ‘Caesar had five moles' is affirmed by the Pope. On its own, this shows nothing about Realism. When we supplement it with an explanation, we might prefer an antirealist one that has the Pope creating the world or a realist one that ascribes to the Pope divine insight into reality. An argument is needed to get from a theory of truth to antirealism.

Return to the Peircean claim. I think that the realist may well be able to give a plausible explanation of this claim under some interpretations of the vague “ideally justified” (1984: 31–2, 187; 1991b: 35–7, 225). If she could, then clearly Peircean truth poses no problem for Realism and so must be unsuitable to cash Putnam's metaphor. Suppose, however, that the realist could not give an explanation. Then, if the claim is accepted (see below), we must find an antirealist explanation: some account must be given of the dependency of reality on our minds that explains Page 11 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * why this claim is true. But if this account of mind dependency could be given we would have exactly what we wanted to cash the metaphor. The excursion into truth has brought us right back where we started. In sum, if we cannot cash the metaphor nonsemantically, talk of truth cannot be brought to bear on the realism issue. If we can cash the metaphor nonsemantically, we do not need talk of truth to cash it. We are in the position to identify another aberration. Since the metaphysical issue is distinct from epistemic and semantic ones the question arises: which issue should we start with? Traditionally, philosophers started with an epistemic issue and argued for antirealism on the ground that the realist's (p.45) world would be unknowable. Recently, philosophers have tended to start with a semantic issue and to argue for antirealism from verificationism; for example, an abduction along the above lines. But suppose that we start with the metaphysical issue. I have argued elsewhere that we can then establish realism (1984/1991b: ch. 5) and proceed by abduction to nonverificationism: the best explanation of language in a realist world is one involving correspondence truth (1981; 1984/1991b: ch. 6; 1997a: 320–30).31 Which starting place is better? Aberration 7

It is a mistake to start building a metaphysics from epistemology or semantics.32 The realism issue should be settled first. Failing to do so is one of the most pervasive and serious aberrations of the realism debate. Consider the priority of Realism over semantics, for example. The argument for Realism, independent of semantics, is very strong. The argument for verificationism, independent of metaphysics, is very weak. Indeed, that argument seems to rest entirely on a priori reflections about linguistic competence. Why should we believe these claims about what meanings we could grasp and what concepts we could have, particularly since they threaten something as plausible as Realism? What is the basis of these claims? Whence cometh this knowledge (1984: 204–20; 1991b: 268–86)? My view of where to start reflects my naturalism. I take the theory of language to be an empirical, conjectural, theory like all others. So there is no question of giving semantics an unearned privileged position in deciding what there is and what it is like. Perhaps naturalism is needed to justify my view of where to start. If so, so be it (1984: 63–9; 1991b: 75–80; Devitt and Sterelny 1987: 7–9, 225–49; 1999: 9–11, 275–305).33 Aberration 7 is often accompanied by a certain caricature of realism. According to Putnam, realism requires a “God's Eye View” (1981: 74); that we have “direct access to a ready made world” (p. 146) and so can compare theories with “unconceptualized reality” (1979: 611); that we can make “a transcendental match between our representation and the world in itself” (p.46) (1981: 134).34 Page 12 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * According to Rorty the realist believes that we can “step out of our skins” (1982: p. xix) to judge, without dependence on any concepts, whether theories are true of reality.35 But, of course, no sane person believes any of this. What realists believe is that we can judge whether theories are true of reality, the nature of which does not depend on any theories or concepts. Diagnosis 2

What lies behind these bizarre, and always undocumented, antirealist fantasies? 36

I think that the answer is clear: the Cartesian picture.

According to this picture we start the quest for knowledge locked in our minds, contemplating our ideas, and asking the following questions. Is there a world out there causing this inner show? Does it resemble the show? How can our ideas reach out to this world? But the naturalist does not start from scratch with epistemic and semantic questions. Those questions arise when we already have wide‐ranging, well‐based, opinions about the world, opinions derived from common sense and science. The questions arise when we focus on a small part of the world: people. We go on to seek empirical answers to those questions; we seek a naturalistic epistemology and semantics. The theories that result have no special status. Indeed, given our lack of confidence in these areas, the theories should have rather a lowly status. To suppose that we can derive the right metaphysics from epistemology or semantics is to put the cart before the horse. From the naturalistic perspective, the relations between our minds and reality are not, in principle, any more inaccessible than any other relations. Without jumping out of our skins we can have well‐based theories about the relations between, say, Ron and Maggie. Similarly, we can have such theories about our epistemic and semantic relations to Ron and Maggie. Aberration 8

It is a mistake to think that what is distinctive about realism is the view that truth is the aim of science.37 Minimally construed, it is certainly hard for a (p. 47) realist to deny this view. On that construal, the truth in question is deflationary. So, semantically descending, the view is that the aim of science is to discover what entities there are—observables and unobservables—and what they are like. Now any realist (who is not so radically eliminativist in her epistemology as to eschew talk of discovery) will believe that science already has discovered many of the entities there are and approximately what they are like. How then could she plausibly deny that the aim is to discover more? So a basic constraint on a realist epistemology pushes one strongly toward a view of the aim of science.38 However, the view is not constitutive of realism nor even peculiar to realism; it could be held by a Kantian idealist, for example. Of course, it is tempting to go for a more robust construal of the view: take the aim as correspondence truth. How else are we to make sense of the notion of Page 13 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * discovery which seems essential to even the minimal construal? Talk of the aim of science, unlike Realism itself, is concerned with the link between scientists and the world. Ultimately, it demands a semantic theory. Correspondence Truth is an attractive candidate—and one that I favor—but whether it is the right one is a long way from being determined by Realism. And if it is the right one, it can be adopted by antirealists as well. Aberration 9

It is similarly a mistake to think that what is distinctive about realism is the convergent view of scientific progress.39 This is another view that is neither constitutive of Realism nor peculiar to it. Once again, Correspondence Truth is an attractive semantics to use in the explanation, but that has little to do with Realism (1984: 113–17, 155–7; 1991b: 123–7, 172–4). Aberration 10

Though Realism is an overarching doctrine, it is a mistake to think that it is, in any interesting sense, about science, a meta‐theory.40 Of course, theories about the aim of science, and of its progress, are, in an interesting sense, meta‐ theories.41 So also are theories in epistemology, methodology, and semantics. But a theory of this kind that stems from Realism is not Realism. Realism, like any other (p.48) metaphysics, constrains theories on such matters, but it does not determine any particular one. And any particular one may be compatible with many metaphysics. My aim in this chapter has been to distinguish the metaphysical issue of realism from other issues and to urge that the metaphysical issue should be settled first. I have not presented a positive argument for realism here. However, once realism has been disentangled from other doctrines, the argument for it is fairly simple and, it seems to me, as persuasive as anything could be (1984/1991b: chs. 5 and 7; 1988a; Chs. 3 and 4 in the present volume). If this is so, there is little to be said for antirealism. Why then is antirealism so popular? One cause has already been indicated: the Cartesian picture and fear of skepticism (Diagnosis 2). I think that there is another. Diagnosis 3

The clue is to be found in the mystical flavor of many statements of antirealism. Antirealism is like religion.42 Religion is even less plausible than antirealism, is supported by even worse arguments, and yet is even more popular; indeed, religion is not just a hazard of philosophy but a hazard of life. If we could explain religion, I think that we could explain antirealism.43 Postscript to “Aberrations of the Realism Debate”

Page 14 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * My approach to the realism issue, set out in this chapter and developed in those to come, has not gone down well in some circles. Thus, Michael Williams argues that the independence dimension of Realism cannot be understood without a theory of truth. Anthony Appiah has a similar view of the existence dimension. Hilary Putnam, Michael Dummett, Simon Blackburn, Richard Rorty, and Arthur Fine think, in effect, that the approach is “not really philosophy”. I shall discuss these views in turn.

(p.49) 1. The Independence Dimension It is common for philosophers to find the realist's talk of “mind independence” obscure and metaphorical, demanding an unpacking in terms of truth; see the authors cited in the discussion of Objection 3 above. Michael Williams has provided a more recent example, claiming that a view of truth is central to realism: The primitive realist intuition is that the world is objective, meaning that how things are is independent of how we think, or even under certain conditions would think, they are. The obvious way to unpack this intuition is in terms of a certain view of truth, the view that, as Putnam says, truth is a radically non‐epistemic notion. (1993: 193) After describing my contrary view that truth has, near enough, nothing to do with realism (1991b), Williams responds as follows:

One obvious difficulty with such a view is that it makes it even less clear than it otherwise might be what form a debate on the question of realism might take. The other is that any such attempt to identify realism with commitment to a certain body of truths, rather than a view about truth, is bound to misfire. For we have to add the proviso that these truths be accepted at “face value” and explaining how and why this is so will inevitably reinvolve us with questions about what the truth of propositions of common sense and science should be understood to consist in. (1993: 212 n.) Set aside the first “obvious difficulty” for a moment and consider the second. Realism has two dimensions, “existence” and “independence”. It seems that Williams is more concerned about the unclarity of the independence dimension.44 He thinks that we need to unpack this dimension in terms of “a radically non‐epistemic” notion of truth. I disagree. Antirealists propose a variety of mind dependencies for the physical world. A Realist might reject such proposals by simply denying all dependencies of the physical world on our minds, but this is too simple. As noted in responding to Objection 3, the Realist must allow certain familiar causal relations between our minds and the world. If we add this qualification to the simple denial, we have succeeded in characterizing the independence of the paradigm Realist objects; of stones, trees, cats, and the like. This easy characterization makes no mention of notions Page 15 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * of truth. Other physical objects that have a more interesting dependence on us— for example, hammers and money—pose more of a challenge. But, with careful attention to the differences between this sort of (p.50) dependence and the dependence that antirealists allege, the challenge can be met (see Ch. 6, sec. 3, in the present volume). Once again truth plays no role. Return now to Williams's first “obvious difficulty”. What should the realism debate be about if not about notions of truth? It should be about the existence and independence dimensions of Realism. And that is what my discussion is about.

2. The Existence Dimension Some philosophers find Realism's existence dimension unclear, demanding that we turn to semantics for clarification. Antony Appiah (1991) is one.45 In a commentary on the paper that forms the present chapter, he finds my identification of the entities to which Realism is committed inadequate and claims that those entities must be identified as the referents of our representations. As a result he thinks that Realism is, in part, implicitly semantic. What is the problem?46 I identify the entities by giving some examples: stones, trees, cats, electrons, muons, curved space‐time, atoms, and molecules. I call such entities “commonsense, and scientific, physical” in my statement of Realism. What is unclear about that? If a person has difficulty catching onto the idea, I could continue the list, right to the end if necessary (1984: 19; 1991b: 21). Or I could say a word about how we construct the list: we look to our best science and common sense. For, we have good reason to believe in the existence of entities that our best science and common sense say exist. There is no genuine problem identifying the entities that Realism is committed to. This having been said, I have no objection to identifying the entities using reference, as Appiah wants. We could state Realism as follows: Realism#: Most referring expressions of commonsense, and scientific, theories refer to objective and mind‐independent entities However, first, this no more makes Realism semantic than did stating it as Realism* above. And Appiah agrees that Realism* does not make Realism semantic. Where Realism* simply exploits the “disquotational” properties of ‘true’ to paraphrase the metaphysical doctrine of Realism, Realism# simply exploits the disquotational properties of ‘refer’ for that same purpose. That is (p.51) to say, Realism# requires a notion of reference no stronger than one that makes all appropriate instances of

‘a’ refers if and only if a exists hold. (Appropriateness is a matter of synonymy, as before: what is substituted for ‘‘a” must name an expression that is synonymous with what is substituted for ‘a’.) All that is needed for this is a deflationary notion of reference that involves no commitment to Page 16 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * any substantive semantic thesis about the representational properties of language. Indeed, it is compatible with an eliminativist semantics (like Quine's, for example).47

Second, though Realism# is unobjectionable, it is no advance over Realism. If there were a genuine problem of identification in the latter case, there would be an analogous problem in the former. Someone might ask Appiah which expressions in which theories concern Realism#. Appiah could give some examples, just as I did before of entities. Someone who doggedly refused to understand how, and on what basis, I would continue the earlier list could be just as difficult with Appiah and his list. And I would be sympathetic if Appiah felt more than a little bit impatient with such doggedness.

3. “It's Not Philosophy” Many feel that my approach to the realism issue is “not really philosophy”. Two of the most eminent philosophers of our time, Michael Dummett and my old teacher Hilary Putnam (who, ironically, had a role years ago in shaping my approach to the realism issue), have treated my approach particularly harshly. And the approach gives no joy to some other leading philosophers: Simon Blackburn, Richard Rorty, and Arthur Fine. 3.1 Putnam and Dummett

In his recent contribution to the Dummett volume of the Library of Living Philosophers, Putnam (2007) concludes a brief defense of Dummett from my criticisms (1984) as follows: “Devitt's dismissive attitude is as unphilosophical as Samuel Johnson's stone‐kicking” (p. 159). Dummett is delighted: “I very much enjoyed Hilary Putnam's criticism of Michael Devitt's attempted refutation of anti‐realism, and thought it wholly to the point”. Dummett concludes that my (p. 52) argument is “a severe case of ignoratio elenchi” (2007: 184). Both these responses to my argument are careless (to put it delicately). Putnam starts by attributing to me the following view: “the realism issue is simply, ‘Is there a mind‐independent reality or not?’ (thump) and that question has nothing to do with semantics” (2007: 158). This is not precisely my view— see the definition of Realism and accompanying discussion in part I of this chapter—but it captures the spirit of it well enough. After a short digression on Lenin, however, Putnam takes my realism to be a commitment to the view that “the behavior of the stars is independent of human sensation and thoughts and beliefs”. I am alleged to portray antirealists as denying this. What I actually, and quite plainly, portray antirealists as denying is that the existence and nature of the stars are in various ways independent of our minds. But this misrepresentation pales into insignificance beside the following: “Devitt's argument . . . simply assumes—what anti‐realists of course deny—that the anti‐ realist cannot interpret the sentence ‘the behavior of the stars is independent of human sensation and thoughts and beliefs' in a ‘justificationist’ way, interpret it so that it is ‘true’ (in the anti‐realist sense)” (p. 158). This claim is the full extent

Page 17 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * of Putnam's argument against my critique of Dummett. (There is an implicit “thump”.) This is his sole basis for the stone‐kicking charge. So, according to Putnam, I make that mistaken assumption about interpretation and thus conclude that the Dummettian anti‐realist cannot accept (something like) my Realism. Hence, presumably, Putnam thinks that I take my Realism to refute that anti‐realist. This is preposterous! Putnam simply ignores the significance of the most prominent distinction in my discussion of realism, the distinction between metaphysical doctrines like my Realism and any semantic doctrine at all.48 The distinction is, of course, crucial to my discussion of Dummett because Dummett identifies realism with a semantic doctrine: he identifies it with a commitment to sentences having “evidence‐transcendent” truth conditions.49 I have argued at great length against this identification; see, for example, Aberration 1 and Objection 3 above. So, in my view, we have (p. 53) two distinct issues: the metaphysical issue over doctrines like my Realism and the semantic issue over doctrines like Dummett's antirealism. It is then, of course, appropriate to probe the relations between these two issues. Realism and Truth does this extensively in discussions that Putnam seems to have missed.50 First, I do not “simply assume” but rather argue that an epistemic doctrine of truth like Dummett's is likely, though not certain, to lead by abduction to the rejection of Realism (1984: 39–40; 1991b: 44–6). It is very difficult for a Realist to find a plausible epistemology to accompany an epistemic doctrine of truth. But my most important point is that this way of proceeding—from semantics to metaphysics—is precisely the wrong way. Semantics is among the weakest places to start from. We should put metaphysics first; see, for example, Aberration 7 above. When we do put metaphysics first, we can present a powerful argument for Realism that makes no appeal to semantics (1984/1991b: ch. 5; Chs. 3 and 5 in the present volume). Then we see what follows about semantics. I have labored mightily to come up with a good abduction from Realism to the semantic realism of a correspondence theory of truth. The first edition of Realism and Truth proposed one (1984: 73–103, 110–12). I had scarcely sent that off than I had second thoughts. In time these led to the very different abduction in the second edition (1991b: 83–107, 121–3). My confidence in this did not last long either and I proposed another abduction in the “Afterword” (1997a: 320–30, based on my 1996). I still stand by that one. If it is right, then any epistemic doctrine of truth, including Dummett's semantic antirealism, is wrong. So Dummett cannot “interpret” my statement of Realism, indeed any statement of anything, “in a ‘justificationist’ way”.

Page 18 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * (p.54) In sum, I argued against Dummett's identification of the metaphysical issue of realism with a semantic issue. With the issues distinct, I argued that we should proceed from the metaphysics to the semantic. I presented a case for Realism and, from that basis, have given three different arguments for correspondence truth and against Dummettian semantic antirealism. All in all, my critique of Dummett is about as far from dismissal by stone‐kicking as one could get. Putnam and Dummett wonder what I mean by ‘independent’ and see a choice between logical and causal independence. I am quite explicit about what I mean and it is neither of these. I mean constitutive independence: thus I say that the known world “is not constituted by our knowledge, by our epistemic values, by the synthesizing power of the mind, nor by our imposition of concepts, theories, or languages; it is not limited by what we can believe or discover” (1991a, which is the present chapter: 45; see also Objection 3). In contrast, Dummett decides, without apparently bothering to check, that I must mean causal independence. He responds: “Of course the behavior of the stars is causally independent of human sensations, desires and beliefs” (1997: 184). So, he thinks, his antirealism is untouched by my argument for Realism. The problem with this is that, with Putnam's help, he has simply invented the argument he attributes to me and ignored my actual one. So his response is “a severe case of ignoratio elenchi” if ever there was one. Versions of my actual argument have been available in several places for twenty‐five years (1983a; 1984: ch. 12; 1991b: ch. 14). It would be interesting to know his response to those. 3.2 Blackburn

My defense of Realism (1999a) raises a dark vision in Simon Blackburn: he sees that defense as denying that attempts “to present a reflective theory about the relationships between ourselves and the world . . . are coherent”; indeed, he sees it as threatening subtle philosophical positions, the history of philosophy, and even our culture (1999: 51). This is a mistake. There are two sorts of relationships between ourselves and the world that are the subject of “reflective theory” in philosophy: epistemic relationships that are the concern of epistemology and semantic ones that are the concern of semantics. Epistemology and semantics can be as coherent as anything could be and I am all for them, as the present volume often exemplifies. Where I differ from the tradition and, I infer, from Blackburn is in thinking that epistemology and semantics should be done empirically, not a priori, and that they should not be the basis for a position on Realism; cf. Aberration 7. I think this is (p.55) what Blackburn finds so threatening. But why? I think that the answer is that naturalism's empirical approach takes the philosophical fun out of the realism debate: it removes a lot of the debate's excitement, a lot of its aura of significance.51 D. C. Williams made the point eloquently long ago:

Page 19 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * Dare I suggest at last that a kind of highmindedness and sportsmanship have conspired against the vulgar plethora of evidence for realism to protect from bathos the “persistent problems” and the laborious ritual of our profession? To bring such gross implements as Mill's methods to the limpid regions of philosophic discourse is like dynamiting a trout stream. It gets the fish, but it misses all the exquisite impractical pleasure of angling with the thin line of dialectic. Besides, it depletes the game supply. These punctilios may, of course, mean simply a resolve that philosophy must be critical of the most obvious of mundane opinions and methods. But in so far as they are a mood of gratuitous superiority, the philosopher who does not think of philosophy as mere courtly pastime like parchesi will abandon them. The disclaimer of the earthier sorts of knowledge has isolated philosophy, made it a mystery or a jest, an escape from reality or a visionary interpretation. Philosophy is not higher and suprascientific. It is the lowest and grubbiest inquiry round the roots of things, and when it answers real questions about the world, it is and can only be an inductive science. (1966: 146–7) 3.3 Rorty and Fine

The rejection of “the reflective enterprise” is part of a view Blackburn calls “minimalism”. Minimalism combines this rejection with the claim that commonsense and scientific discourse “are self‐interpreting and need no further metatheoretical commentary” (1999: 51). Although it is wrong to attribute minimalism to me, I think it would be right to attribute it, or something close, to Rorty and Fine. I have already indicated some differences with Rorty and Fine; see Aberration 2, particularly. In the present context, it is worth being really explicit about two differences. Rorty and Fine differ from me not only in rejecting the need for the metatheory that interests Blackburn but also in identifying the philosophical issue of realism with that rejected metatheory; cf. Objections 1 and 2 above.52 They make this identification because they think that simply going along with science and common sense hardly constitutes a philosophical position. It is important to see that my Realism does more than just go along. First, there is a critical aspect to its existence dimension: science and common sense may posit entities that, given their purposes, they need not posit (1984: 16–17; 1991b: 18–19). Second, and more important, there is the independence (p.56) dimension. This is a philosophical doctrine par excellence. Now, of course, if one thinks that any attempt to make sense of the independence dimension must appeal to the philosophical metatheory that Blackburn likes and Rorty and Fine do not, then one might think that this metatheory is indeed the philosophical part of the realism debate. But no such appeal is necessary; see section 1 above.

Page 20 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * Notes:

(*) First published in Philosophical Studies (Devitt 1991a). Reprinted with kind permission from Springer. (1) [2009 addition] The 2nd edn. of Realism and Truth (1991b) adds a fifth maxim. The defense of realism was further developed in the “Afterword” that was added to the 2nd edn. when it was reprinted in 1997. (2) Susan Haack (1987) distinguishes nine “senses” of ‘realism’! (3) See also Boyd 1984: 41–2; Fales 1988: 253–4. [2009 addition] Some other examples: Hesse 1967: 407; Hooker 1974: 409; Papineau 1979: 126; Ellis 1979: 28; R. Miller 1987; Jennings 1989: 240; Matheson 1989; Kitcher 1993; J. R. Brown 1994. (4) Two examples are Hilary Putnam's “metaphysical realism” (1978: 123–5), and the account of realism by Arthur Fine (1986a: 115–16, 136–7). (5) [2009 addition] For fairly accessible accounts of these worlds see, respectively: Kant 1783; Kuhn 1962; Dummett 1978: preface and chs. 10 and 14; Goodman 1978; Putnam 1981. (6) For the reasons for the qualification, see my 1984: 16–17, 121–2; 1991b: 18– 19, 131–2. Brian Ellis demonstrates nicely some further reasons for qualification; 1985: 52–8. (7) Cf. my 1984/1991b: maxim 2 and ch. 4; 1997a: 304–7. (8) More needs to be said to allow for the paradoxes, ambiguity, indexicals, and truth value gaps, but these problems are irrelevant to our concerns. (9) For more on this distinction, see Quine 1970: 10–13; Grover et al. 1975; Leeds 1978; Field 1986; Brandom 1988; also Devitt and Sterelny 1987: 162–5; 1999: 202–5. Grasping the distinction is not aided by the fact that deflationary truth is sometimes called correspondence; e.g. Ellis 1985: 53. [2009 addition] See Ch. 8, particularly 8.7, in the present volume for a much more thorough discussion of deflationism. This discussion emphasizes that, according to deflationism properly conceived, the equivalence thesis explains the meaning of the truth term but not the nature of truth. Indeed, truth has no nature to explain. (10) I agree with Putnam (1985: 63) that Tarski's famous article would have been better called “The Nonsemantic Conception of Truth”. (11) Which I mistakenly went along with; 1984: 26–8. (12) [2009 addition] Antony Appiah, in his comments on this paper, still thinks that a truth‐conditional theory including a causal theory of reference entails Page 21 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * Realism (1991: 70–1). I do not see the force of his argument. On this theory, as he says, “it is hardly surprising . . . that most of the terms succeed sometimes in referring to something” (p. 70). Nevertheless, the theory does not entail this. Furthermore, it does not entail that any particular term refers, hence not that any particular entity exists. In brief, it does not yield the existence dimension of Realism. Further still, it does not yield the independence dimension. So it could hardly be further from entailing Realism. Of course, if we add the two dimensions of Realism to the semantic theory, then it entails Realism. But if we added the doctrine of the Trinity, it would entail that too. (This note draws on my 2001d response to Appiah.) (13) For more on the verbal point, see my 1988b: 160–1; 1991b: 40. (14) Cf. my 1984/1991b: maxim 4 and ch. 6. [2009 addition] Ch. 8 of the present volume argues for a correspondence theory. (15) “. . . whatever be the interest in the philosophy of language, it has very little value for understanding science” (Hacking 1983: 45). (16) See e.g. Field 1978; Churchland 1979; Stich 1983. (17) For the reason for the qualification, “almost”, see my 1984: 114–17, 230–1; 1991b: 124–7, 298–9. (18) Two recent examples of this swift move from metaphysics to semantics are Wright 1988: 25–7; Luntley 1988: 1–2. Luntley claims (p. 12) to have “shown” my earlier criticism (1983a: 75–83; 1984: particularly 198–204; 1991b: particularly 261–7) of this move to be wrong. He's bluffing. He has simply repeated the move without supplying any justification at all. (19) I am indebted to Fiona Cowie for many thoughts about Fine. For more on Rorty, see my 1988b and 1991b: ch. 11. Much of Rorty's argument against correspondence truth, and all of Fine's, depend on gratuitously saddling that notion with an implausible epistemology; see my discussion of the God's Eye View in the next part. Fine's “natural ontological attitude” mostly seems just like Realism. However, at times one wonders; 1986b: 163–5. Van Fraassen's claim that Fine's view would be compatible with his “with minor modifications” (1985: 246) should give the realist pause. (20) 1984: 87–91, 106–10, 113–17; 1987; 1991b: 97–101, 113–17, 123–7. See also Levin 1984. See Ch. 4, secs. 3.1 and 4.2, of the present volume for a discussion of Laudan's argument against Scientific Realism. (21) Lewis 1984 is a particularly helpful one. My own response criticized Putnam's dismissal of the causal‐theory solution: the point of the solution is not that the causal theory's use of ‘cause’ determines reference but that causation itself does (1983b; 1984: 188–91; 1991b: 225–9). Putnam was not impressed Page 22 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * (1983: pp. xi–xii, 295–6). [2009 addition] I have responded (1997a: 330–8; 1997b). (22) Cf. e.g. three recent responses to Putnam: Heller 1988; Fales 1988; Lepore and Loewer 1988. See also T. Blackburn 1988: 179. Putnam seems to find the idea of realism without correspondence truth inconceivable (1985: 78; 1987: 15– 16). (23) [2009 addition] Putnam criticizes other views that he associates with metaphysical realism and that are also inessential to Realism. One example is the view that there is exactly one true and complete description of the world (1981: 49), a view which, with Correspondence Truth, is alleged to require “a ready made world” (1983: 211). Another example is a sort of individualistic essentialism (1983: 205–28). Even if Putnam's criticisms of these views were correct, they would leave Realism largely untouched (Devitt 1991b: 229, 245). (24) The idea is an aberration of such magnitude that others pale almost into insignificance beside it. I have not listed it because there is no room to discuss it in this chapter. For some helpful criticisms of the idea and what leads to it, see Aune 1987; Wolterstorff 1987; C. Brown 1988; McMichael 1988; “Philosophy and Lunacy: Nelson Goodman and the Omnipotence of Words” in Stove 1991. See also Devitt and Sterelny 1987/1999: chs. 12–13 and Ch. 5, secs 5–7, in the present volume. (25) S. Blackburn 1980: 354; Fine 1986a: 138–9, 152; 1986b: 175–6. (26) 1983c: 671–2; 1984: 40–6; 1991b: 50–8; 1997a: 307–20; Ch. 7 in the present volume. (27) Dummett 1973: 507–8; 1977: 380–9; 1978: pp. xxiv–xxix, 229–47 (cf. my 1983a: 79–82; 1984: 200–4; 1991b: 263–7). See also S. Blackburn 1980 (cf. my 1983c). Luntley (1988) wrings his hands frequently about the unclarity of talk of mind‐independent reality. Rorty thinks that we need Davidsonian semantics to clarify the talk; 1976: 327; 1982: 12–15 (cf. my 1988b: 165–6; 1991b: 208–9). Barry Taylor spoils a subtle and helpful discussion of what this talk amounts to with a sudden, largely unsupported, move to talk of truth; 1987: 59–61. (28) Cf my 1984/1991b: maxim 1 and ch. 7. (29) Cf. Fine 1986a: 137; Margolis (1986: 113); Fales 1988: 254. Margolis thinks that realism ‘(at least initially) refuses to disjoin so‐called ontological and epistemological questions' (p. 111). I think that the reverse is the case. Blurring the distinction between ontology and epistemology is a mark of the antirealist. (30) Ellis (1985) subscribes to a Peircean notion of truth and yet seems to agree with much of Realism. However, at bottom, his position is antirealist: “there is no way that the world is absolutely, only ways in which it is relative to various kinds Page 23 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * of beings” (p. 71). How does he arrive at this view? He believes that choice of the ideal theory, hence truth, is not entirely determined by empirical evidence but partly by “pragmatic” considerations which are our “epistemic values” (p. 68). So, he concludes, the world is dependent on those values. I assume that the intermediate step here involves the equivalence thesis in the way I illustrate. (31) [2009 addition] So Appiah is wrong in thinking that, on my view, Realism “should make no difference to our semantical beliefs” (1991: 68). Different metaphysics can make different semantics seem plausible or implausible; in particular, I have argued that Realism makes verificationism seem implausible. Similarly, as we have seen, different semantics can make different metaphysics seem plausible or implausible; in particular, verificationism can make Realism seem implausible. (32) Cf. my 1984/1991b: maxim 3 and ch. 5. (33) [2009 addition] “Putting Metaphysics First” is a pervasive theme of the present volume; see particularly, Ch. 3, secs. 4–6; Ch. 5, sec. 6; Ch. 15, sec. 4. (34) Putnam attributes this sort of idiocy to realist friends “in places like Princeton and Australia” (1979: 611). The British School has more bad news for Australians (particularly black ones): “there is no sense to supposing that [Australia] either determinately did or did not exist [in 1682]” (Luntley 1988: 249–50). (35) See also 1979: 293; Fine 1986a: 131–2; 1986b: 151–2. (36) Bill Lycan has nicely mocked the fantasies with his name “Turtle Realism”: antirealists should go all the way and accuse realists of believing that the earth sits on the back of a giant turtle (1988: 191). Australian realists believe that the turtle sits on the back of a giant crocodile: “Crocodile Realism”. (37) Cf. van Fraassen 1980: 8. [2009 addition] See Ch. 4, n. 10, in the present volume for more on the mistake of defining realism in terms of the aims of science. (38) “The different views about the aim of science lead naturally to different views concerning its theoretical achievements” (Ellis 1985: 48). I think that this is the wrong way round as an account of both the actual and rational order of thought. (39) Cf. S. Blackburn 1980; Laudan 1981; Boyd 1984. (40) Cf. van Fraassen 1985: 255 n.; Fine 1986a: 113–14, 145; 1986b: 150, 161, 171.

Page 24 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * (41) Fine is dismissive of all such theories about science because he sees them as requiring a stance outside science (1986a: 147–8). The theories do not require this stance. Science can be discussed from within science. See also Ch. 4, n. 23, in the present volume. (42) For some other speculations along these lines, see “Idealism: A Victorian Horror Story” in Stove 1991. [2009 addition] See also my 1991b: 257 n. 11, for a suggestion of Georges Rey's. (43) My thanks to Fiona Cowie for comments on an earlier draft. The version of this chapter that was distributed, and partly delivered, at the Oberlin Colloquium (Apr. 1989), included a discussion of van Fraassen and scientific realism. This has been dropped because of space limitations. Parts of the longer paper were also delivered at Davidson College and the University of North Carolina at Chapel Hill (Apr. 1989) and at the University of Maryland Baltimore County (May 1989). (44) I draw here on my 1997a: 304–7. I respond to Williams's possible concern about the unclarity of the existence dimension in my 1996b, which is Ch. 7 of the present volume, sec. 3. (45) See also Christopher Gauker's problem (2006: 127) with my talk of “unobservables” in defining Scientific Realism. I respond in Ch. 4, n. 4, of the present volume. (46) I draw here on my response (1991d) to Appiah. (47) See Ch. 7 in the present volume for a lot on deflationary truth and a little on deflationary reference. (48) Putnam's Lenin scholarship is faulty too. He rightly supposes (2007: 166 n. 12) that the title of my early criticism, “Realism and the Renegade Putnam” (1983b), was a play on the title of a famous article by Lenin. However, that title was not “Marxism and the Renegade Kautsky” but rather “The Proletarian Revolution and the Renegade Kautsky”. (49) In Realism and Truth (1984: 198–200; 1991b: 261–3), I cite evidence that Dummett makes this identification. In the “Afterword” to that book (1997a: 307), I find further evidence in Dummett's valedictory lecture in Oxford (1993: 468). Panu Raatikainen has drawn my attention to an earlier part of that lecture where Dummett seems, however, to accept that the metaphysical and semantic issues are distinct whilst finding the semantic one more interesting (p. 465). Indeed, as I noted in the book, “Dummett attaches no significance to the difference between these two views” of the issues (1984: 199; 1991b: 262). The view that the semantic issue has a certain priority over the metaphysical one is certainly more reasonable than the identification of the issues but a central tenet of my Page 25 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Aberrations of the Realism Debate * book is that the view is very wrong (except, perhaps, in the realm of Dummett's favorite example, mathematics; 1984: 200–4; 1991b: 263–7). (50) Curiously, Putnam is not alone in missing these discussions. (i) Alexander Miller (2003), who agrees with my rejection of the Metaphor Thesis and of Dummett's identification of the realism issue with a semantic issue, thinks that I have “overlooked” (p. 192) “that Dummett's arguments against semantic realism can be viewed as attempting to establish that common‐sense realism cannot be conjoined with [the Truth‐Conditional Conception of meaning and understanding]” (p. 207). (ii) Drew Khlentzos—whose 2004 is, according to Putnam, “a convincing criticism” of my response to Dummett (p. 166 n. 10)— includes me among philosophers who think that “all they need do to disarm the antirealist's challenge is show that the metaphysical issue of realism has nothing to do with disputes about the nature of truth” (16). In fact, to repeat, I consider the relationship between the metaphysical and semantic issues at some length (see also 1999a: 93–8). Another point to note in assessing Khlentzos's criticisms (31–5) is that my commitment to “evidence‐transcendent” truth conditions is only a commitment to the view that “statements have truth conditions that are not in any way constrained by our epistemic capacities (Maxim 5). So it is possible that a statement might be true and yet we might not be able to detect this (which is not to say that the truth of any true statement is actually undetectable; 3.5, 7.4)” (1991b: 260). (51) I draw here on my response to Blackburn (1999b). (52) My differences with Rorty emerge nicely in his St Andrews comments on my view (2006: 239).

Access brought to you by:

Page 26 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Underdetermination and Commonsense Realism *

Putting Metaphysics First: Essays on Metaphysics and Epistemology Michael Devitt

Print publication date: 2009 Print ISBN-13: 9780199280803 Published to Oxford Scholarship Online: February 2010 DOI: 10.1093/acprof:oso/9780199280803.001.0001

Underdetermination and Commonsense Realism * Michael Devitt (Contributor Webpage)

DOI:10.1093/acprof:oso/9780199280803.003.0004

Abstract and Keywords This chapter defends commonsense realism from the underdetermination theses of extreme skepticism. If true, these would count against nearly all our knowledge and hence undermine realism about the observable world. The traditional responses to these theses rest on a priori knowledge. The chapter rejects a priori knowledge, but even if there were such knowledge, these traditional responses tend to involve bizarre metaphysics and to be otherwise unsatisfactory. Instead, it offers a Moorean response: realism is much more firmly based than the epistemological theses that are thought to undermine it. The Moorean response is supported by a naturalistic one that appeals to scientific practice. Rather than proceeding form an a priori epistemology to an a priori metaphysics, we should proceed from an empirical metaphysics to an empirical epistemology. We should put metaphysics first. Keywords:   underdetermination, commonsense realism, skepticism, a priori, Moore, naturalism, Moorean response

1. Introduction The underdetermination of theories by evidence often leads to skepticism about the theories and hence to antirealism about the worlds described by the theories. At its most extreme, this skepticism is about “Commonsense Realism”, about our knowledge of the observable world of stones, trees, cats, and the like. This skepticism must spread to “Scientific Realism”, to our knowledge of the unobservable world of atoms, viruses, photons, and the like. But underdetermination sometimes leads to a less extreme view aimed only at Page 1 of 10

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Underdetermination and Commonsense Realism * Scientific Realism: against a background acceptance of realism about the observable world, realism about unobservables is rejected. I shall set this criticism of Scientific Realism aside until Chapter 4 of this volume. The underdetermination theses that concern us claim that a theory (belief) has rivals that stand in some sort of equivalence relation to it with respect to certain evidence. The theses vary with the equivalence relation and with the evidence. In particular, sometimes the relation is deductive and sometimes ampliative; and sometimes the evidence is the actual given evidence and sometimes it is some sort of possible evidence. There are two dimensions to the realisms that are challenged by underdetermination. “The existence dimension” of Commonsense Realism is a commitment to the existence of most observables such as stones, trees, and cats and to these entities having, for the most part, the properties attributed to them by science (p.58) and common sense. Typically, idealists, the traditional opponent of realists, have not denied this dimension; or, at least, have not straightforwardly denied it. What they have typically denied in response to the skeptical challenge is “the independence dimension”. According to some idealists, the entities identified by the first dimension are made up of mental items, “ideas” or “sense data”, and so are not external to the mind. In recent times, under the influence of Kant, another sort of idealist has been much more common. According to these idealists, the entities are not, in a certain respect, “objective”: they depend for their existence and nature on the cognitive activities and capacities of our minds. Realists reject all such mind dependencies. Relations between minds and those entities are limited to familiar causal interactions long noted by the folk: we throw stones, plant trees, kick cats, and so on. We could say a lot more to make these doctrines precise and I have done so elsewhere.1 But these definitions will suffice for our purposes. Let us now consider whether underdetermination has any consequence for Commonsense Realism.

2. Extreme Skepticism In the First Meditation Descartes famously doubted the evidence of his senses. We can see this as an argument about underdetermination. Descartes believes that he is sitting by the fire. But perhaps he is suffering from an illusion, perhaps he is dreaming, perhaps he is being stimulated by an evil demon. There seem to be a range of alternative hypotheses to Descartes's belief, each equally compatible with the evidence available to him: his belief is underdetermined by the sensory evidence. This is the first step in the argument of the extreme skeptic. We do not need to imagine evil demons and the like to find support for this underdetermination: it is supported by our psychological and neurophysiological theories of perception. Consider this example: I look in front of me and come to believe that there is a Page 2 of 10

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Underdetermination and Commonsense Realism * cat there. Our scientific explanation of this is roughly: the cat reflects light waves which provide stimulus to my retina causing sensory neurons to fire leading to my belief. But that scientific account also tells us that the cat is not necessary for the belief, for a multitude of reasons. The cat is not necessary for the light waves that strike the retina: the waves might have (p.59) other causes. The light waves are not necessary for the stimulus: there are other ways of providing that stimulus. Similarly the stimulus is not necessary for the neuron firings and those firings are not necessary for the belief. So the belief has a range of rivals that are compatible with the sensory evidence. Let us call beliefs like Descartes's and mine—beliefs about the external environment caused by perception—“observational beliefs”. We seem to have established the following basic underdetermination thesis: D1: Any observational belief has rivals that are equally compatible with the actual given sensory evidence for that belief.2 D1 needs some clarification. What exactly is “the sensory evidence”? We might take it to be the perceptual experiences themselves, but then it is not clear what it is for an observational belief to be “compatible with” the evidence. If we suppose that perceptual experiences produced beliefs about ideas or sense data then we could take beliefs about them to be the sensory evidence; for example, the belief that the sense datum I am now perceiving is cat‐like. But the supposition is controversial at best. We can be more ontologically cautious, taking the sensory evidence to be beliefs, prompted by perceptual experience, about how things appear (with no commitment to mental entities); for example, the belief that it appears to me that there is a cat in front of me. Then we can take D1 to be claiming that this belief is logically consistent not only with the belief that there is a cat in front of me but with many other beliefs. That is surely true. We move to the second step in the extreme skeptic's argument. Given D1, how could an observational belief be justified? What basis is there for eliminating rivals that are equally compatible with the evidence? The skeptic's position is that there is no basis. She infers the troubling epistemological thesis: A1: Any observational belief has rivals that are equally supported by the actual given sensory evidence for that belief.3 D1 is concerned with the deductive relation of logical compatibility between the sensory evidence and observational beliefs; hence the ‘D’ in its name. A1 is a much stronger underdetermination thesis because it is concerned with the ampliative relation of epistemic support between the sensory evidence and observational beliefs; hence the ‘A’ in its name.4 If A1 is (p.60) correct, even ampliative inferences like induction and abduction (inference to the best explanation) do not provide a basis for preferring any observational belief over many alternative hypotheses. Indeed, the Page 3 of 10

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Underdetermination and Commonsense Realism * skeptic is dubious of all such inferences. Our observational beliefs about stones, trees, cats and the like are all unjustified. Commonsense Realism must be abandoned.

Suppose that we found some way to reject A1 and justify our observational beliefs. We still have a way to go to escape extreme skepticism. We want to move from observational beliefs to singular beliefs about unobserved objects and to general beliefs that cover such objects; to take a boring but familiar example, we want to move from the evidence of many ravens all observed to be black to the belief that Oscar, an unobserved raven, is black, indeed to the “theory” that all ravens are black. The skeptical tradition once again presents us with an underdetermination problem: D2a: Any theory has rivals equally compatible with the actual given observational evidence for that theory. This seems to be indubitably so. Thus our evidence of many sightings of ravens is compatible not only with the theory that all ravens are black but also with the theory that all ravens are black except the ones on island X which has never been investigated. Another related underdetermination thesis also causes trouble:

D2b: Any theory has rivals that entail the same actual given observational evidence. Thus the theories that all ravens are black and that they are all black except on X both entail our evidence of raven sightings.

These deductive underdetermination theses are the first step in the extreme skeptic's argument. She then, once again, infers an ampliative thesis. Given these deductive theses, how could a theory be justified? She arrives at another troubling epistemological thesis: A2: Any theory has rivals that are equally supported by the actual given observational evidence for that theory. According to this ampliative underdetermination thesis, a theory enjoys no epistemic support over some of its rivals.5 Our general beliefs about (p.61) stones, trees, cats and the like are unjustified. Again Realism must be abandoned.

We note that the pattern of the skeptic's argument is: Deductive underdetermination → ampliative underdetermination → antirealism. Now, it might well be objected that my discussion does not do skepticism justice for it takes skepticism to be committed to A1 and A2. The cautious skeptic would avoid this commitment, like almost all others, simply putting the onus on the nonskeptic to undermine A1 and A2. I think this is a good objection but, in the end (section 6), we shall see that it makes no difference whether the skeptic is committed or not. Meanwhile, for convenience, I shall take the skeptic to be committed.

Page 4 of 10

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Underdetermination and Commonsense Realism * 3. “First Philosophy” Responses We could accept the inferences to A1 and A2 and give up on our knowledge of the world: we do not know what there is and what it is like. This abandons the existence dimension of Realism and is very unappealing. The traditional responses of “First Philosophy” were different. One response started by seeking a more basic area of knowledge than (what I have called) our observational beliefs, an area that was not open to skeptical doubt and that could serve as a foundation for all or most claims to knowledge. This foundation was found in the sensory evidence for our observational beliefs: we were thought to have indubitable knowledge of our own ideas (sense data); this knowledge was not underdetermined. Even if we go along with this highly dubious claim, we still have to get from this foundation to knowledge of the world of stones, trees, cats, and the like, thus rejecting A1 and A2. In attempting to solve this problem, foundationalists nearly always gave up the view that the world is external to the mind, thus abandoning the independence dimension of Realism. It was thought that only by constituting the world somehow out of ideas could we hope to save our knowledge of it. Realism leaves a “gap” between our ideas and the world that makes knowledge of the world impossible. Idealism closes the gap by bringing the world into the mind. Another traditional response, currently much more popular, is also idealist. It seeks to reject A1 and A2 by taking the world to be partly constituted by the mind's imposition of concepts, theories, or languages. We can know about (p. 62) that world because we partly create it. Realism's independence dimension is abandoned once again.6 So the price of saving our knowledge in the face of underdetermination and skepticism was typically an idealist metaphysics of one sort or another. Even if we were prepared to pay the price of such a bizarre metaphysics, these responses would not be too convincing. Although First Philosophy aims to take skepticism seriously and hence meet the very demanding skeptical standards for rational belief, it often seems to fall short of those standards: it assumes what no self‐respecting skeptic should allow (for example, indubitable knowledge of ideas). So even with an idealist metaphysics we still seem not to have the knowledge we want. I have argued for this, and against idealism, elsewhere (1984, 1991b, 1997a, 1999a, and 2001a, which is Ch. 5 in the present volume). We are faced with a choice between skepticism and idealism. Surely something has gone seriously wrong. It is time to think again. I shall first make a Moorean response to the skeptical challenge, then a naturalistic one.

4. A Moorean Response7 The disaster has come from the epistemological theses A1 and A2 which make Commonsense Realism untenable: we are supposed to doubt our commonsense beliefs in an external world. But why should we accept these skeptical theses? Page 5 of 10

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Underdetermination and Commonsense Realism * How much confidence should we have in a view that undermines Realism? Realism is a compelling doctrine almost universally held outside intellectual circles. From an early age we come to believe that such objects as stones, cats, and trees exist. Furthermore, we believe that these objects exist even when we are not perceiving them, and that they do not depend for their existence on our opinions nor on anything mental. This Realism about ordinary objects is confirmed day by day in our experience. It is central to our whole way of viewing the world, the very core of common sense. A Moorean point is appropriate: Realism is much more firmly based than the epistemological theses A1 and A2 that are thought to undermine it. We have started the argument in the wrong place: rather than using A1 and A2 as evidence against Realism, we should use Realism as evidence against A1 and A2. We should, as I like to say, “put metaphysics first”. (p.63) Descartes puts us in an armchair and asks us to start by clearing our minds of all knowledge and doing some epistemology. The Moorean puts us in an armchair and asks us to start by assessing the evidence for Realism. In so doing we must resolutely decline to theorize about standards of good and bad evidence, for that epistemological path was what led to the disaster: we simply apply our ordinary evidential standards, just as we presumably did in childhood when we became Realists in the first place.8 Once we have done that, we turn to the epistemological theses A1 and A2. In assessing them we have little to go on but the skeptical argument that puts these theses at odds with Realism. We ask: is it more likely that the theses are mistaken than that Realism is? Is it more likely that there is a flaw in the skeptical argument than that, contrary to what we have always supposed, we lack knowledge of the external world? The Moorean answers these questions with a resounding “Yes”.

5. A Naturalistic Response What else do we have to go on in assessing A1 and A2? D1 does not entail A1, nor do D2a and D2b entail A2. Where can we look for more evidence? Not, First Philosophy assumes, to empirical science, for science itself is doubted by the extreme Cartesian skeptic. So the evidence must be of some nonempirical sort. Thus the various idealist positions that rejected the theses were thought to be, like mathematics and logic, known a priori. The a priori approach is the very essence of First Philosophy and its response to underdetermination. Reflecting from the comfort of armchairs First Philosophers decide what knowledge must be like, and from this infer what the world must be like. If the world were the way the Realist says it is, we could not know about it. Yet, it is typically thought, we surely do know about it. So the Realist cannot be right. The Moorean response alone casts doubt on any such arguments. Given the strength of Realism, it is simply not plausible that we could know something a priori that undermined it, whether that something is the skeptic's A1 and A2 or the idealist's response to them. But the Moorean response is not, of course, Page 6 of 10

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Underdetermination and Commonsense Realism * sufficient. We need to do better and we can. Manifestly we could not have a priori knowledge damaging to Realism if we could not have a priori knowledge at all. According to Quinean naturalism, we could not: there is no a priori knowledge. There is only one way of knowing, the empirical way that (p.64) is the basis of science (whatever that way may be); in Quine's vivid metaphor, the web of belief is seamless.9 So we could not know A1 and A2 a priori because we could not know anything a priori (1996a, 1998, which is Ch. 12 in the present volume, 2005a, b, and 2009c, which is Ch. 13 in the present volume). So, we should now consider A1 and A2 from a naturalistic perspective.

6. Naturalism and the Underdetermination Arguments Naturalism is an overarching epistemological doctrine claiming that the only way of knowing anything is the empirical way of science: for each area of knowledge x, naturalized x. When the area is physics, this yields naturalized physics, when the area is biology, it yields naturalized biology, and when the area is epistemology, it yields naturalized epistemology. Everyone believes in a naturalized physics. Everyone but a few benighted creationists in places like Kansas believes in a naturalized biology. But those in the tradition of First Philosophy do not believe in a naturalized epistemology. The radical consequence of naturalism is that philosophy, including epistemology, becomes continuous with science. From this naturalistic perspective, the troubling epistemological theses, A1 and A2, have no special status. They have to be assessed empirically, contrary to the assumptions of First Philosophy, because there is no other way to assess them. As empirical theses, they do not compare in evidential support with our view of stones, trees, cats, and the like. Experience has taught us a great deal about such objects but rather little about how we know about them. So epistemology is just the wrong place to start the argument: it is one of the weakest threads in the web of belief (cf. Quine's vivid image of Neurath's boat).10 Instead, we should start with an empirical metaphysics and use that as the basis for our naturalized epistemology, as the basis for our empirical study of what we can know and how we can know it. Instead of the traditional pattern of argument, exemplified by the underdetermination arguments against Realism, a priori epistemology → a priori metaphysics, (p.65) we should follow the pattern,

empirical metaphysics → empirical epistemology. The underdetermination arguments not only use the wrong methodology, they proceed in the wrong direction.

Proceeding empirically in the right direction, we start with metaphysics. Realism is then irresistible. Indeed, it faces no rival we should take seriously.11 We then turn to naturalized epistemology. This is a very difficult matter. Still, with Page 7 of 10

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Underdetermination and Commonsense Realism * Realism established, we already know that A1 and A2 are false. We can go a bit further. It is clear from scientific practice that we are entitled, despite D1, to dismiss implausible theses like the evil demon one; and that we are usually entitled, despite the equivalence that a theory has with some rivals according to D2a and D2b, to prefer that theory to its rivals; we are entitled to believe that all ravens are black, for example. The epistemic standards implicit in scientific practice clearly give us these entitlements. We would like to know, of course, exactly what those standards are but it has proved notoriously difficult to say. Nevertheless, it is indubitable that, whatever the standards are, they give us these entitlements. Finally, we must consider the objection raised in section 2. For convenience, we have taken the skeptic to be committed to A1 and A2. The objection is that this does not do the skeptic justice. The cautious skeptic would be dubious of these epistemological theses as of all other substantive theses. After all, she is a skeptic. Since she is not committed to the theses she is not committed to knowing them a priori. So what then is the nature of her challenge to Realism? She puts the onus on the Realist to justify his rejection of A1 and A2. She does not boldly assert the badness of ampliative inferences, thus embracing skeptical standards of justification. She simply points to these epistemic standards, which yield A1 and A2, and asks for a justification for ruling the standards out in favor of Realist alternatives that reject A1 and A2. From our naturalistic perspective the challenge of this cautious skeptic is no more difficult to meet than that of the incautious one. For what I have presented is just what the cautious skeptic wants: a case against A1 and A2 and against any epistemic standards that would sustain them. On the one hand, the strength of the case for Realism counts against them. On the other hand, the practices of science count against them. And these practices are the only place to look in assessing epistemic standards because there is no a priori knowledge. These practices support the use of ampliative inferences for preferring a theory (p.66) to many, if not all, of its rivals. Sometimes, of course, a theory will face a rival that cannot be ruled out in this way but it is not the case that all theories always face such rivals. Commonsense Realism is not threatened by the underdetermination that remains.

7. Conclusion We have considered the underdetermination theses A1 and A2 of extreme skepticism. If true these would count against nearly all our knowledge and hence undermine Commonsense and Scientific Realism. The traditional responses of First Philosophy to these theses rest on a priori knowledge. I have argued elsewhere that there is no such knowledge. Even if there were, these traditional responses tend to involve bizarre metaphysics and to be otherwise unsatisfactory. Instead, I urge a naturalistic response that gives priority to metaphysics over epistemology. The case for Realism is then irresistible and Page 8 of 10

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Underdetermination and Commonsense Realism * alone shows that A1 and A2 must be false. Furthermore, the epistemic standards implicit in science count against these theses and hence against extreme skepticism. Notes:

(*) This chapter draws heavily on part I of “Underdetermination and Realism” (Devitt 2002a) and appears with the kind permission from Wiley‐ Blackwell. Part I of that publication includes a discussion of a priori knowledge which has been omitted from this chapter; but see Chs. 12 and 13. Part II discusses underdetermination and scientific realism, on which see Ch. 4, sec. 3.1 and Postscript. (1) 1991b, 1997a, 1999a. My definitions are unfashionable in not being, or even appearing to be, semantic. I argue that it is very important to disentangle the metaphysical doctrine of realism from any semantic doctrine. See also Ch. 2, part I, in the present volume. (2) Note that the thesis concerns the actual given evidence. There are much stronger underdetermination theses that concern possible evidence. See Ch. 4, sec. 4.1, in the present volume. (3) Quine thinks this likely even given all the sensory evidence that there will ever be, “man's surface irritations even unto eternity” (1960: 23). (4) Larry Laudan emphasizes the importance of the distinction between the deductive and the ampliative in his helpful discussion of underdetermination theses (1996: ch. 2). (5) Laudan calls this “the nonuniqueness thesis”. A stronger thesis, “the egalitarian thesis” claims that a theory enjoys no support over all of its rivals. He is sadly persuasive in attributing the latter thesis to Quine, Thomas Kuhn, and Mary Hesse (1996: 33–43). (6) E.g. Kuhn 1962 (on which see Hoyningen‐Huene 1993); Feyerabend 1975, 1981a; Goodman 1978; Putnam 1981; Latour and Woolgar 1986. See Ch. 5 of the present volume for discussion. (7) The discussion in the next three sections takes a similar line to that against constructivism in Ch. 5, sec. 6, and against semantic ascent in Ch. 15, sec. 4, of the present volume. (8) Devitt 1984/1991b: sec. 5.7, in effect, does this (albeit without emphasizing that the independence of the world is also confirmed day by day in experience). (9) See particularly Quine 1952: pp. xi–xvii; 1961: 42–6. Quine uses ‘naturalism’ to stand for this epistemological doctrine. Others use it to stand for a reductive metaphysical doctrine like physicalism. Page 9 of 10

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Underdetermination and Commonsense Realism * (10) For details of the image, see Ch. 5, sec. 6, of the present volume. (11) Devitt 1984/1991b: sec. 5.9, demonstrates the difficulties of arguing for antirealism from a naturalistic perspective.

Access brought to you by:

Page 10 of 10

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism *

Putting Metaphysics First: Essays on Metaphysics and Epistemology Michael Devitt

Print publication date: 2009 Print ISBN-13: 9780199280803 Published to Oxford Scholarship Online: February 2010 DOI: 10.1093/acprof:oso/9780199280803.001.0001

Scientific Realism * Michael Devitt (Contributor Webpage)

1. Introduction What is scientific realism? The literature provides a bewildering variety of answers. I shall start by addressing this question (sec. 2). I shall go on to discuss the most influential arguments for and against scientific realism. The arguments for are the “success argument” and related explanationist arguments (sec. 3). The arguments against are the “underdetermination argument” which starts from the claim that theories always have empirically equivalent rivals; and the “pessimistic meta‐induction” which starts from a bleak view of the accuracy of past scientific theories (sec. 4). My approach is naturalistic.

2. What is Scientific Realism? Science appears to be committed to the existence of a variety of unobservable entities—to atoms, viruses, photons, and the like—and to these entities having certain properties. The central idea of scientific realism is that science really is committed and is, for the most part, right in its commitments. As Hilary Putnam once put it, realism takes science at “face value” (1978: 37). So, for the most part, those scientific entities exist and have those properties. We might call this the “existence dimension” of realism. It is opposed by those who are skeptical that science is giving us an accurate picture of reality. Scientific realism is about unobservable entities. Science appears also to be committed to lots of observable entities—to a variety of plants, mollusks, (p.68) moons, and the like. Folk theory appears to be committed to observables like stones, trees, and cats. A skepticism that extends to observables is extreme, “Cartesian”, skepticism. It yields the issue of “realism about the external world”. This issue is both different from and prior to the issue of scientific realism. It addresses doubts about the very clearest cases of knowledge about observables, Page 1 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * doubts occasioned by skeptical hypotheses such as that we are manipulated by an evil demon. The issue of scientific realism arises only once such doubts about the observable world have, somehow or other, been allayed. Given the obvious truth of the following weak underdetermination thesis, WU: Any theory has rivals that entail the same actual given observational evidence, allaying those doubts will involve accepting some method of nondeductive ampliative inference. Not even a theory about observables can be simply deduced from any given body of evidence; indeed, not even the very existence of an observable can be deduced “from experience”. If we are to put extreme skepticism behind us and gain any knowledge about the world, we need some ampliative method of inference.1 Armed with that method, and confident enough about the observable world, there is thought to be a further problem believing what science says about unobservables. So the defense of scientific realism does not require that we refight the battle with extreme skepticism, just that we respond to this special skepticism about unobservables. We shall see that this point has not been kept firmly enough in mind.

The general doctrine of realism about the external world is committed not only to the existence of this world but also to its “mind independence”: it is not made up of “ideas” or “sense data” and does not depend for its existence and nature on the cognitive activities and capacities of our minds. Scientific realism is committed to the unobservable world enjoying this independence. We might call this the “independence dimension” of realism. The very influential philosophers of science, Thomas Kuhn and Paul Feyerabend, think that scientific entities are not independent but are somehow “constructed” by the theories we have of them. This “constructivism” has its roots in the philosophy of Kant and is extremely influential. An important feature of constructivism, for the purposes of this paper, is that it applies in the first instance to observables: there is no special problem about the independence of (p.69) unobservables (as there is thought to be about their existence).2 The struggle between constructivism and realism is appropriately conducted at the level of observables. I shall therefore not engage in it here.3 Before attempting a “definition” of scientific realism, some further clarification is called for. First, talk of the commitments “of science” is vague. In the context of the realism debate it means the commitments of current scientific theories. The realist's attitude to past theories will be the concern of section 4.2. Second, the realist holds that science is right, “for the most part”. It would be foolhardy to hold that current science is not making any mistakes and no realist would hold this. Third, this caution does not seem to go far enough: it comes too close to a blanket endorsement of the claims of science. Yet scientists themselves have many epistemic attitudes to their theories. These attitudes range from outright disbelief in a few theories that are useful for predictions but known to be false, through agnosticism about exciting speculations at the frontiers, to a strong commitment to thoroughly tested and well‐established theories. The realist is Page 2 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * not less skeptical than the scientist: she is committed only to the claims of the latter theories. Furthermore, realism has a critical aspect. Theories may posit unobservables that, given their purposes, they need not posit. Realism is committed only to “essential” unobservables. In brief, realism is a cautious and critical generalization of the commitments of well‐established current theories. More clarification would be appropriate but this will have to do.4 Utilizing the language of the clarification we can define a doctrine of scientific realism as follows: SR: Most of the essential unobservables of well‐established current scientific theories exist mind‐independently. (p.70) With a commitment to the existence of a certain unobservable goes an implicit commitment to its having whatever properties are essential to its nature as that unobservable. But, beyond that, SR is noncommittal on the properties of the unobservables, on the scientific “facts”. Yet the scientific realist is often committed not only to the entity realism of SR but to a stronger “fact” realism:5

SSR: Most of the essential unobservables of well‐established current scientific theories exist mind‐independently and mostly have the properties attributed to them by science. The existence dimensions of these doctrines are opposed by those who are skeptical of what science is revealing; the independence dimension is opposed by the constructivists.

Although not generally skeptical of scientific theories, SR and SSR do reflect some skepticism. By varying the amount of skepticism, we could define some other doctrines; for example, instead of claiming that most of the unobservables exist we could claim that a large proportion do or, even weaker, that some do. Clearly there is room for argument about how strong a position should be defended against the skeptic. Related to this, but less interesting, there is room for argument about which doctrines warrant the label “scientific realism”. But this does not prepare one for the bewildering variety of definitions of scientific realism in the literature, many of them very different from SR and SSR. SR and SSR are about what the world is like, they are metaphysical (or ontological). Some philosophers favor epistemic definitions of scientific realism (for example, Kukla 1998: 10; see also Psillos 1999: pp. xix–xxi). Thus, instead of claiming that most of the unobservables of science exist, one could claim that a belief that they do is justified; or, instead of claiming SSR, one could claim that SSR is justified. This illustrates that epistemic definitions (p.71) are generally parasitic on metaphysical ones. And although the epistemic ones are clearly different from metaphysical ones, they are not different in a way that is significant for the realism debate. For, if one believes that, say, SSR is justified, one should believe SSR. On the other hand, if one believes SSR, one should be able to produce a justification for it. And someone who urges SSR in the realism Page 3 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * debate would produce (what she hopes is) a justification because she would argue for SSR. A metaphysical doctrine of scientific realism and the epistemic one that is parasitic on it stand or fall epistemically together.6 It is common to propose what may seem to be semantic definitions of scientific realism, definitions using the terms ‘refer’ and ‘true’.7 For example, we might propose: “Most of the theoretical terms of currently well‐established scientific theories refer to mind‐independent entities and the theories' statements about those entities are approximately true”. This should be seen as simply a paraphrase of the metaphysical SSR, exploiting only the “disquotational” properties of ‘refer’ and ‘true’ captured in the schemas “ ‘F’ refers iff Fs exist” and “ ‘S’ is true iff S”. Such paraphrases are often convenient but they do not change the subject matter away from atoms, viruses, photons, and the like. They are not in any interesting sense semantic. In particular, they do not involve commitment to a causal theory of reference or a correspondence theory of truth, nor to any other theory of reference or truth. Indeed, they are compatible with a totally deflationary view of reference and truth: a deflationist can be a scientific realist (Horwich 1990).8 So there are epistemic and apparently semantic definitions of scientific realism which do not differ in any significant way from straightforwardly metaphysical definitions like SR and SSR. However, there are others that do differ significantly. Most important are the ones that really have a semantic (p.72) component. ‘Scientific realism’ is often now taken to refer to some combination of a metaphysical doctrine like SSR with a correspondence theory of truth.9 The combination is strange. Skepticism about unobservables, which is indubitably at the center of the realism debate, is simply not about the nature of truth. The issue of that nature is surely fascinating but is orthogonal to the realism issue.10 Of course there may be evidential connections between the two issues: there may be evidential connections between any issues (Duhem‐Quine). But no doctrine of truth is constitutive of metaphysical doctrines of scientific realism.11 In what follows I shall be concerned simply with the latter, using SR and SSR as my examples. We move on to consider the explanationist arguments for scientific realism, and the underdetermination argument and the pessimistic meta‐induction against realism.

(p.73) 3. Arguments for Scientific Realism 3.1 The Success Argument

The most famous argument for realism is the argument from the success of science. The argument has its origins in the work of Grover Maxwell (1962) and J. J. C. Smart (1963) but its most influential expression is by Putnam (1978: 18– 19), drawing on Richard Boyd. Scientific theories tend to be successful in that their observational predictions tend to come out true: if a theory says that S then Page 4 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * the world tends to be observationally as if S. Why are theories thus successful? The best explanation, the realist claims, is that the theories' theoretical terms typically refer—SR—and the theories are approximately true—SSR: the world is observationally as if S because, approximately, S.12 For example, why are all the observations we make just the sort we would make if there were atoms? Answer: because there are atoms. Sometimes the realist goes further: it would be “a miracle” that theories were so successful if they weren't approximately true. Realism does not just have the best explanation of success, it has the only good explanation.13 Larry Laudan (1981, 1984, 1996) has mounted a sustained attack on this argument. In the first prong of this attack, Laudan offers a list of past theories— phlogiston theory is a favorite example—that were successful but are now known not to be approximately true. The realist has a number of responses. First, the success of a theory can be challenged: although it was (p.74) thought to be successful, it was not really so (McAllister 1993). But unless the criterion of success is put so high that not even contemporary theories will qualify, some theories on Laudan's list will surely survive. Second, it can be argued that a theory was not, in the appropriate sense, well‐established and hence not the sort that the realist is committed to; or that entities it posited were not essential to its success (Kitcher 1993: 140–9). But surely some theories on the list will survive this test too. Third, the realist can insist that there are many other past theories, ones not on Laudan's list, for which the realist's explanation of success works fine (McMullin 1984). Still, what about the theories that survive on Laudan's list? The realist must offer some other explanation of their success. So even if the approximate truth of most theories is the best, perhaps only, explanation of their success, it cannot be so for all theories. But then the realist should not have needed to struggle with Laudan's list to discover this need to modify the success argument. After all, the sensible realist does not suppose that no well‐established scientific theory has ever been very wrong in its entities and its claims about them. And some theories that have been very wrong have surely been successful; indeed, scientists sometimes continue to use theories known to be false simply because they are so successful. So the realist must offer some explanation of this success that does not depend on the rightness of the theories. Here is a suggestion that is very much in the spirit of the original success argument. The success of a theory T that is very wrong is explained by the approximate truth of a replacement theory T’.14 It is because the unobservables posited by T’ exist and have approximately the properties attributed to them that T is successful. Indeed, we expect the very same theory that shows T to be wrong also to explain T's observational success.15 So the realist modifies the success argument: the best explanation of a theory's success is mostly that its unobservables exist and have approximately the properties specified by the Page 5 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * theory; otherwise, the best explanation is that the unobservables of a replacement theory exist and have approximately the properties specified by (p. 75) that theory. Furthermore, the realist may insist, the only way to explain the success of a theory is by appeal to its unobservables or those of another theory. The three earlier realist responses greatly reduce the challenge that Laudan's list poses for the original success argument. Still, some theories on the list will survive these responses. The modified argument offers an explanation of the success of those theories and is sufficient to support SR and SSR.16 Now, of course, this modification that accepts past mistakes raises the specter of the pessimistic meta‐induction.17 We shall consider that later (3.2). Meanwhile the modification does seem to save the success argument. But perhaps antirealists can explain success? There have been attempts. (1) Bas van Fraassen has offered a Darwinian explanation: “any scientific theory is born into a life of fierce competition, a jungle red in tooth and claw. Only the successful theories survive” (1980: 39). But this explanation is not relevant because it is not explaining the same thing as the realist's success argument. It is explaining why we humans hold successful theories. It is not explaining why those particular theories are successful (Lipton 1991: 170–2; Devitt 1991b: 116; Kitcher 1993: 155–7; Leplin 1997: 7–9). (2) Arthur Fine claims that antirealism can explain success as well as realism can by appealing to a theory's instrumental reliability (1986b; Fine is not committed to this antirealist explanation). Jarrett Leplin (1987, 1997) develops this proposal and labels it “surrealism”. The basic idea is that although the world has a “deep structure” this structure is “not experientially accessible”. “The explanation of the success of any theory . . . is that the actual structure of the world operates at the experiential level as if the theory represented it correctly” (1997: 26). Leplin goes on to argue, in my view convincingly, that the surrealist explanation is not a successful alternative to the realist one.18 In the second prong of his attack on realism, Laudan (1981: 45–6) has criticized the realist's success argument for its dependence on inference to the best (p. 76) explanation, or “abduction”. Fine (1986a: 113–22) has made a similar criticism.19 In presenting this criticism they charge the realist with “question‐ begging”. This charge is not apt. The realist argument could be question‐ begging only if it assumed abduction and the dispute with the antirealist was over abduction. But the primary dispute, at least, is not over abduction but over a doctrine like SSR. So, presented in this way, the criticism seems to miss its target. (One wonders if the cause of this mistake is that the many definitions of scientific realism have left the target unclear.)

Page 6 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * The criticism should be simply that the realist argument relies on abduction and this is a method of inference that an antirealist might reject. Van Fraassen (1980, 1989), for one, does reject it. Is the realist entitled to rely on abduction? Boyd (1984: 65–75) has argued that the antirealists are not in a position to deny entitlement because scientists regularly use abduction to draw conclusions about observables. Boyd's argument illustrates an important, and quite general realist strategy to defend unobservables against discrimination, to defend “unobservable rights”.20 The realist starts by reminding the antirealist that the debate is not over extreme skepticism: the antirealist claims to have knowledge of observables (sec. 2 above). The realist then examines the antirealist's justification for this knowledge. Using this justification she attempts to show, positively, that the epistemology it involves also justifies knowledge of unobservables. And, she attempts to show, negatively, that the case for skepticism about unobservables produced by the antirealist is no better than the case for skepticism about observables, a skepticism that all parties to the scientific realism dispute have rejected. So the antirealist's criticism of the success argument leaves him with the task of showing that he can save his beliefs about observables without using abduction. If he cannot manage this, the criticism fails. If he can—and van Fraassen (1989) has made an attempt—then the realist seems to face the task of justifying abduction. How concerned should the realist be about this? Perhaps not as much as many suppose. After all, the antirealist must rely on some methods of ampliative inference, even if not on abduction, to overcome extreme skepticism. How are those methods justified? The antirealist may well have little to say about this, relying on the fact that these methods are widely and (p.77) successfully used in science and ordinary life and on there being no apparent reason to abandon them. But, of course, that seems to be true of abduction as well. If further justification for a method is required, where could we find it?21 The naturalistically inclined will have trouble with any attempt at an a priori justification. And it is hard to see how an a priori approach could be effective either for or against abduction. What about an empirical justification within a naturalized epistemology? Any justification must of course use some methods of inference. If it uses the very method it seeks to justify, circularity threatens. Stathis Psillos argues that this circularity is not vicious (1999: 81–90). Be that as it may, Neurath's famous image of rebuilding a boat while staying afloat on it suggests another procedure for justifying a method: we hold fast to all other methods and use them to justify the method in contention. So, perhaps we can justify abduction using the methods of induction and deduction. Indeed, perhaps the very success of abduction in science and everyday life—its tendency to produce conclusions that are later observationally confirmed—provides the basis Page 7 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * for such an inductive justification.22 In any case it is not obvious that the justification of abduction will be more problematic than the justification of the methods of inference relied on by the antirealists. 3.2 The Success of Methodology Argument

Our scientific methodology is “instrumentally reliable” in that it leads to successful theories, theories that make true observational predictions. Everyone (p.78) agrees that our methodology does this. Why does it? What is the explanation? Boyd (1973, 1984, 1985) has posed this question and offered an answer that is both realist and naturalist: the methodology is based in a dialectical way on our theories and those theories are approximately true. He argues that antirealists of various sorts cannot explain this methodological success satisfactorily and so his realist explanation is the best. I think that he is probably right. Like the earlier success argument this argument relies on abduction, but it has a different explanandum. Where the earlier argument sought to explain the success of theories, this one seeks to explain the success of scientific methodology in producing successful theories. 3.3 The Basic Abductive Argument

The two abductive arguments for realism that we have considered are somewhat sophisticated. A more basic argument is strangely overlooked: by supposing that the unobservables of science exist, we can give good explanations of the behavior and characteristics of observed entities, behavior and characteristics which would otherwise remain inexplicable. This basic argument differs from the success argument in the following way. Where the success argument uses realism to explain the observational success of theories, the basic argument uses realism to explain observed phenomena. This is not to say that observational success is unimportant to the basic argument: the explanation of observed phenomena, like any explanation, is tested by its observational success. So according to the basic argument, realism is successful; according to the popular one, it explains success.23 In sum, there are some good arguments for scientific realism provided the realist is allowed abduction. Some critics reject abduction but this rejection seems dubious. Perhaps our knowledge of observables depends on abduction. In any case, abduction seems to be on an equal footing, at least, with other ampliative methods of inference.

(p.79) 4. Arguments against Scientific Realism

Page 8 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * 4.1 The Underdetermination Argument

There is an appealing and influential empiricist argument against scientific realism that starts from a doctrine of empirical equivalence.24 Let T be any theory committed to unobservables. Then, EE: T has empirically equivalent rivals. This is taken to imply the strong underdetermination thesis:

SU: T has rivals that are equally supported by all possible observational evidence for it. So, realist doctrines like SR and SSR are unjustified.25

Some preliminaries. First, what exactly is it for two theories to be “empirically equivalent”? The basic idea is that they have the same observational consequences. We shall later see the importance of looking very closely at this basic idea. Secondly, where EE talks simply of T having equivalent rivals, the premise of the argument is sometimes that T has indefinitely many rivals (e.g. Kukla 1998: 58) and sometimes, that it has at least one (e.g. Psillos 1999: 164). For convenience, I shall mostly treat EE as if it were only committed to one rival because its commitment to more does not seem to make a significant difference to the conclusions we should draw. Thirdly, SU should not be confused with various other underdetermination theses26 including the weak and obviously true one, mentioned in section 2, that leads to the challenge of extreme skepticism: WU: Any theory has rivals that entail the same actual given observational evidence. SU is stronger than WU in two respects. First, SU concerns an ampliative relation between theories and evidence and not merely a deductive one. Second, SU is concerned with T's relation to all possible evidence not merely (p.80) to the given evidence.27 If we are to avoid skepticism in the face of WU, we noted, some ampliative method of inference must be accepted. But if SU is true, we face a further challenge: ampliative methods do not support T over its rivals on the given evidence nor even on all possible evidence. So what T says about the unobservable world can make no evidential difference. Surely, then, commitment to what the theory says is a piece of misguided metaphysics. Even with extreme skepticism behind us, realism is threatened.28

Now, consider EE. A good reason for believing EE is that there is an empiricist algorithm for constructing an equivalent rival to T. Consider To, the theory that the observational consequences of T are true. To is obviously empirically equivalent to T. Still, it may not count as a rival because it is consistent with T.

Page 9 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * That is easily fixed: T* is the theory that To is true but T is not. T* is an empirically equivalent rival to T. So EE is established. It is tempting to respond that T* is produced by trickery and is not a genuine rival to T (Laudan and Leplin 1991; Hoefer and Rosenberg 1994). But this response seems question‐begging and unconvincing, as Andre Kukla argues (1998: 66–81). A better response is that, in counting theories generated by the empiricist algorithm as rivals, EE, as it stands, is too weak to sustain SU. For, with extreme skepticism behind us, we are justified in choosing T over T*. In considering this choice, the first half of T*, To, is key. In van Fraassen's terminology, To is the claim that T is “empirically adequate”. He has some famous remarks comparing this claim with the bolder claim that T is true: “the empirical adequacy of an empirical theory must always be more credible than its truth” (1985: 247); “it is not an epistemological principle that one may as well hang for a sheep as for a lamb” (p. 254). The extra boldness (p.81) of T comes, of course, from its realist commitment to certain truths about unobservables. Because van Fraassen thinks that T takes no further empirical risk than To, he claims that this extra boldness “is but empty strutting and posturing”, a “display of courage not under fire” (p. 255). We should prefer the weaker To. Now if van Fraassen were right about this, no evidence could justify a move from To to the bolder T. So it could not justify a preference for T over its rival T* (= To & not‐T). SU would be established. Here is a reason for thinking that van Fraassen is not right. If it were really the case that we were only ever justified in adopting the weakest theory compatible with the possible evidence for T, we would have to surrender to extreme skepticism. For, To is far from being the weakest such theory. For example, consider Te, the theory that T is “experientially adequate”. Where To claims that the observable world is as if T, Te claims only that the observable world appears to be as if T. Te is much weaker than To: it does not require that there be an observable world at all; perhaps an evil demon is at work. Those, like van Fraassen, who believe theories of the observable world are displaying courage not under fire all the time.29 This argument exemplifies the negative side of the realist strategy described earlier: arguing that the case for skepticism about unobservables produced by the antirealist is no better than the case for skepticism about observables. We can apply the positive side of the strategy too. Any methods of ampliative inference that support the move from Te to To and free us from extreme skepticism must justify the dismissal of the evil‐demon hypothesis and a whole lot of others. The methods must justify many singular hypotheses about unobserved objects and many general hypotheses that cover such objects (‘All ravens are black’ and the like). Whether or not these methods alone support the Page 10 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * further move to T, hence support scientific realism, they will surely justify the dismissal of T's rival T*, produced by the empiricist algorithm. And they will justify the dismissal of another empirically equivalent rival produced by Kukla's algorithm according to which the world changes when unobserved (1993). It would be nice to know, of course, what these methods are. But it is a strategic error for the scientific realist to attempt to say what they are in responding to the antirealist. For the antirealist believes in observables and whatever ampliative inferences support that belief will justify the dismissal of the likes of T*. The antirealist might, of course, simply insist that inferences that work for observables do not work for unobservables. Certainly there is no logical (p.82) inconsistency in this insistence.30 Nevertheless, the insistence is arbitrary and unprincipled. The realist need say no more.31 We conclude that EE as it stands cannot sustain SU: T is indeed justified over empirically equivalent rivals like T*. If the underdetermination argument is to work, it needs to start from a stronger equivalence thesis, one that does not count any theory as a rival to T that can be dismissed by whatever ampliative inferences enable us to avoid extreme skepticism. Let us say that the rivals that can be thus dismissed are not “genuine”. T* and the output of Kukla's algorithm are surely not genuine. Precisely how far we can go in thus dismissing rivals remains to be seen, of course, pending an account of how to avoid extreme skepticism. And, given the realist strategy, the account that matters is the one given by the antirealist. With EE now restricted to genuine rivals, the next step in assessing the underdetermination argument is a careful consideration of how to interpret EE's talk of empirical equivalence. The basic idea is that empirically equivalent theories have the same observational consequences. What does this amount to? A natural first stab at an answer is that the theories entail the same observations. This yields the following version of EE: EE1: T has genuine rivals that entail the same possible observational evidence. Whether or not EE1 is true, it is easy to see that it is inadequate to support SU. This inadequacy arises from the fact that T is likely to entail few observations on its own and yet the conjunction of T with auxiliary hypotheses, theories of instruments, background assumptions, and so on—briefly, its conjunction with “auxiliaries”—is likely to entail many observations. T does not face the tribunal of experience alone (Duhem–Quine). By failing to take account of these joint consequences, EE1 leaves many ways in which evidence could favor T over its rivals, contrary to SU. To sustain SU and challenge realism, we need another interpretation of EE. Page 11 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * Consider Laudan and Leplin's influential critique of the underdetermination argument (1991). They propose the thesis, “The Instability of Auxiliary Assumptions”, according to which “auxiliary information providing premises for the derivation of observational consequences from theory is unstable in two respects: it is defeasible and it is augmentable” (p. 57).32 As the accepted (p. 83) auxiliaries that can be conjoined with T change, so do its consequences. So, any determination of T's empirical consequence class “must be relativized to a particular state of science”, the state that supplies the auxiliary hypotheses. Thus “any finding of empirical equivalence is both contextual and defeasible” (p. 58). To determine the consequences of T we need more than logic, we need to know which auxiliaries are acceptable, an “inescapably epistemic” matter (p. 59). To avoid the consequences of this argument, Kukla (1993) proposed an answer to our interpretative question along the following lines: for two theories to be empirically equivalent at time t is for them to entail the same observations when conjoined with At, the auxiliaries that are accepted at t. This yields: EE2: T has genuine rivals which are such that when T and any of the rivals are conjoined with At they entail the same possible observational evidence. Set aside for a moment whether or not EE2 is any threat at all to realism. It is clearly too weak to sustain the threat posed by SU. Let T’ be an empirically equivalent rival to T according to this interpretation. So T&At and T’&At entail the same observations. This sort of equivalence is relative to At, to the auxiliaries accepted at a certain time. It amounts to the claim that T and T’ cannot be discriminated observationally if conjoined only with those auxiliaries. But this does not show that T and T’ could not be distinguished when conjoined with any acceptable auxiliaries at any time. And that is what is needed, at least, to sustain the claim that T and T’ cannot be discriminated by any possible evidence, as SU requires. SU demands a much stronger answer to the interpretative question: for two theories to be empirically equivalent is for them to entail the same observations when conjoined with any (possible) acceptable auxiliaries.33 This yields: EE3: T has genuine rivals which are such that when T and any of the rivals are conjoined with any possible acceptable auxiliaries they entail the same possible observational evidence. If T and T’ were thus related they would be empirically equivalent not just relative to certain auxiliaries but tout court, absolutely equivalent. Only then (p.84) would they be observationally indiscriminable. So if EE is to support SU, it must be interpreted as EE3.34

Page 12 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * The main point of Laudan and Leplin's critique can be put simply: we have no reason to believe EE3.35 If T and T’ cannot be discriminated observationally relative to, say, currently accepted auxiliaries, they may well be so relative to some future accepted auxiliaries. Some currently accepted auxiliaries may cease to be accepted and some new auxiliaries are likely to become accepted. This point becomes particularly persuasive, in my view (1991b: 119), when we note our capacity to invent new instruments and experiments to test theories. With a new instrument and experiment come new auxiliaries, including a theory of the instrument and assumptions about the experimental situation. Given that we can thus create evidence, the set of observational consequences of any theory seems totally open. Of course, there is no guarantee of successful discrimination by these means: a theory may really face a genuine empirically equivalent rival. Still, we are unlikely to have sufficient reason for believing this of any particular theory.36 More importantly, we have no reason at all for believing it of all theories, as EE3 requires. We will seldom, if ever, have a basis for concluding that two genuine rivals are empirically equivalent in the absolute sense required by EE3.37 This argument against EE3 does not depend on any assumption about the breadth of T. So EE3 cannot be saved by taking it to apply to “total sciences” (Boyd 1984: 50). Should such a broad conjunction of theories seem to face an equivalent rival at a certain time, we are unlikely to have sufficient reason for believing that experimental developments will not enable us to discriminate the conjunction from its rival by supplying new auxiliaries. There is no known limit to our capacity to generate acceptable auxiliaries. (p.85) I have argued that we have no reason to believe EE3. But suppose, nonetheless, that EE3 were true. Would this establish SU and undermine scientific realism? It might well do so.38 If EE3 were true, realists would have to appeal to “nonempirical virtues” to choose between empirically equivalent theories. Empirical virtue is a matter of entailing (in conjunction with accepted auxiliaries) observational truths and not entailing observational falsehoods. The nonempirical virtues are explanatory power, simplicity, and the like. For the reason indicated earlier in discussing abduction (sec. 3), I think that the realist is entitled to appeal to explanatory virtues, at least. But if it really were the case that all theories faced genuine rivals equally compatible with all possible evidence, the appeal to these virtues would seem epistemologically dubious.39 For, in those circumstances, there could be no way to judge the empirical success of these virtues, no way to show, for example, that theories that provide the best explanation tend to be observationally confirmed. So the defense of realism might well depend on there being no good reason for believing EE3. What about EE2? We have already seen that EE2 will not sustain SU. But perhaps it is otherwise threatening to realism. So, first, we need to consider whether it is true; then, whether, if it were, it would undermine realism. Page 13 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * There are surely some theories that face a genuine rival that is empirically equivalent relative to the accepted auxiliaries at a certain time. But do all theories face such rivals at that time, let alone at all times? EE2 guarantees that all theories do at all times. But the ampliative methods, whatever they may be, that support our knowledge of the observable world and avoid extreme skepticism will count many rivals as not genuine, so many as to make this guarantee seem baseless. How could we know a priori that T must always face such a genuine rival? Suppose, nonetheless, that EE2 were true. So, if T and its rivals are restricted to the accepted auxiliaries at a certain time, T could not be justified over some rivals on the basis only of the observations that the theories and auxiliaries entail and the ampliative methods that save us from extreme skepticism. So, without recourse to some further ampliative methods, T would be underdetermined by the evidence that the restriction allows into play. Of course, once new (p.86) acceptable auxiliaries were discovered and the restriction changed, the further methods might well not be needed to justify T over those old rivals. So this underdetermination would not be as serious as SU, but it would be serious enough: at any time, we would not know what to be realist about. But then perhaps the realist would be entitled to the further ampliative methods that would remove this underdetermination. For the reasons already indicated, and given that the case for EE3 has not been made, I think that the realist might be so entitled.40 In sum, we have no reason to believe EE2 or EE3 and so the underdetermination argument fails. However, if EE3 were true, it might well undermine scientific realism and if EE2 were true, it could. Once we have set aside extreme skepticism, then, contrary to received opinion, the nonempirical virtues are not central to defending realism from the underdetermination argument; the rejection of the equivalence thesis is. In drawing these conclusions I have mostly construed EE2 and EE3 as if they were committed only to T having at least one genuine empirically equivalent rival. Their actual commitment to more rivals does not significantly change the conclusions we should draw. 4.2 The Pessimistic Meta‐Induction

The most powerful argument against scientific realism, in my view, is what Putnam (1978) calls a “meta‐induction”. It does not rest on a prejudice against abduction or exaggerated concerns about underdetermination. It rests on plausible claims about the history of science. The basic version of the argument is aimed at an entity realism like SR: the unobservables posited by past theories do not exist; so, probably the unobservables posited by present theories do not exist. Another version, largely dependent on the basic one, is aimed at a “fact” realism like SSR: past scientific theories are not approximately true; so, probably present theories are not approximately true. (This is an example of the convenience of exploiting the disquotational property of ‘true’ to talk about the Page 14 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * world.) Both versions of the argument rest on a claim about past theories from the perspective of our present theory.41 And the pessimistic suggestion is (p.87) that, from a future perspective, we will have a similarly critical view of our present theories. Laudan (1981, 1984, 1996) has supported these claims about the past with a list of theoretical failures. Laudan's list is the one used to discuss the realist's success argument (2.1), but the purpose of the list is different here. The purpose before was to show that past theories were successful without being true, thus undermining the argument for realism that “realism explains success”. The list's purpose here is to show that past theories were not approximately true and their unobservables did not exist, thus establishing the premise of an argument against realism, against the view that present theories are approximately true and their unobservables exist. Scientific realism already concedes something to the meta‐induction in exhibiting some skepticism about the claims of science. It holds that science is more or less right but not totally so. It is committed only to well‐established theories not exciting speculations. It leaves room for a theoretical posit to be dismissed as inessential to the theory. According to the meta‐induction, reflection on the track record of science shows that this skepticism has not gone nearly far enough. The realist can respond to the meta‐induction by attacking the premise or the inference. Concerning the premise, the realist can, on the one hand, resist the bleak assessment of the theories on Laudan's list, claiming that while some of the unobservables posited by these theories do not exist, others do; or claiming that while there is a deal of falsehood in these theories, there is a deal of truth too (Worrall 1989;42 Kitcher 1993: 140–9; Psillos 1999: chs. 5–6). On the other hand, the realist can claim that the list is unrepresentative, that other past theories do seem to be approximately true and to posit entities that do exist (McMullin 1984). In the light of history, some skepticism about the claims of science is clearly appropriate. The argument is over how much, the mild skepticism of the realist, or the sweeping skepticism of the meta‐induction. Settling the argument requires close attention to the historical details. This is not, of course, something that I shall be attempting. However, I shall make some general remarks about the attempt. How can we tell whether Fs, posited by a past theory, exist? Given the disquotational schema, “ ‘F’ refers iff Fs exist”, many approach this question by considering another: how do we tell whether ‘F’ refers? This common approach would be harmless if it exploited only the disquotational property of ‘refer’ captured by the schema, a property acceptable to the deflationist. (p.88) For Page 15 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * then the reference question is just a paraphrase of the existence question. However, it is usual (and, in my view, right) to take ‘refer’ to pick out a substantive semantic relation between ‘F’ and the world, a relation that needs to be explained by a theory of reference. So it is natural to take the reference question to concern this substantive relation and to be answered by appealing to some theory of reference. But then the common approach is far from harmless. The first problem is that the theory of reference appealed to on this approach is usually a description theory. According to this theory, the reference of ‘F’ depends on the descriptions (other terms) that its containing theory associates with it: it refers to whatever those descriptions pick out. It is likely that, from our present scientific perspective, those descriptions do not pick anything out. So, the conclusion is drawn that ‘F’ does not refer and hence there are no Fs.43 Yet, the arguments of Saul Kripke (1980) and others have made it likely that reference for some terms, at least, is to be explained not by a description theory but by a theory that links a term to its referent in a more direct causal way.44 So it may well be that a description theory is the wrong theory for ‘F’. This points to the second, deeper, problem with the common approach: in attempting to answer the existence question by answering the reference question, the approach has its epistemic priorities all wrong. For, we know far less about reference, particularly about when to apply a description theory and when to apply a causal theory, than we know about what exists. In light of this, the rational procedure is to let our view of what exists guide our theories of reference rather than let our theories of reference determine what exists.45 So, we should not use a theory of reference to answer our existence question. How then should we answer it? Consider how, in general, we argue directly for the nonexistence of Fs. On the basis of the established view of Fs, we start, implicitly if not explicitly, with an assumption about the nature of Fs: something would not be an F unless it were G. Then we argue that nothing is G. So, there are no Fs. Very often this argument is persuasive and generally accepted. But someone might respond by denying the assumption about nature. “Fs do not have to be G, they are just mistakenly thought to (p.89) be G. So the argument proves nothing”. How do we settle this disagreement? It may be difficult. We can try saying more about the established view of Fs, but this may not do the trick. After all, the responder does not deny that Fs are thought to be G, just that being G is part of the nature of being an F. And the established view may not be clear on the nature issue. We may be left with nothing but a “clash of intuitions” over that issue. In such a situation, we should wonder whether there is a genuine issue to settle: there may be no determinate matter of fact about the nature issue. If there is not, then there is no determinate matter of fact about whether the absence of G things establishes the nonexistence of Fs.

Page 16 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * Consider two humdrum examples. Most people are antirealist about witches because they believe that nothing casts spells, rides on a broomstick through the sky, and so on. Some people may be antirealist about God because they are convinced by the Problem of Evil that nothing is both all powerful and all good. But these are grounds for antirealism only if casting of spells, riding on broomsticks, and so on, and being all powerful and all good, are essential to witches and God, respectively. There may be disagreement about that. And there is room for worry that disagreement may not be entirely over matters of fact.46 In light of this, we can expect that close attention to the historical details about past unobservables will reveal some ontologically determinate cases but very likely some indeterminate ones too. The determinate cases will surely include some of nonexistence; phlogiston is a good candidate. But it will surely also include some of existence; the atoms posited in the nineteenth century are good candidates.47 So, we should conclude that the premise of the meta‐induction is overstated, at least.48 But how much is it overstated? That depends on the “success ratio” of past theories, the ratio of the determinately existents to the determinately nonexistents + indeterminates. Where is this ratio likely to leave scientific realism? To answer this we need to consider the meta‐induction's inference. (p.90) I think (1984: 144–7; 1991b: 162–5) that there is a good reason for being dubious about the inference. Suppose that our past theories have indeed failed rather badly to get the unobservable world right. Why would that show that our present theories are failing similarly? It clearly would show this if we supposed that we are no better at finding out about unobservables now than we were in the past. But why suppose that? Just the opposite seems more plausible: we are now much better at finding out about unobservables. A naturalized epistemology would surely show that science has for two or three centuries been getting better and better at this. Scientific progress is, to a large degree, a matter of improving scientific methodologies often based on new technologies that provide new instruments for investigating the world. If this is so—and it seems fairly indubitable—then we should expect an examination of the historical details to show improvement over time in our success ratio for unobservables. If the details do show this, it will not matter to realism that the ratio for, say, two centuries ago was poor. What will matter is that we have been improving enough to now have the sort of confidence reflected by SR.49 And if we have been improving, but not fast enough for SR, the realist can fall back to a more moderate commitment to, say, a high proportion of the unobservables of currently well‐established theories. Improvements in scientific methodologies make it much harder to mount a case against realism than seems to have been appreciated. For the appeal to historical details has to show not only that we were nearly always wrong in our unobservable posits but that, despite methodological improvements, we have not Page 17 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * been getting significantly righter. It seems to me most unlikely that this case can be made.50

5. Conclusions Scientific realism is best seen as a straightforwardly metaphysical doctrine along the lines of SR or SSR. Various explanationist arguments for scientific realism (p.91) succeed provided that the realist is entitled to abduction. I have suggested that the realist is entitled. The underdetermination argument against realism fails because we have no good reason to believe an empirical equivalence thesis that would serve as its premise. The pessimistic meta‐ induction, with its attention to past theoretical failures, does pose a problem for realism. But the problem may be manageable. For, the antirealist must argue that the historical record shows not only that past failures are extensive but also that we have not improved our capacity to describe the unobservable world sufficiently to justify confidence that the accounts given by our current well‐ established theories are to a large extent right. This is a hard case to make.51 Postscript to “Scientific Realism”

1. Restricting Possible Evidence In assessing SU: T has rivals that are equally supported by all possible observational evidence for it and its consequences, “Scientific Realism” (sec. 4.1) takes what seems to be an appropriately liberal view of what counts as “possible observational evidence”, taking note of our capacity to create evidence by inventing new instruments and conducting new experiments. Quine has a more restrictive view reflecting, no doubt, his distaste for modality. He takes the possible evidence to be what would have been observed had there been an observer at each point of actual space‐time (1970b: 179). This is also van Fraassen's view of the phenomena he wishes to save: all actual observable things and events, past, present and future, whether or not anyone in fact observes them (1980: 12, 60, 64). On this restrictive view, acts of observation are the only nonactual aspects of possible evidence. On the liberal view, in contrast, the possible evidence includes many things that we do not do, but could have done, other than merely observing. If we had had more time, energy, and money perhaps we could (p.92) have invented the right instruments and conducted the right experiments to discriminate between T and T’. There may be many differences between them which we never detected because we passively observed points of actual space‐time where we could have actively intervened (Hacking 1983) to change what happened.52

Let us make explicit what SU becomes on the restrictive view of possible evidence: SU(r): Any theory positing unobservables has rivals that are equally supported by all the actual observable facts.

Page 18 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * Quine may believe SU(r)53 but, as Laudan points out (1996: 41–2), he offers no evidence for it. Yet it needs a powerful argument for much the same reason that SU did (sec. 4.1).54 Set that aside for a moment.

What would the consequences be for a scientific realism like SR or SSR if SU(r) were true. In a previous work I found SU(r), in effect, “too weak for the task of undermining Realism” (1991b: 121). Certainly, SU(r), unlike SU, would not show that we could not find evidence that would discriminate between a theory and its genuine rivals, just that we ran out of time before we did. So SU(r) would not show that commitment to a theory's unobservables was a piece of misguided metaphysics. But I overlooked that SU(r) would still be a problem for realism because it would have the consequence that we would never as a matter of fact know what to be realist about; we would never as a matter of fact be justified in preferring the unobservables of our chosen theory over those of its rivals. So the realist must resist SU(r) as well as SU. Where might we find an argument for SU(r)? Once again we must look to an empirical equivalence thesis. What thesis? We found no reason to believe EE3 but, in any case, its talk of “any possible acceptable auxiliary” would not recommend itself to the restrictive view of possible evidence. On that view, (p. 93) any empirical equivalence thesis that is to support SU(r) must surely restrict the relevant auxiliaries to actual ones. And it will not be sufficient to restrict them to actual ones that are accepted “at any point of space‐time” because some that are accepted at one point are later rejected; think, for example, of auxiliaries about the number of planets. It seems that we must restrict auxiliaries to the ones that are still standing “at the end of human inquiry”! This yields: EE4: T has genuine rivals which are such that when T and any of the rivals are conjoined with auxiliaries that are accepted at the end of human inquiry they entail the same actual observable facts. But why should we believe EE4? In discussing EE2 we found no reason to believe that all theories face genuine equivalent rivals relative to the auxiliaries at any time. So we have no reason to believe that all theories do relative to the auxiliaries at the end of human inquiry. We have no reason to believe even that some theories do. EE4 is baseless. Suppose that EE4 were true? Would it sustain SU(r)? If not, would it be otherwise damaging to scientific realism? It is not clear that the answer to either question is “Yes”. In thinking about these questions we should look, once again, for ampliative methods beyond those that save us from extreme skepticism. In sum, SU(r), with its restrictive view of the possible evidence, would threaten scientific realism, albeit not as badly as SU with its liberal view. But SU(r), like SU, needs a powerful argument that it does not have. In particular it can get no

Page 19 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * support from the equivalence thesis EE4 because the thesis is baseless. And it is not clear that EE4 would support SU(r) or antirealism even if it were true.

2. When are Abductions Good? The paper defends abduction as a method of inference and claims that certain abductions support scientific realism (sec. 3). To provide this support the abductions must, of course, be good inferences. But clearly not all attempts at abductive inference are good. Why suppose that these ones are? An answer cannot appeal to an algorithm for deciding which abductions are good for no such algorithm exists. Still, we do know some factors that should influence our decisions.55 (p.94) One thing that worries people about abductive arguments is that they can seem too easy. Whenever we have some previously unexplained phenomena, we can easily come up with some putative explanation if we do not operate under any constraints; thus, if we are prepared to tolerate absurdity, we can appeal to gremlins, acts of God, or Martian invaders. So, criterion (A) of a good abduction is that it involves a good explanation. As is well known, it is hard to say much about what makes an explanation good. But goodness requires at least two things. (A1) The explanation must not only not be absurd but it must be plausible given what else we know, it must be plausible relative to background knowledge. A corollary of this requirement is that if none of our candidate explanations meet this requirement, then we should not accept an abduction involving the best of this implausible bunch. Rather than such a rush to judgment we should follow a course of action the virtues of which are sadly underappreciated: we should suspend judgment and keep looking. (A2) Next, a good explanation must have an appropriate level of detail: if the explanation posits x as the cause of y, it must say enough about the mechanism by which x causes y to not leave this mysterious; a wave of the hand is not sufficient.56 If we do not have the details we need, we should, once again, suspend judgment. Criterion (B) of a good abduction is that the explanation it features must be better than any actual alternative or even any alternative that is likely, given what we already know. Attention to likely alternatives is important. Even if the featured explanation does not face any worked‐out alternatives, we may have some ideas for alternatives that, given our background knowledge, seem promising. Until those ideas have been explored sufficiently to be set aside, we should suspend judgment on the abduction. (This is not to be confused with the extreme‐skeptical view that we should not embrace an abduction until all possible alternative explanations have been set aside. We need wait only until no better alternatives seem likely, given what we already know.) In my view the abductions that feature in the explanationist arguments for realism meet these criteria.

Page 20 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * 3. Stanford's Unconceived Alternatives Kyle Stanford starts a recent book with the claim that “the most powerful challenge to scientific realism has yet to be formulated” (2006: 9). He goes (p. 95) on to formulate what he takes to be that challenge, offering a version of the pessimistic meta‐induction that includes elements from the underdetermination argument. I have already labeled the meta‐induction “the most powerful argument against scientific realism”. I did so because “it rests on plausible claims about the history of science”. Stanford brings out just how plausible such claims can be. I think his version of the meta‐induction is indeed the most powerful challenge. Stanford is concerned with underdetermination, but not with the sort alleged to arise from empirical equivalence. So he is not urging a version of the argument discussed in section 4.1 of my “Scientific Realism”. He is concerned rather with “recurrent, transient underdetermination”: “the possibility that there might be . . . empirically inequivalent but nonetheless well‐confirmed, serious alternative among the theories that we have not yet even imagined or entertained” (p. 17). The worry here is not with alternatives that I have labeled “not genuine”, ones that are Cartesian fantasies or concocted by some philosopher's algorithm. The worry is with alternatives that are as genuine as could be, “ordinary theoretical alternatives of the garden variety scientific sort that we have nonetheless simply not yet managed to conceive of in the first place” (p. 18); the worry is with “scientifically serious theoretical possibilities” (p. 31). This is “the problem of the unconceived alternative” (p. 18). Laurence Sklar has already raised this problem in his delightfully‐named article, “Do Unborn Hypotheses Have Rights?” (1981). Now, of course, there is a problem for realism only if theories really do face such “scientifically plausible competitors” (p. 18). Sklar assumes they do. Stanford sets out to argue that history supports this assumption: “we have repeatedly occupied a predicament of recurrent, transient underdetermination across a wide and heterogeneous variety of scientific fields and domains”. This is so because “subsequent inquiry would routinely (if not invariably) reveal further, radically distinct alternatives as well confirmed by the previous available evidence as those we were inclined to accept on the strength of that evidence” (p. 19). The power of Stanford's argument comes from his examination of the historical record in support of this thesis. Now history surely shows us that, typically, there comes a time when any theory T1 is replaced by a successor theory T2 that was previously unconceived. And T2 is likely to be at least as well supported by the evidence for T1 as T1 was. This should be uncontroversial. But Stanford goes much further: T1 and T2 are “radically distinct”. This is the crucial claim. For, suppose that the entities posited by T2 include those posited by T1, then clearly the change from T1 (p. 96) to T2 is no threat to an entity realism like SR. And suppose that, from the perspective of T2, T1 is approximately true, then clearly the change from T1 to Page 21 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * T2 is no threat to a “fact” realism like SSR. Stanford's threat to scientific realism comes from his view that this sort of “continuity” is not typical in the history of science. Despite his argument, Stanford is not generally skeptical about science: he is not suggesting that we should “never trust the deliverances of our scientific investigations” (p. 184; see also p. 37). And I have already allowed that some skepticism about the claims of science is obviously called for. The wise realist will commit only to the findings of well‐established theories not to those of exciting speculations and, even there, will allow for some errors. Stanford is clearly calling for a more sweeping skepticism. Still, the difference between him and the realist must remain a bit uncertain. The dispute can sometimes look like one over whether a glass is half empty or half full (cf. Godfrey‐Smith 2008). As noted in “Scientific Realism”, the realist can reply to the meta‐induction by attacking its premise or its inference. The premise is, of course, a gloomy view of the past. And the strength of Stanford's discussion is that he has done a good job of deepening the gloom. He presents a range of cases suggesting a lack of continuity between a theory and its successor. Three chapters (3 to 5) are devoted to one case in particular, biological theories of generation and inheritance. He devotes two chapters (6 and 7) to rebutting realist attempts to lessen the gloom. A lot of this is fairly convincing and I shall not challenge him on it.57 In any case I made no attempt to calculate (futilely?) the “success ratio” of past posits. Still, perhaps Stanford's discussion shows that I implied too much optimism about this ratio. So the issue comes down to the inference. The realist's reply here is to argue that gloom about the past does not require gloom about the present. Standard versions of this reply emphasize the “immaturity” of past sciences. These versions, as Stanford notes, “point out differences in the breadth, precision, novelty, or other important features of the predictive and explanatory accomplishments of past and present theories” to invalidate the inference in the induction (p. 43). Briefly, these versions base their rejection of the inference on the claim that present theories are more successful than past ones. But this claim has, as Stanford nicely notes, “the lingering whiff of ad‐hoc‐ery” (p. 10). The version of the realist's reply that I have always preferred bases rejection of the inference on the claim that our present methodology is better than past ones. This version explains why present theories are more successful and hence removes the whiff of ad‐hoc‐ery. (p.97) My version of the realist's reply, summarized in section 4.2 of the chapter, is to argue that we have very good reason to believe that we have been getting better and better at learning about the unobservable world; good reason to believe that, aided by technological developments, there has been, over recent centuries, a steady improvement in the methodology of science. That's Page 22 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * why our present theories are more successful. Indeed, serious sciences like physics take this improvement for granted, as methodological instruction in the classroom demonstrates. A naturalized epistemology surely supports this confidence. And, I have argued, there seems to be no basis for a pessimistic meta‐induction about this epistemology (1984: 146; 1991b: 163–4). So, we have serious scientific support for the view that theories are indeed getting more successful and hence, the realist argues, more true. Stanford responds to standard versions of the realist's reply, not to mine. His response is surprisingly brief and quite inadequate. He claims that the realist's reply simply does not apply to the problem of unconceived alternatives or to the new induction that supports it . . . because the latter arguments concern the theorists rather than the theories of past and present science. That is, they point out that past theories . . . were at one time the best or only theories we could come up with, notwithstanding the availability of equally well‐confirmed and scientifically serious alternatives. . . . that present theorists are no better able to exhaust the space of serious, well‐confirmed possible theoretical explanations of the phenomena than past theorists have turned out to be. (2006: 44) Stanford has simply missed the point of the realist's reply. He is surely right that present scientists are not likely to be better at exhausting that theoretical space. But this is just to endorse what I earlier called “uncontroversial”: every present theory is likely to have a successor that we have not yet conceived. An effective response to the realist needs to support what I earlier called “the crucial claim”: the claim that the present theory and its successor will be “radically distinct”. For that claim is the one that threatens realism. And that claim is what the realist's reply is in effect challenging. The realist is pointing out that even if in the past theories were often radically distinct from their successors they are not likely to be now. Continuity is the issue. The realist reply is that discontinuity in the past does not mean discontinuity now. Stanford has failed to address this point.

“Scientific Realism” did not claim that this realist reply should remove all worries occasioned by the meta‐induction, and I do not suggest this now. Still, I do think that this reply does a lot to alleviate the problem for doctrines like SR and SSR. And, as I pointed out, we can always fall back to a weaker (p.98) realist position. Stanford thinks that all such fallbacks leave realism with only a “Pyrrhic” victory (p. 142). I don't think that this is so. The basic realist view that we are getting to know more and more about the unobservable world remains intact even if there remains some doubt as to just how successful we have been at this so far.

Page 23 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * Notes:

* First published in The Oxford Handbook of Contemporary Philosophy, ed. Frank Jackson and Michael Smith (Devitt 2005c). Reprinted with kind permission from Oxford University Press. (1) [2009 addition] Extreme skepticism is discussed in Ch. 3 of the present volume. WU is the same as D2b in Ch. 3, sec. 2. (2) But what about quantum theory? The notorious Copenhagen interpretation responds to the mysterious picture of the world suggested by quantum theory by taking the quantum world to be observer‐dependent. This would offend against the independence dimension of realism. But, of course, this interpretation of the theory is not obligatory. Many interpretations have been proposed that do not involve observer dependence. I'm told that all of these are somewhat weird in one way or another. Some philosophers respond to the mysteries by taking quantum theory instrumentally and hence not as an accurate guide to reality; so, the existence dimension of realism is not embraced for the theory. But the enormous success of the theory makes this a difficult choice (see sec. 3 below on the significance of success). In brief, controversy rages (see e.g. the papers in Cushing and McMullin 1989; Cushing et al. 1996). This situation is both fascinating and worrying. What conclusions should we draw from it about scientific realism? In my view, we should draw none until the dust begins to settle. (3) Elsewhere I take a very dim view of constructivism (1991b, 1999a, and 2001a, which is Ch. 5 in the present volume). (4) [2009 addition] Christopher Gauker, in his commentary on this paper, seems to think that my talk of “unobservables” needs clarification (2006: 126). I don't think so. Certainly the talk is a bit vague but that is no problem at all for someone who is a realist about observables too; e.g. someone committed to my Ch. 2 doctrine Realism. And it is not obvious that the vagueness is a serious problem for those like van Fraassen whose realism is limited to observables. Their serious problem is the apparent epistemic insignificance of unobservability (1984: 130–4; 1991b: 143–7). (In his attempt to clarify the talk of unobservables, Gauker reformulates SR using ‘refers'. This is unnecessary but nonetheless harmless; see the discussion of “apparently semantic” definitions below. But he then goes on to argue that this reformulation makes scientific realism “a thesis about the nature of truth”, indeed a thesis that “makes truth mind‐ dependent” (p. 126). His argument has no force, in my view.) (5) The scare quotes around ‘facts' are to indicate that the use of the term can be regarded as a mere manner of speaking, not reflecting any commitment to the existence of what many regard as very dubious entities. Ian Hacking (1983) calls this sort of doctrine “theory realism”. I prefer to talk of “facts” rather than Page 24 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * theories to emphasize that the doctrine is about the world itself not our account of it. (6) Indeed, one can generalize: (p)((p is justified) is justified iff p is justified). (Talk of justification here should be construed broadly so that “externalist” accounts of knowledge are not ruled out.) Jarrett Leplin (1991: 26) has defined an epistemic doctrine, “minimal epistemic realism”, that is not parasitic on a metaphysical one. This doctrine does not claim that a belief in any of the claims of science is justified, just that such a belief could be justified. A realism that concedes this much to the skeptic is indeed minimal (although still too strong for an antirealist like Bas van Fraassen). (7) E.g. Hesse 1967: 407; Hooker 1974: 409; Papineau 1979: 126; Ellis 1979: 28; Boyd 1984: 41–2; Leplin 1984b; Fales 1988: 253–4; Jennings 1989: 240; Matheson 1989. (8) [2009 addition] Gauker claims that my argument rests on an “equivalence thesis” for ‘true’ or ‘refer’ and that the explanation of that thesis will be semantic. So, he concludes, scientific realism is a semantic thesis after all (2006: 129–30). This is a non sequitur (and if it were not it could be used to show that all theses were semantic). My argument needs the equivalence thesis to be true, that's all. The explanation of the thesis, important and interesting as that is (see Ch. 8, sec. 6, below), is beside the point. (9) See e.g. Putnam 1978: 18–20, 123–5; 1987: 15–16; Fine 1986a: 115–16, 136– 7; R. Miller 1987; Kitcher 1993: 127–33; J. R. Brown 1994. Arthur Fine dismisses the realism issue altogether because he takes it to involve an issue about truth. Despite this dismissal, Fine urges the mysterious “Natural Ontological Attitude” (1986a, b) which often seems to be a realist doctrine like SSR! However, some passages (1986b: 163–5) make it hard to take Fine as a realist. (10) This objection also counts against definitions of realism that include the idea that truth is the aim of science (e.g. van Fraassen 1980: 8) wherever this talk of truth is taken to commit realism to correspondence truth. Even if the talk does not have that commitment, and so is acceptable to a deflationist, such definitions have problems. On the one hand, if the idea that truth is the aim of science is added to a doctrine like SSR, the addition is uninteresting: if science is discovering the truth, nobody is going to propose that it is not aiming to, that truth is a happy accident. On the other hand, if the idea is not added to a doctrine like SSR, the definition will be too weak: realism will require that science aims for truth without any commitment to it ever having achieved that aim. Indeed, if science had never achieved that aim despite the efforts of the last few centuries, it would hardly be rational now to have the aim. In the distant past, of course, the situation was different (as Howard Sankey has emphasized to me). Then the realistically inclined philosopher should have had the aim Page 25 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * without the commitment. But now she should have the commitment with the result that the aim goes without saying. [2009 addition] For more against defining realism as the view that truth is the aim of science, see Ch. 2, part II, Aberration 8 in the present volume. (11) For more on this and other matters to do with defining realism, see my 1991b, particularly chs. 2–4; 1997a, part I; 1999a, part I. See also Ch. 2, part I and Postscript, in the present volume. Here are two excuses for the intrusion of semantics into the definition of scientific realism. (1) We seem to need semantics to capture the “nonfactualist” antirealism of classical positivistic instrumentalism (for a learned account of the history of this instrumentalism, see Psillos 1999: ch. 3). This instrumentalism is like the moral antirealism of “noncognitivism” in claiming that what appear to be descriptive and factual statements are not really so. So these “statements” are not really committed to what they appear to be committed to. I argue that these antirealisms are, nonetheless, at bottom metaphysical not semantic (1996b, which is ch. 7 in the present volume; 1997a: part II). Aside from that, positivistic instrumentalism is no longer a player in the dispute over scientific realism (although instrumentalism in general certainly is, for it simply involves doubting the theoretical claims of science without reinterpreting them). (2) Although a doctrine like SSR need not be combined with a correspondence theory of truth it very likely cannot be plausibly combined with an epistemic theory of truth. But still a nonepistemic theory is not constitutive of SSR (1991b: 44–6; Ch. 2, Aberration 6, in the present volume). (12) Note that although this argument is usually stated using ‘refer’ and ‘true’, this is not essential. And such usage should be seen as exploiting only the disquotational properties of the terms with no commitment to a robust correspondence relation between language and the world. The realist argument should be that success is explained by the properties of unobservables, not by the properties of truth and reference. (“Truth, like Mae West's goodness, has nothing to do with it”; Levin 1984: 124.) So the argument could be urged by a deflationist (Devitt 1984: 106–10; 1991b: 113–17). (13) [2009 addition] Gauker objects that SR cannot explain the success of our theories. (His objection is actually aimed at one of my earlier versions of SR; 1991b: 21. But if the argument were good it would apply as much to SR. So I shall adapt it so that it is aimed at SR to avoid unnecessary complications.) He rightly points out that “part of the explanation of why the electron theory enables us to predict what will happen in a cathode ray tube might be that electrons exist”. Yet SR is not committed to the existence of electrons. It is simply committed to the existence of most of the essential unobservables of well‐ established current scientific theories, and many of those unobservables “are quite irrelevant to the explanation of our success with cathode ray tubes” (2006: 128). So how could SR explain that success? Gauker has misunderstood the Page 26 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * success argument. The argument does not claim that SR explains the success of electron theory in particular. It claims that SR explains the success of theories in general: theories tend to be successful because the entities they posit typically exist. Gauker has the wrong explanandum. (14) Laudan is scornful of the realists' appeal to the explanatory role of approximate truth. He complains that the notion is undefined. He acknowledges that many scientifically useful notions are undefined, but thinks that approximate truth is so specially unclear that the realists' appeal to it is “so much mumbo‐jumbo” (1981: 32). This is excessive. First, approximately true theories can have fully true parts that do the explaining. Second, the talk of approximate truth is simply a convenience (n. 12 above). The claim that the approximate truth of ‘a is F’ explains an observation amounts to the claim that a's being approximately F explains it. Science and life are replete with such explanations; e.g. a's being approximately spherical explains why it rolls. (15) This expectation should not be construed as a requirement on a replacement theory: that no theory should replace T until it has explained T's success. Laudan rightly objects to this requirement (1981: 44–5). (16) Insofar as we have reason to believe that the success of a current theory is to be explained by some as yet unknown future theory, that success provides no reason to believe in the current theory's entities or approximate truth. So the modified success argument's support for SR and SSR does depend on the three earlier responses showing that it is mostly the case that the success of a past theory can be explained by the reference of its terms and its approximate truth. (I am indebted to Mathias Frisch for drawing my attention to this.) (17) Peter Lewis (2002) summarizes Laudan's discussion as a meta‐induction (which differs from the meta‐induction we shall be considering): Many false past theories were successful. So the success of a theory is not a reliable test for its truth. Lewis argues persuasively that Laudan's meta‐induction exemplifies “the false positives paradox”. To establish his conclusion, Laudan needs to find periods of science where most theories were successful and yet most were false.

(18) For more on this, see Kukla 1998: 20–4; Psillos 1999: 90–3. (19) Realists tend to be fond of abduction; e.g. Glymour 1984; Boyd 1984; McMullin 1984; Musgrave 1985; Leplin 1991. (20) For examples of this strategy, see Churchland 1985; Gutting 1985; Musgrave 1985; Clendinnen 1989; Devitt 1991b (pp. 147–53); Psillos 1999 (pp. 186–91). Van Fraassen 1985 responds to the first three of these.

Page 27 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * (21) One problem in finding it is that we cannot give a precise specification of any of these methods: as Georges Rey says, “no one yet has an adequate theory of our knowledge of much of anything” (1998: 29). (22) The suggestion is that experience, according to the empiricist “the sole legitimate source of information about the world” (van Fraassen 1985: 286), supports abduction. For arguments in favor of abduction see Boyd 1984; McMullin 1984; Lipton 1991; Devitt 1991b: 111–13; Leplin 1991: 116–20. Analogous problems arise, of course, over the justification of deduction; see Field 1996 and 1998; Rey 1998; Devitt 1998, which is Ch. 12 in the present volume. [2009 addition] Ch. 13, sec. 2.4, expresses doubts about the Neurathian suggestion, and has a detailed discussion of the rule‐circularity problem for justifying inference rules. Van Fraassen (2000: 261–71) seems to misunderstand the relation of naturalized epistemology to science. It goes without saying that epistemology implies the methods of science. But van Fraassen seems to take the naturalist view to be that basic science, or special sciences like biology, medicine, and psychology, imply the methods of science, a view that he rejects. This view misrepresents naturalism. Naturalism holds that epistemology is itself a special science. As such it is no more simply implied by another science than is any other special science: it has the same sort of relative autonomy, and yet dependence on basic science, as with other special sciences. Naturalized epistemology, like any special science, applies the usual methods of science, whatever they may be, mostly taking established science for granted, to investigate its special realm. In the case of epistemology that realm is those very methods of science. The aim is to discover empirically how we humans learn, and should learn, about the world (1984: 63–7; 1991b: 75–9). We have no reason to suppose that the methods that have yielded knowledge elsewhere cannot yield knowledge in epistemology. (23) Devitt 1984: 106–10; 1991b: 113–17. Hacking's arguments (1983) for the reality of entities manipulated in experiments and perceived under a microscope are persuasive examples of the basic argument (although he, strangely, does not regard them as abductions). Hacking's point about manipulation is clearly related to Alan Musgrave's (1988) insistence that novel predictions give us the best reason for believing a theory. Ernan McMullin (1991), responding to Fine (1991), provides some nice examples of the basic argument in geology, biology, and astrophysics. An advantage of the basic argument is that it makes clear that, contrary to Fine's frequent claims, the use of abduction to justify realism is not at some “philosophical” level above science: “the argument is properly carried on at one level only, the level of the scientist” (McMullin 1991: 104). (24) My discussion of this argument draws on a more detailed one in Devitt 2002a.

Page 28 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * (25) The argument has no one clear source. But see Duhem 1906; Quine 1960, 1961 (“Two Dogmas”), and 1975a; van Fraassen 1980; Putnam 1983 (“Equivalence”). (26) My presentation reflects the influence of Laudan's excellent discussion of this variety (1996: ch. 2). (27) Since SU is concerned with all possible evidence the premise about empirical equivalence that is supposed to support SU in the underdetermination argument must be also. Psillos's version of the argument fails on this score. It starts: “for any theory T and any body of observational evidence E, there is another theory T’ such that T and T’ are empirically equivalent with respect to E” (1999: 164). The quantifiers need to be reordered if this is to support SU: for any theory T, there is another theory T’ such that for any (possible) body of observational evidence E, T and T’ are empirically equivalent with respect to E. (28) [2009 addition] Since the possible evidence must include the actual given evidence, SU would entail A2: Any theory has rivals that are equally supported by the actual given observational evidence for that theory were it not for SU's limitation to theories positing unobservables; so our rejection of A2 in Ch. 3 of the present volume would suffice to reject SU. It is only because of the limitation to theories positing unobservables that, even with extreme skepticism behind us, we must still be concerned about SU. We need a powerful argument to show why the limitation makes this difference. The argument from empirical equivalence is supposed to fulfill this need. That argument is another illustration of the pattern noted in Ch. 3, sec. 2: deductive underdetermination → ampliative underdetermination → antirealism.

(29) I develop this argument more thoroughly in my 1991b: 150–3. (30) Kukla emphasizes this (1998: 25–6, 84). (31) However, I think that an examination of the epistemic significance of observation helps to bring out the arbitrariness (1984: 130–4; 1991b: 143–7). (32) See also Ellis 1985 and Devitt 1991b: 117–21. (33) This demand arises out of a liberal and, it seems to me, intuitive view of what counts as “possible evidence”. Quine and van Fraassen have a more restricted view which I discuss in my 2002a: sec. 13. See also Postscript, sec. 1, to the present chapter. (34) Tim Williamson pointed out to me a problem with EE3 as it stands. Suppose that T1 and T2 are two allegedly equivalent rivals and that A1 is an acceptable auxiliary relative to T1 but not T2 and A2 is an acceptable auxiliary relative to T2 but not T1. Thus A1 might be a theory of a testing instrument from the Page 29 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * perspective of T1 and A2 a theory of that instrument from the perspective of T2. So the acceptability of the auxiliaries is not independent of the theories being tested. Now suppose that T1&A1 and T2&A1 have different observational consequences. That alone should not show that T1 and T2 are not empirically equivalent. For, T1&A1 and T2&A2 might have the same observational consequences. Clearly, what need to be assessed for empirical equivalence are theories together with their dependent auxiliaries. And EE3 should be taken as referring to any possible independently acceptable auxiliaries. (35) Note that this is not the claim that EE3 is “demonstratively false”; cf. Kukla 1998: 58. (36) For some theories where we may have sufficient reason, and for some past ones where we wrongly thought we had, see Psillos (1999: 166–8) and the works he cites. (37) [2009 addition] Behind this argument lies the following Realist picture. T and T’ describe different causal structures alleged to underlie the phenomena. We can manipulate the actual underlying structure to get observable effects. We have no reason to believe that we could not organize these manipulations so that, if the structure were as T says, the effects would be of one sort, whereas if the structure were as T’ says, the effects would be of a different sort. (38) Laudan and Leplin (1991: 63–8) think not, arguing that T can be indirectly supported over its rival by evidence that confirms another theory that entails T but not its rival; and that some consequences of T and its rival might support only T. But, as Kukla points out (1998: 84–90), this argument begs the question: if EE3 really were true, this evidential support would seem to disappear. (39) I emphasize that since it has not been established that all theories do face such rivals, it might well be appropriate to appeal to explanatory virtues, or indeed to the evidential support mentioned by Laudan and Leplin (1991: 63–8), to prefer some theory that does face such a rival. (40) In a reply to Kukla 1993, Leplin and Laudan (1993: 10), in effect, doubt EE2 but in any case emphasize that EE3 is what matters to the underdetermination argument. Kukla disagrees, claiming, in effect, that EE2, when applied to total sciences, “brings in its train all the epistemological problems that were ever ascribed to the doctrine of EE” (1998: 64). According to my discussion, EE2 would bring some epistemological problems if it were true, but they are not as extreme as those that would be brought by EE3 if it were true. (41) So there is a “tension” in the argument: it seems to rest on a realist view of present science and yet concludes that this realist view is mistaken; see Leplin

Page 30 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * 1991: 141–5. I suppose that we should see the meta‐induction as some sort of reductio. (42) Worrall takes the truth to be not about the nature of entities but about structures that contain the entities. For a critical discussion of this “structural realism”, see Psillos 1999: ch. 7. (43) See Kuhn (1962) for an argument along these lines. Stephen Stich (1983) and others have argued similarly for various forms of eliminativism about the mind. Stich has since recanted (1996: 3–90). (44) For a summary and development of these and other moves in the theory of reference see Devitt and Sterelny 1999. (45) My 1984/1991b argues for this priority. See also, Ch. 2, Aberration 7, and Ch. 15, sec. 4, in the present volume. There is, of course, a truth underlying the mistaken approach: to determine whether the posits of a theory exist we have to know what those posits are and for that we have to understand the language of the theory (pp. 50–3). But understanding a language is a practical skill that does not require theoretical knowledge about the language, else we would understand very little (pp. 270–5). (46) [2009 addition] See Ch. 15 in the present volume for further discussion. (47) Also molecules and microbes; see R. Miller 1987. (48) [2009 addition] So the pessimistic meta‐induction needs to be modified. It is important to see that, to maximize the chance of the meta‐induction being effective against SR, it should take the following form: most of the unobservables posited by theories at all, or most, past times do not exist; so, probably, most of the unobservables posited by current theories do not exist. Marc Lange (2002) has shown that this form is much better than the following common one: most unobservables posited in the past do not exist; so probably most unobservables posited now do not exist. The problem with the latter form is that it commits “the turnover fallacy”. Because false theories turn over much more often than true ones, its premise might be true even though, at any time, most of the unobservables posited at that time exist. (I was prescient enough, or lucky enough, to take the former form as the one to be criticized in my 1984: 143–5 and 1991b: 161–2.) (49) So this realist response does not take the failures of “immature” science to be irrelevant to the defense of realism, thus threatening the defense with “vacuity” (Laudan 1981: 34). Rather, it takes the relevance of a science's failures (and successes) to that defense to increase with the degree of that science's maturity, a degree assessed by an empirical epistemology.

Page 31 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * (50) [2009 addition] Jacob Busch (2006b) criticizes my discussion of the pessimistic meta‐induction but, as I point out in response (2006b), he misplaces the onus. The meta‐induction is offered as an argument against scientific realism. So there is an onus on those who propose it to show that it is a good argument. This requires that they establish its premise and its inference. The onus on realists is simply to show that the premise and/or the inference has not been established. Realists have already provided abductive arguments for realism and their further onus is simply to undermine arguments against it. I have accepted that onus. I don't think that proponents of the meta‐induction have fully accepted theirs. (51) Versions of this chapter were delivered in many places, starting with a conference, “Logic and Metaphysics”, held in Genoa in Sept. 2001. I am indebted to these audiences for comments. I am also indebted to Radu Dudau for helpful advice on the literature, to Peter Godfrey‐Smith for a helpful prior exchange on the topic, to the members of my graduate class on scientific realism in Fall 2001, and to the following for comments on a draft: Jeff Bub, Radu Dudau, Frank Jackson, Mikael Karlsson, Andre Kukla, Jarrett Leplin, David Papineau, Stathis Psillos, and Howard Sankey. (52) This section follows closely Devitt 2002a: sec. 13. (53) Quine starts, in effect, by saying that SU(r) is conceivable (1960: 22), which it surely is, and then moves on to the claim, mentioned in Ch. 3 n. 3 of the present volume, that the very different doctrine, “any observational belief has rivals that are equally supported by the actual given sensory evidence for that belief”, is likely even when the given evidence is “man's surface irritations even unto eternity” (1960: 23). (54) Since the actual observable facts must include the observed facts, SU(r), like SU, would entail A2: Any theory has rivals that are equally supported by the actual given observational evidence for that theory were it not for its limitation to theories positing unobservables; see n. 28 above. So our rejection of A2 in Ch. 3 of the present volume would suffice to reject SU(r). We need a powerful argument to show why, with extreme skepticism behind us, this limitation makes a difference.

(55) This section draws heavily on Devitt 2006d: 198–201. I have discussed abduction before: 1991b: 111–13. (56) Berkeley rightly criticizes an abduction for Locke's “representative realism” on this score. The abduction is that the existence of a material world causing and resembling our ideas provides the best explanation of those ideas. Berkeley

Page 32 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Scientific Realism * points out that we do not have any idea of the mechanism by which a material body can cause a resembling spiritual idea (1710: sec. 19). (57) Stanford is particularly harsh on realist attempts to lessen the gloom by appealing to theories of reference (2006: 147–55). Clearly this is not an attempt I would make.

Access brought to you by:

Page 33 of 33

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics *

Putting Metaphysics First: Essays on Metaphysics and Epistemology Michael Devitt

Print publication date: 2009 Print ISBN-13: 9780199280803 Published to Oxford Scholarship Online: February 2010 DOI: 10.1093/acprof:oso/9780199280803.001.0001

Incommensurability and the Priority of Metaphysics * Michael Devitt (Contributor Webpage)

I share the common view that Thomas Kuhn and Paul Feyerabend enormously improved our understanding of the history and epistemology of science. Their suggestions about the semantics of science are interesting but, in my view, dubious at best. Their antirealist views on the metaphysics of science are a disaster. I shall be concerned with that disaster and with its connection to the incommensurability issue.

1. Incommensurability I shall argue against an incommensurability thesis. My argument is partly a response to the views of Paul Hoyningen‐Huene and his co‐authors, particularly to his Reconstructing Scientific Revolutions (1993), a sympathetic and wonderfully scholarly account of Kuhn's views. I shall draw on my Realism and Truth (1984/1991b/1997a). The thesis I shall reject is: Incommensurability: Terms in rival comprehensive theories in a domain differ sufficiently in meaning, especially in reference, to make the theories incomparable. What sorts of theory comparison is Incommensurability against? A straightforward sort would arise if the two theories share referents. As a result, some parts of one theory would be logically inconsistent with some parts of the other, and some parts would entail some parts of the other. But there could be more complicated sorts of comparison where referents were not shared; for example, Michael Martin's case where terms in the two theories have (p.100) overlapping referents (1971); or Hartry Field's case where the theories share partial referents (1973). The intuitive idea of theory comparison is that what one Page 1 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics * theory says about x or about Fs is in agreement or disagreement with what the other says about x or about Fs. Sometimes the disagreement might be over the very existence of x or Fs. Agreement and disagreement at the observational level is, of course, particularly important to theory comparison. Incommensurability denies that any of these sorts of comparison of rival theories is possible. Clearly Incommensurability, italicized, has been stimulated by the writings of Thomas Kuhn (e.g. 1970) and Paul Feyerabend (e.g. 1981a, b) on incommensurability. But do Kuhn and Feyerabend actually embrace Incommensurability? Their writings on the matter are so notoriously various and vague as to leave plenty of room for disagreement over interpretation. Hoyningen‐Huene argues that the incommensurability that Kuhn has in mind does not imply incomparability (1993: 218–22). I will address this interpretative issue briefly in a moment, but it is not really my concern.1 Whatever the truth of it, the rejection of Incommensurability is worthwhile for two reasons. First, the writings of Kuhn and Feyerabend have been commonly taken to imply Incommensurability: Hoyningen‐Huene gives a long list of philosophers “among many others” who have construed Kuhn and Feyerabend in this way (p. 218 n.). Second, rejecting Incommensurability is sufficient to remove the worries occasioned by those writings for a realist view of science. Indeed, if the claims in (p.101) those writings about meaning change and translation failure do not threaten theory comparison then, whatever their purely semantic interest, they do not pose any special problems for the epistemology and metaphysics of science.

2. Realism and Constructivism My position on the realism issue is captured by the following doctrine: Realism: Tokens of most commonsense, and scientific, physical types objectively exist independently of the mental. At the observable level, the tokens in question are of stones, trees, cats, and the like; at the unobservable level, they are of electrons, muons, curved space‐time, and the like.2

Realism stands opposed to a variety of doctrines. The one that concerns us most is usually called “Constructivism”. It starts from two Kantian ideas. The first of these is that the knowable world of “appearances”—stones, trees, cats, and the like—is partly constituted by the cognitive activities of the mind. Kant called this world “the phenomenal world”. According to Kant himself, the mind constitutes the phenomenal world by imposing a priori concepts. In recent times the restriction to a priori concepts has been removed: the mind may impose any concept, or a theory, or a language. The second Kantian idea distinguishes objects as we know them from objects as they are independent of our knowledge. The latter objects make up “the noumenal world” of “things‐in‐ themselves”. Only the noumenal world, forever inaccessible and beyond our ken, has the objectivity and independence required by Realism. The phenomenal Page 2 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics * world of familiar objects does not, as it is partly our construction.3 So, where the Realist thinks that stones, trees and cats are both (p.102) knowable and independent, Kant thinks that they are knowable but dependent, being partly constituted by us and partly by an unknowable independent world. Constructivism adds a third idea to these two Kantian ones: relativism. Kant was no relativist: the concepts imposed to constitute the known world were common to all humankind. Constructivists drop this universality. Different languages, theories, and world views are imposed to create different known worlds. In sum: Constructivism: The only independent reality is beyond the reach of our knowledge and language. A known world is partly constructed by the imposition of concepts. These concepts differ from (linguistic, social, scientific, etc.) group to group and hence the worlds of groups differ. Each such world exists only relative to an imposition of concepts. So Constructivists believe in what Hoyningen‐Huene aptly calls “the plurality‐of‐ phenomenal‐worlds thesis” (1993: 36).

Examples of Constructivist worlds include the stars made by a Goodman “version” (1978); the constructed worlds of Putnam's “internal realism” (1981); the worlds built by a Whorfian language (1956); the many worlds created by the “discourses” of structuralists and post‐structuralists. Most important for our purposes, according to Kuhn and Feyerabend, the ontologies of scientific theories are constructivist worlds. It is common to interpret Kuhn and Feyerabend as subscribing to Constructivism, but the interpretation has not been without its problems. It is comforting, therefore, to find Hoyningen‐Huene (1993) giving an authoritative argument for this interpretation of Kuhn.4

3. The Relation between the Realism and Incommensurability Issues No position on Realism entails a position on Incommensurability because Incommensurability is a semantic doctrine whereas Realism is not. Still arguments can (p.103) be mounted from a position on Realism to a position on Incommensurability with the help of semantic assumptions. Consider this argument first: Constructivism → Incommensurability, Adopt the common assumption that meaning is to be explained at least partly in terms of reference. Let T1 and T2 be examples of rival comprehensive theories. What might the terms of T1 and T2 refer to? According to Constructivism, they cannot refer to the same entities, because with the move from T1 to T2 the world changes. Indeed, although some entities exist‐relative‐to‐T1 and others exist‐relative‐to‐T2, no sense can be made of any of these entities existing “absolutely”. So the potential referents, or even potential partial referents, of T1 terms are different from the potential referents Page 3 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics * of T2 terms. So there is no way that T1 and T2 can be compared in the way that concerns Incommensurability: they cannot agree or disagree about x or Fs because they are not talking about the same x or Fs.

In light of this let us briefly consider the earlier‐mentioned issue about the interpretation of Kuhn. Since Hoyningen‐Huene thinks that Kuhn is a Constructivist, what could be his basis for claiming that Kuhn holds T1 and T2 to be nonetheless comparable? One basis (1993: 219–20) comes from Kuhn's talk in his later writings of incommensurability being merely “local” (1983: 670–1). This implies a more moderate relativism, hence more moderate Constructivism, than we have been discussing, a move back toward Kant. The local conceptual differences between T1 and T2 may leave the theories with a lot conceptually in common. To the extent of what is in common, T1 and T2 construct the same world. Reference to that common world could enable some theory comparison.5 But, to repeat, Incommensurability is worth rejecting whether or not, and to whatever extent, Kuhn is committed to it. Call the negation of Incommensurability, “Commensurability”, and consider: Realism → Commensurability This argument is not so easy. According to Realism, the world remains constant through theory change and a certain part of that world is the common domain of T1 and T2. It can then be argued that, by and large, these theories succeed in referring, or at least partially referring, to parts of this domain and this provides sufficient basis for theory comparison.6

(p.104) Assume that I am right about this. So we have a plausible route from Realism to Commensurability as well as one from Constructivism to Incommensurability. There are surely other plausible routes to Commensurability, ones that do not involve Realism. However, I claim that there are no plausible routes to Incommensurability that do not involve establishing, or simply assuming, Constructivism or some other form of antirealism. So, Incommensurability depends on antirealism. Hoyningen‐Huene, and his co‐authors Eric Oberheim and Hanne Andersen, may very well agree. In their review (1996) of Howard Sankey's book defending a commensurability thesis (1994), they criticize Sankey's argument on the ground that he “presupposes realism” (1996: 133). In another work, Oberheim and Hoyningen‐Huene insist that “incommensurability was not introduced within a realist context” but rather in (what I am calling) a Constructivist one (1997: 450). They go on to talk of the “blatant inefficacy” of my own arguments against incommensurability “from the perspective of the non‐realist proponent” (p. 452). It fits my prejudices nicely that the Incommensurability issue should rest on the Realism issue. And I am rather delighted by the stereotypical nature of this

Page 4 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics * particular debate: German Kantianism versus Australian Realism (1991b: pp. vii, x). In the light of this, it would, of course, beg the question against Incommensurability to presuppose Realism. Oberheim and Hoyningen‐Huene allege (pp. 451–2) that I did beg this question, but they are wrong. I did not presuppose Realism, I argued for it and against Constructivism.

4. Summary of an Argument for Realism Here is a summary of my argument for Realism.7 I start by observing that Realism about the ordinary observable physical world is a compelling doctrine. It is almost universally held outside intellectual circles. From an early age we come to believe that such objects as stones, cats, and trees exist. Furthermore, we believe that these objects exist even when we are not perceiving them, and that they do not depend for their existence on our opinions nor on anything mental. This Realism about ordinary objects is confirmed day by day in our experience. It is central to our whole way of viewing the world, the very core of common sense. Given this strong case for Realism, we should give it up only in the face of powerful arguments against it and for an alternative. There are no such arguments. That concludes the case for Realism. (p.105) What about scientific Realism, Realism about unobservables? The argument for it rests upon Realism about observables. The argument consists mainly in a simple but powerful inference to the best explanation: by supposing that the unobservables of science exist, we can give good explanations of the behavior and characteristics of observed entities, behavior and characteristics which would otherwise remain completely inexplicable.8 It can be seen, then, that the case for Realism partly rests on rejecting alternatives. In the present context, the alternative that we particularly need to reject is Constructivism. I shall start with some criticisms of Constructivism and then consider the arguments for it.9

5. Criticizing Constructivism Constructivism is surely the metaphysics of the twentieth century. It has its origins in the work of a great philosopher and has been urged by some outstanding ones. Despite this popularity, and the respect due to our elders and betters, we should not close our eyes to the fact that Constructivism is prima facie absurd, a truly bizarre doctrine. This emperor has no clothes.10 To start with, the idea of noumenal things‐in‐themselves is explanatorily useless and probably incoherent. Constructivists are attracted to things‐in‐themselves to provide an external constraint on theorizing. The plausibility of the view that there is some external constraint is, of course, overwhelming: there must be something outside us determining that some theories are better than others. However, things‐in‐themselves provide the appearance of constraint without the Page 5 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics * reality. Since we can, ex hypothesi, know nothing about things‐in‐themselves, we can know nothing about the mechanisms by which they exercise their constraint, nor can we explain or predict any particular (p.106) constraint. For Kant himself, the very idea of causal constraint by the noumenal world is incoherent because CAUSALITY is one of the concepts imposed by us. So causality is part of the phenomenal world and cannot hold between the noumenal and phenomenal worlds. If this is not the position of a Constructivist, it surely ought to be. Why should causality be exempt from the rule of creation by imposition?11 If it is not exempt, the Constructivist faces the same problem that has baffled Kant scholars for years: the nature of the non‐causal constraint exercised by things‐in‐ themselves. Frederick Jameson captures the mysteriousness of the noumenal world (in discussing structuralism): it is “a formless chaos of which one cannot even speak in the first place” (1972: 109–10). It is hardly ever mentioned without the protection of scare quotes or capital letters. Yet mentioned it often is. And, given the role that Constructivists want the noumenal world to play, it is not surprising that they should try to tell us about it. Yet, ex hypothesi, this is to attempt the impossible. For example, consider the problem of specifying the common domain of rival theories. The Realist can do this in terms of shared referents or, at least, shared partial referents. How can a Constructivist like Kuhn do it? Hoyningen‐ Huene points out that Kuhn appears to his critics to have the view that rival incommensurable theories “bear the same relation to one another as, for example, theories of the unconscious bear to theories on the stability of globular star clusters . . . [they] address differently constituted regions of the world” (1993: 218–19). Hoyningen‐Huene rejects the analogy, claiming that whereas “the latter theories have totally different domains . . . incommensurable theories target roughly the same object domain, as far as the world‐in‐itself is concerned” (p. 219). But this talk, which treats the noumenal world as if it were the Realist world, makes no sense:12 we can know nothing about what is targeted in an unknowable world.13 The noumenal world adds only an invisible figleaf to the naked idealism of Constructivism. If the leaf is dropped we are left with a modified Constructivism, which seems to be the preferred view of Putnam (1981: 61–2, 83) and the later (p.107) Kuhn (1979). On this view, no account of constraints on our theorizing can be given. The modified Constructivist might deny that there are any constraints: we can think anything we like. That is not plausible (to put it delicately). Alternatively, he might claim that there are constraints but we can, in principle, say nothing about them: it is just an inexplicable brute fact that we cannot think anything we like. This replaces the earlier incoherence with silence. It is hardly an appealing position.

Page 6 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics * Worse still, if that is possible, is the idea that we make the known world of stones, trees, cats, and the like with our concepts. It is common to convey this idea with the help of the cookie‐cutter metaphor: the dough (the noumenal world) is independent of the cook (us); the cook imposes cookie‐cutters (concepts) on the dough to create cookies (appearances). But how could cookie cutters in the head literally carve out cookies in dough that is outside the head? How could dinosaurs and stars be dependent on the activities of our minds? It would be crazy to claim that there were no dinosaurs or stars before there were people to think about them. Constructivists do not seem to claim this. But it is hardly any less crazy to claim that there would not have been dinosaurs or stars if there had not been people (or similar thinkers).14 And this claim seems essential to Constructivism: unless it were so, dinosaurs and stars could not be dependent on us and our minds.15 Finally, there is an old problem for relativism: arbitrarily excluding from the scope of the theory something dear to the theorist's heart. In this case, why do the languages, concepts, cultures, and so forth, that do the worldmaking not themselves exist only relative to . . . ? Relative to what? Themselves? The “texts” themselves start to shimmer and lose their reality. Constructivists typically vacillate between talk of theories or experience and talk of the world.16 This vacillation is important to the appeal of their message. For, although it is false that we construct the world by imposing concepts on the world, it is plausible to suppose that we construct theories of the world by (p. 108) imposing concepts on experience of the world. The vacillation helps to make the falsehood seem true.

6. Rejecting Two Kantian Arguments for Constructivism What then is the case for Constructivism?17 It arises out of alleged problems for Realism. I shall start with two Kantian arguments and then consider two arguments that Hoyningen‐Huene proposes on behalf of Kuhn. The main ingredients for one argument come straight from Kant. How can we save knowledge in the face of Cartesian doubt? The gap between the knowing mind and the Realist world of independent objects is alleged to make knowledge of those objects impossible. So the world we know about cannot be the Realist world. The only sort of world we could know about is one we partly constitute with our theories. Finally, we add relativism to this Kantian brew: when our theories change radically, so must our worlds. A semantic variant of this argument can be abstracted from contemporary antirealist discussions. The gap between the knowing mind and the Realist world makes it impossible to refer to that world. So the world we refer to cannot be that world but must be a world we construct. With radical theory change goes reference to a different world.

Page 7 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics * I think that these sorts of argument are implicit in the discussions of Kuhn and Feyerabend and have frequently attributed a semantic one to them. Hoyningen‐ Huene and Oberheim dismiss the attribution (1997: 448–9). I am not convinced by their dismissal but will not argue the matter. It is important to see what is wrong with these sorts of argument, whether or not they are to be found in Kuhn and Feyerabend. The argument I have attributed to Kuhn and Feyerabend starts from a description theory of reference according to which the reference of a term in a scientific theory depends on the descriptions (other terms) the theory associates with it: it refers to whatever those descriptions (or most of them) pick out. Now with theory change, particularly radical theory change, is likely to go the view that those descriptions do not pick anything out. So, if we take (p.109) the Realist view that a referent must exist independently of theory, we must conclude that, from the new perspective, the term in the old theory does not refer. This will be true even of an “observational” term; think, for example, of the change in descriptions associated with ‘The Earth’ that came with the Copernican revolution. However, if we abandon Realism we can take the old terms to refer to entities constituted by the old theory, entities that exist relative to that theory but do not exist “absolutely”. Such arguments should give us pause.18 Speculations about what and how we can know and refer have led to disaster: a bizarre metaphysics. But why should we have any confidence in these speculations? In particular, why should we have such confidence in them that they can undermine a view as commonsensical as Realism? A Moorean point is appropriate: Realism is much more firmly based than these speculations that are thought to undermine it.19 We have started the argument in the wrong place: rather than using the speculations as evidence against Realism, we should use Realism as evidence against the speculations. We should, as I like to say, “put metaphysics first”. Indeed what support are these troubling speculations thought to have? Not the empirical support of the claims of science. This is most obvious with the epistemological speculations, but it is fairly obvious with the semantic ones. Thus, no attempt is ever made to establish empirically that a description theory of reference is appropriate for these scientific terms. In brief, the support for these speculations is thought to be a priori.20 Reflecting from the comfort of (p. 110) armchairs, Constructivists decide what knowledge and reference must be like, and from this infer what the world must be like: A priori epistemology/semantics → a priori metaphysics. The Moorean point alone casts doubt on this procedure and the philosophical method it exemplifies, the a priori method of “First Philosophy”. But we can do better: the doubt is confirmed by the sorts of considerations adduced by Quine Page 8 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics * (1952: pp. xi–xvii; 1961: 42–6). These considerations should lead us to reject a priori knowledge and embrace “naturalism”, the view that there is only one way of knowing, the empirical way that is the basis of science.21 From the naturalistic perspective, philosophy becomes continuous with science. And the troubling speculations have no special status: they are simply some among many empirical hypotheses about the world we live in. As such, they do not compare in evidential support with Realism. Experience has taught us a great deal about the world of stones, cats, and muons but rather little about how we know about and refer to this world. So epistemology and semantics are just the wrong places to start the argument. Instead, we should start with an empirically based metaphysics and use that as evidence in an empirical study of how we know and refer: epistemology and semantics themselves become part of science, they become “naturalized”: Empirical metaphysics → empirical epistemology/semantics. And when we approach our metaphysics empirically, Realism is irresistible. Indeed, it faces no rival we should take seriously.

Quine is fond of a vivid image taken from Otto Neurath. He likens our knowledge —our “web of belief”—to a boat that we continually rebuild whilst staying afloat on it. We can rebuild any part of the boat but in so doing we must take a stand on the rest of the boat for the time being. So we cannot rebuild it all at once. Similarly, we can revise any part of our knowledge but in so doing we must accept the rest for the time being. So we cannot revise it all at once. And just as we should start rebuilding the boat by standing on the (p.111) firmest parts, so also should we start rebuilding our web. Epistemology and semantics are among the weakest places to stand.22 We start with metaphysics. We have already summarized our argument for a Realist one. Does the history of science, so nicely revealed by Kuhn and Feyerabend, demand any modification of this Realism. As theories have changed, have we abandoned our belief in entities that we previously thought existed? First, consider observables. Theoretical progress certainly results in the addition of new observables, terrestrial and celestial, to our catalogue. But there have been very few deletions. Cases like witches, Piltdown Man, and Vulcan are relatively rare. There have been some mistakes, but there is nothing in our intellectual history to shake our confidence that we have steadily accumulated knowledge of the make‐up of the observable world. We have been wrong often enough about the nature of those entities, but it is their nature we have been wrong about. We have not been wrong about the fact of their existence. In brief, theory change is no threat to Realism about observables. Furthermore, we should be sufficiently confident of this metaphysics to reject any theory of language that fails to fit it. It is not that the historical facts of theory change, together with a description theory of reference for scientific Page 9 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics * terms, show Realism to be false. Rather, those facts, together with Realism, show the description theory for those terms to be false. Many ideas for other theories of reference compatible with Realism have emerged in recent times.23 It is less easy to rebut Kuhn and Feyerabend on unobservables. It is plausible to suppose that we have often been wrong in thinking that an unobservable exists. Even there, Kuhn and Feyerabend's commitment to the description theory leads them to exaggerate our degree of error. Without these exaggerations, scientific realism is not in much trouble: while our views of, say, the subatomic particles have changed and evolved, we still believe in the entities posited by Bohr and Rutherford. At most, the history of science should make us cautious in our commitment to unobservables. It should not lead us into Constructivism (1984/ 1991b: 9.4).24 (p.112) The discussion in this section is intricate enough to warrant a summary. The background to the discussion is our earlier argument for the plausibility of Realism (sec. 4) and the implausibility of Constructivism (sec. 5). The Kantian arguments for Constructivism and against Realism rest on speculations in epistemology and semantics. Against the background—the plausibility of Realism and implausibility of Constructivism—the Moorean point is that we should prefer Realism to the Kantian speculations; we should put metaphysics first. This point is good on its own but when supported by naturalism it is formidable. From the naturalistic perspective these speculations cannot be supported a priori and they do not come close to having the empirical support enjoyed by Realism. The arguments for Constructivism use the wrong methodology and proceed in the wrong direction. I turn now to the arguments that Hoyningen‐Huene offers on behalf of Kuhnian Constructivism.

7. Rejecting Two Kuhnian Arguments for Constructivism Oberheim and Hoyningen‐Huene dismiss the idea that Kuhn and Feyerabend were led to incommensurability and Constructivism by their semantics. Rather, these doctrines “were the result of attempting to achieve a historical understanding of the development of science” (1997: 449). How did they result? We need to be particularly concerned with how the history of science is supposed to support Constructivism, because without Constructivism the case for Incommensurability collapses (sec. 3). In his book, Hoyningen‐Huene emphasizes Kuhn's revolutionary historiography of science which was to be the basis for his view of science: “Kuhn's goal is to propose a new picture of science and scientific development, in particular of scientific progress, grounded in this new historiography” (1993: 13). According to this historiography, episodes in the history of science are best studied and explained in their own right and not from the perspective of contemporary science. We need to see the world as the scientists of those times did, which may Page 10 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics * be difficult because those scientist saw the world so differently. This practice “requires an exact reconstruction of the period's conceptual system” (p. 20). It leads to “a more alien, yet at the same time more reasonably alien, scientific past than the old historiography” (pp. 22–3). Hoyningen‐Huene goes on to claim that the experience of the historian practicing this new historiography justifies the plurality‐of‐phenomenal‐worlds thesis for the practice “may produce a (p. 113) different phenomenal world—different, that is, by comparison with the historian's own phenomenal world” (p. 38). But, of course, this experience could justify the plurality of phenomenal worlds only if it has already been established that there is a phenomenal world at all in the relevant Kantian sense, a world partly constructed by the imposition of concepts. Perhaps Hoyningen‐Huene takes this to have been established already by Kant. Mooreans and naturalists think that nothing could be further from the truth (sec. 6). Kuhn and Feyerabend's illuminating view of the history of science alone does not support Constructivism. No more does their illuminating view of the epistemology of science. They claim that all statements, even “observation” statements, are epistemically theory‐laden.25 This claim is not novel—it is central to the Duhem–Quine thesis—but Kuhn and Feyerabend did more than anyone to establish it, producing a marvelous array of scientific evidence in its favor. Hoyningen‐Huene sees an argument here for the plurality‐of‐phenomenal‐ worlds thesis: that thesis helps explain theory‐ladenness (pp. 36–7). And so, in a way, it does. But this is not the best explanation because it is based on an implausible metaphysics. The best explanation is based on Realism. Against a Realist background theory‐ladenness is readily explained by a naturalized epistemology that appeals to the psychological facts of belief formation. Theory‐ ladenness would provide a reason for going beyond Kant's metaphysics by adding a plurality thesis if we had already established that metaphysics. But we have not. Constructivism combines the Kantian idea of a phenomenal world, the Kantian idea of a noumenal world, and relativism. Kuhn and Feyerabend's views of history and epistemology would support the addition of relativism to the two Kantian ideas but they do nothing to support the ideas themselves. Let us take stock. I have argued that Constructivism leads to Incommensurability and Realism leads to Commensurability. Furthermore, I claimed that without Constructivism, or some other form of antirealism, there is no plausible route to Incommensurability (sec. 3). So, we can refute Incommensurability by establishing Realism. The case for Realism is so strong that we should give Realism up only in the face of powerful arguments against it and for an alternative (sec. 4). I have argued against Constructivism, the alternative that concerns us, emphasizing the bizarre and mysterious nature of the doctrine (sec. 5). Adopting first a (p.114) Moorean and then a naturalistic perspective, I have rejected two Kantian arguments for Constructivism (sec. 6). We have just seen Page 11 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics * that the arguments that Hoyningen‐Huene proposes on Kuhn's behalf presuppose rather than argue for the Kantian core of Constructivism. Constructivism is not only implausible, it is unsupported. Realism still stands. If I am right about all this, the case against Incommensurability is made. But one matter remains.

8. Rejecting Meta‐Incommensurability Hoyningen‐Huene, Oberheim, and Andersen, struck by the inconclusiveness of the realism debate, make the tentative proposal that the debate itself involves incommensurability, what they call “meta‐incommensurability”: because of meaning differences “effective means of rational meta‐theory choice are not yet at hand” (1996: 138). If they are right about this, of course, the argument I have just summarized, purporting to give a rational basis for choosing Realism over Constructivism, must be a failure. In support of meta‐incommensurability, they claim: there are several terms that change meaning when one crosses the line from realism to nonrealism: namely, ‘reality’, ‘world’, ‘theory comparison’, ‘fact’, and even ‘reference’ itself . . . they purportedly refer to different things, based on the different metaphysical assumptions each party brings to the debate. (pp. 138–9) This simply assumes that because the realist and the non‐realist make different assumptions involving a word, and hence associate different predicates with it, its meaning and reference differ for them. This is to assume a description theory for these words. Why should we do that? We already know that description theories cannot be true for all words: these theories explain the meaning and reference of a word in terms of the meaning and reference of other words, a process that cannot go on for ever. Some words must “stand on their own feet” being explained (at least partly) in terms of some sort of relation to reality that is not mediated by another word (1996a: 159– 60). We have been given no reason to suppose that the words in question are not of that sort.

The initial proposal of meta‐incommensurability was cautious but the caution of Oberheim and Hoyningen‐Huene soon disappears: “ ‘truth’, ‘world’, ‘fact’, ‘theory comparison’, and ‘reference’ . . . clearly have different meanings for the realist and the non‐realist” (1997: 453; my emphasis). Furthermore, they reject the idea that this thesis rests on semantic considerations, in particular (p.115) on a description theory (p. 460).26 Rather, it is “based on a contemporary historical case study” (p. 453). They see meta‐incommensurability as the best explanation of a range of phenomena which they observe in the realism literature: “communication difficulties”; accusations that arguments are “circular” or “question‐begging”; the sense that arguments are “indecisive” (pp. 453–9). This is ingenious. The phenomena that they identify cannot be denied. And if there is meta‐incommensurability we would certainly expect these Page 12 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics * phenomena. But is meta‐incommensurability really the best explanation of them? I think the answer is, “No”. I start with a qualification. I would be the last to claim that there are no meaning differences in the realism debate, nor that such differences play no role in producing phenomena of the sort identified: I have often complained of the confusion over the word ‘realism’ itself.27 So I accept that incommensurability sometimes plays an explanatory role of the sort Oberheim and Hoyningen‐Huene suggest. But this sort of incommensurability can easily be removed with a bit of terminological care. It is very different from their meta‐incommensurability rooted in metaphysical difference, difficult if not impossible to remove. Here is a reason for thinking that meta‐incommensurability may not be the best explanation of the phenomena that they identify. Communication difficulties, accusations of circularity, and similar phenomena often occur in disagreements where it is very implausible that there are meaning differences. These may be humdrum disagreements of everyday life or “low‐level” scientific disagreements within a paradigm. It is clear, then, that the occurrence of such phenomena does not depend on meaning differences between positions. Sometimes we need another explanation of these phenomena. What other explanations are available? I shall briefly describe four features of the cognitive life that might contribute to such explanations. In light of these, it seems to me clear that meta‐incommensurability is not the best explanation of the phenomena of the realism debate. The first feature is the theory‐ladenness of observation, a feature much emphasized by Kuhn and Feyerabend: what we make of our experience depends very much on what we already believe, on what we expect. This applies to our linguistic experience as well. We interpret sentences that mean p, in our language as much as the speaker's, to mean q because, given our beliefs, (p. 116) including those about the speaker, that is what we expect her to be saying in the circumstances. We have “communication difficulties”. This is surely part of the explanation of a phenomenon experienced by all journal reviewers: the author of a critical paper seems unable to read what is plainly on the page criticized. The second is the general difficulty of putting together good arguments, and the undeniable extra difficulty of doing so in philosophy. It is very hard to think straight and particularly hard to do so about such abstract topics as realism. This difficulty is surely the explanation of many arguments in ordinary life and science that are circular, question‐begging, or indecisive. And it is plausible to think that it is part of the explanation of these failures in many philosophical arguments.

Page 13 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics * The third is a bit more speculative. It seems plausible to suppose that we humans suffer from a certain rigidity in our thinking that makes it difficult for us to contemplate alien views. And the more alien and the more global the view, the greater the difficulty. So we should expect great difficulty in the realism debate. Sometimes, no doubt, difficulties with alien views arise from meaning differences but there seems no good reason for supposing that they all must. Why should we not regard it as a brute fact about us that, even with meanings constant, we find it difficult to “get our heads around” alien views? Certainly we do not know enough about psychology and semantics to rule this out. Finally, a rather obvious feature of many debates in philosophy and elsewhere is the ego involvement of the participants: the participants are wedded to their views and “want to win”. This can blind them to the faults in their arguments leading to question begging, circularity, and the like. Given the general availability of explanations built out of these four features, it is appropriate to invoke meta‐incommensurability as an explanation of a particular phenomenon of question begging, circularity, etc., only if we have some independent reason for thinking that meaning differences are involved. This independent reason must arise from a semantic theory, however primitive. So an historical case study that does not invoke semantic considerations will not do the job of justifying a meta‐incommensurability explanation of the phenomena of the realism debate, contrary to what Oberheim and Hoyningen‐Huene desire. And, so far as I can see, the only semantic consideration that they can invoke to do the job is a description theory of reference. As I have pointed out, we have been given no reason to suppose that this theory is appropriate here. I conclude that the meta‐incommensurability thesis is false. My earlier argument for Realism and hence against Incommensurability stands. (p.117) Finally, it is just as well that the meta‐incommensurability thesis is false because it is dangerous. If the thesis were true both sides of the realism debate would be immune to rational criticism. Post‐modernists would relish such a conclusion but it should dismay the rest of us.28 Postscript to “Incommensurability and the Priority of Metaphysics”

1. Lynch's Metaphysical Pluralism I describe Constructivism as “the metaphysics of the twentieth century”. Michael Lynch's ingenious book, Truth in Context (1998),29 closes out that century with an argument for a Constructivist doctrine he calls “pluralism”: “there can be more than one true story of the world; there can be incompatible, but equally acceptable, accounts of some subject matter” (p. 1); “reality is not absolute” (p. 2) “true propositions and facts are relative to conceptual schemes or worldviews” (p. 3). The doctrine admits “the existence of what Putnam has derisively called THE WORLD and what Kant called the noumenal realm” (p. 5). Page 14 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics * “Incompatible, but equally acceptable, conceptual schemes . . . are ways of dividing reality into objects and kinds of objects” (p. 11). An engaging feature of Lynch's discussion is the way it reveals that, deep down, he recognizes what bad news this doctrine is. We see this in his tendency to shy away from the consequences of the doctrine, to tone it down. Lynch would like to offer us “Goodman Lite”, “Goodman with a Human Face”. Thus, in discussing Goodman, Lynch remarks that “the idea that there is literally more than one world . . . is anti‐intuitive in the extreme” (p. 95). Yet this idea seems to be at the very core of his pluralism. So how does Lynch avoid it? How can we all share the one world despite our different conceptual schemes? Lynch suggests that people “share . . . a minimal world . . . [which] can present itself to them in radically distinct and incompatible ways” (p. 95). What world could this be? Those in the Kantian tradition are attracted to the idea that we share the unknowable noumenal world. But this, as Goodman (p.118) rightly says, is a world not worth fighting for (1978: 20). And Lynch does not want to fight for it: “a single, shared, ‘real’ world is clearly impossible if what we mean by ‘the real world’ is a realm of being literally beyond ourselves, separate from us and our concepts. Such a concept of noumenal reality is incoherent.” He is fighting for another world: “But the concept is coherent if we mean by the term ourselves and all other things” (p. 155). So his view seems to be that the shared world is the knowable phenomenal one of stones, trees, cats, and ourselves. This view is also suggested by his rejection of Goodman's idea that there are many phenomenal worlds. But how could we share the one phenomenal world, given his Constructivist view that phenomenal worlds are created by conceptual schemes that differ? His answer is to move toward Kant's universality:30 all conceptual schemes have a minimal common part. To the extent that we share conceptual schemes, we share the phenomenal worlds we create by imposing those schemes.31 But clearly we are not yet out of the Goodmanian woods. For, to the extent that we do not share conceptual schemes, we are all creating different phenomenal worlds. And, of course, if there is to be any interesting pluralism left in Lynch's view, the extent to which we do not share schemes must be considerable. Indeed, the implication of Lynch's discussion, particularly his motivation for pluralism (to be discussed below), is that the extent is considerable. We live in different worlds but Lynch, unlike Goodman, lets us all share a common part. There is a way to avoid this metaphysical abyss. People do indeed share a world which “can present itself to them in radically distinct and incompatible ways”. It is not a minimal, noumenal, or phenomenal world; it is the Realist world. This world is not constructed by our way of presenting it: it is independent of our cognitive activities. Throughout his discussion, we see signs of Lynch being drawn willy‐nilly toward Realism. He should give in and enjoy it.

Page 15 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics * Lynch's pluralism seems to make a person's world dependent on her mind for its existence and nature. For, the world is relative to the conceptual scheme she uses to construct it and that scheme is dependent on her mind; scheme relativity implies mind dependence. So, “if there were not minds, the earth would not be spherical, not rotate around the sun, nor occupy an orbit between that of Venus and Mars”. With admirable good sense, Lynch describes this sort of idealism as “absurd” (p. 97). But how can he avoid the absurdity? (p.119) He tries to avoid it by extending his relativity to modal facts: “It is consistent with pluralism to acknowledge that, relative to one's conceptual scheme, there would still be stars without conceptual schemes” (p. 98). Ingenious, but it simply won't do! Any Realist should, of course, accept a relativity claim like this. It amounts to the claim that according to a certain conceptual scheme, stars are independent of conceptual schemes. The Realist certainly agrees with that. But the Realist wants to go further: stars are independent of conceptual schemes not just relatively but absolutely; there would have been stars even if there had been no conceptual schemes period. It is the denial of this that is both idealist and absurd. Lynch is only too well aware of this Realist response, seeing it as another sign of the “intractability that surrounds such debates” (p. 98). I shall discuss this alleged intractability in a moment. Meanwhile, I think Lynch would do better to see the Realist response as undermining any motivation that his pluralism has. He can of course dig in his heels and insist that no claim is true except relative to a conceptual scheme, not even a claim that is common to all conceptual schemes. He can claim that his pluralism itself is true relative to a pluralist scheme and Realism is true relative to a Realist scheme, but neither is true absolutely. But why should we believe any of this? Lynch has interestingly little to say in answer to this question. Constructivists typically make three related moves to bolster their case. These moves are initially beguiling, but spurious on closer inspection. Lynch follows in this tradition. My paper has already mentioned the first move: vacillation between talk of theories or experience and talk of the world. Thus Lynch sees the Kantian idea that “conceptual schemes . . . are ways of dividing reality into objects and kinds of objects” as rooted in the idea that “a conceptual scheme . . . shapes raw sensory experience in a certain way (p. 11). But these are radically different ideas. Whereas the latter idea of imposition on experience is plausible, the former idea of imposition on the world is wildly implausible. The vacillation is important to the appeal of Lynch's message for it helps to make the wildly implausible seem plausible.

Page 16 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics * The second move is to draw Constructivist conclusions from truisms. (i) “The pluralist believes that we can say nothing outside of a conceptual scheme, and therefore that all facts are relative to conceptual schemes” (p. 97). This is an awful non sequitur. Everyone believes that we can say nothing outside of a conceptual scheme because to say something is to express a thought and a thought must be in some conceptual scheme. Nothing metaphysical follows from this triviality. (ii) “We cannot consider an object as it is independently of our conceptualizations” (p. 11). Indeed, we cannot think about an object (p.120) without thinking, hence using concepts. But, contrary to what Lynch suggests, this gives no support at all to the Kantian view that our objects are constructed by this thinking. The third move is to suppose that Realism is committed to denying such truisms; indeed, Realism is thought to be committed to knowing the unknowable and speaking the unspeakable. Thus the Realist is thought to believe that we can “step outside of our skins and compare our thoughts to the world as it is in itself” (p. 145). In Putnam's immortal phrase, the Realist claims a God's Eye View. But this view of Realism is a caricature, as I pointed out earlier (Chapter 2, part II, aberration 7). No sane Realist believes any of it. What Realists believe is that we can judge whether theories are true of reality, the nature of which does not depend on any theories or concepts. Lynch's most explicit motivation for his pluralism is reminiscent of Hoyningen‐ Huene's “meta‐incommensurability”. Lynch offers pluralism as the explanation of “the apparent intractability of metaphysical disputes” (p. 11); for example, disputes over personal identity, counting objects, substance, and numbers (pp. 17–18). Yet in all cases the Realist has explanations available that are far more plausible than pluralism. Thus, the Realist might claim that in some disputes there is simply no fact of the matter which position is correct. This seems a plausible response to some split‐brain personal identity disputes. Or the Realist might claim that some disputes can be settled only by acknowledging an implicit relativity to frameworks. This seems a plausible response to the problem of counting objects. Or the Realist might claim that the dispute is over a pseudo problem. This seems a plausible response to the dispute over substance (and, in my Quinean view, the correct response; cf. Chapter 1). Or the Realist might claim that the dispute is too difficult for us at present but we can hope to do better in future. This response seems a plausible one to the dispute over numbers. In conclusion, Lynch supplies surprisingly little motivation for his Constructivist doctrine and what he does supply is extremely unconvincing. He writes as if the doctrine is prima facie plausible and appealing. Sadly, he is not alone in this view.

Page 17 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics * Notes:

(*) First published in P. Hoyningen‐Huene and H. Sankey (eds.), Incommensurability and Related Matters (Devitt 2001a). Reprinted with kind permission from Kluwer Academic Publishers. (1) Eric Oberheim and Hoyningen‐Huene (1991) raise another interpretative issue. They distinguish between “Feyerabend and Kuhn's conception of incommensurability, in the sense that it was a result arrived at through historical analysis and . . . a semantic conception of incommensurability within a realist framework”. Whereas “contemporary literature” has been concerned with the semantic conception, Feyerabend and Kuhn's conception was not “restricted to such semantic issues” (p. 447). I have two comments. First, here and elsewhere, Oberheim and Hoyningen‐Huene confuse the question “What is the argument for Kuhn and Feyerabend's incommensurability?” with the question “What is the nature of that incommensurability?”. Whether or not the argument for (or against) their incommensurability is from historical analysis, within a realist framework, or whatever, is one thing, what that incommensurability is is another. (My “Maxim 1” makes an analogous point about the realism issue: “In considering realism, distinguish the constitutive and evidential issues”; 1984/ 1991b: 3.) In particular, whether or not Kuhn and Feyerabend's incommensurability is semantic is a distinct matter from whether or not their argument for it is from “semantic theory” (pp. 447–52). My second comment addresses what Oberheim and Hoyningen‐Huene have to say on the former question, the one about the nature of incommensurability. They reject the common view, which I share, that Kuhn and Feyerabend's incommensurability thesis is semantic. They insist that the thesis “was intended to involve more than semantic issues”: it was intended to involve a neo‐Kantian antirealism (p. 450). I have no doubt that Kuhn and Feyerabend subscribe not only to a semantic thesis like my Incommensurability but also to an antirealist metaphysics. I do doubt that they conflated them under the one term ‘incommensurability’. But whether or not they did, they shouldn't: nothing but harm comes from the conflation of semantics with metaphysics. Or so I have argued (1984/1991b/1997a; Ch. 2, part I, of the present volume). To insist on distinguishing the semantic thesis from the metaphysical one is not, of course, to deny that the theses may be related. I shall explore the relation in sec. 3. (2) I explain and argue for this nonsemantic characterization of realism in my 1984/1991b and Ch. 2, part I, of the present volume. (3) [2009 addition] It is hard to avoid describing this mysterious metaphysics in ways suggesting that Kant is proposing two distinct worlds. But this suggestion is mistaken. The idea is not that there is a world of noumenal objects and a distinct world of phenomenal objects. Rather, the idea is that there is the one world consisting of objects both as they are in themselves, the noumenal, and as we know them because of the imposition of our concepts, the phenomenal. Page 18 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics * Perhaps an analogy will help. Consider the world of “natural” physical objects and stuffs: stones, trees, cats, water, oxygen, and the like. Now consider the world of “tools”: chairs, hammers, paperweights, and the like. Tools are usually artifacts made by us out of the natural world. Sometimes tools are not artifacts but simply parts of the natural world used as tools: stones, even cats, can be paperweights; chairs might have grown on trees. Stretching our imaginations, suppose that all natural objects and stuffs (apart from us) were parts of tools. There would not be a world of natural objects and a distinct world of tools. Rather, there would be one world consisting of objects and stuffs both as they are naturally, and as they are as tools because of the way we made and use them. (Given the obvious dependence of tools on us it is interesting to consider what it is to be realist about them; see Ch. 6, sec. 3, of the present volume for a discussion.) (4) Howard Sankey talks of Hoyningen‐Huene's “novel interpretation of Kuhn's philosophy of science, which presents the latter within a neo‐Kantian anti‐realist framework” (1991: 437). What is novel, of course, is not the neo‐Kantian (= Constructivist) interpretation itself, which has been widespread for years, but the thoroughness of the case for it, and the clear and detailed presentation of Kuhn's philosophy of science from that interpretative perspective. (5) Hoyningen‐Huene offers two other bases (pp. 220–1). I find these very unconvincing but will not argue the matter. (6) The argument for Commensurability in my 1979 is implicitly an argument from Realism along these lines and that in my 1984/1991b: sec. 9.6, is explicitly so (also Devitt and Sterelny 1999: 227–8). Sankey 1994 is an extended argument of this sort; see also Sankey 1998. (7) 1984/1991b: particularly ch. 5; see also 1999a and Ch. 3 of the present volume. (8) This simple argument should not be confused with a popular one captured by the slogan “Realism explains success”. The popular argument uses Realism to explain the observational success of theories where the simple one uses Realism to explain the observed phenomena (1984: 7.2; 1991b: 7.3). [2009 addition] See Ch. 4 of the present volume for detailed arguments for scientific realism. (9) For more details, see my 1991b: particularly ch. 13. (10) Richard Rorty thinks it absurd to say “that we make objects by using words” (1979: 276). Nicholas Wolterstorff thinks that in saying this the Constructivist must be “speaking in metaphor. If we took him to be speaking literally, what he says would be wildly false—so much so that we would question his sanity” (1987: 233). David Stove has this to say in his chapter “Philosophy and Lunacy: Nelson Goodman and the Omnipotence of Words”: “the statement Page 19 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics * that worlds can be made with words: a statement which, as Hume said of the doctrine of the real presence, ‘is so absurd that it eludes the force of all argument’ ” (1991: 31). [2009 addition] Mark Johnston asks: “Is there really a debatable issue here? Who could really believe that quite generally the world's structure is dependent on our classificatory activity?” (1993: 99). (11) According to Hoyningen‐Huene, “Kuhn stipulates [the noumenal world] to be spatiotemporal, not undifferentiated, and in some sense causally efficacious (1993: 34). One wonders what Kuhn's justification could be for this departure from Kant. (12) As, indeed, Hoyningen‐Huene admits in a footnote (n. 119)! The admission totally undermines his rejection. See also secs 2.2.c–2.2.e where Hoyningen‐ Huene wrestles mightily with Kuhn's attempts to describe the constraining role of the noumenal world (in the guise here of what Hoyningen‐Huene calls “object‐ sided stimuli”). This discussion brings out nicely the futility of such attempts to speak the unspeakable. (13) Alternatively, if we can know what is targeted in the noumenal world, why can we not know other things about that world? Why then does the noumenal world not collapse into the knowable Realist world? For more along this line, see Sankey 1991: 439–40. (14) [2009 addition] Constructivism is so bizarre and mysterious that one is tempted to seek a charitable reinterpretation of constructivist talk. But, sadly, charity is out of place here (1991b: 239–41). (15) [2009 addition] Our earlier tool analogy in n. 3 may help to highlight these problems of Constructivism. (1) The natural world constrains what can be made into a tool and used as a tool: we cannot make a chair out of oxygen; a feather cannot be a hammer. There is nothing unknowable about these constraints: we can explain the mechanisms by which the constraints are exercised; we can explain and predict particular constraints. Contrast this with the mysterious constraint that the noumenal is alleged to have on the phenomenal. (2) We make something a tool either by literally constructing it from the natural world or by using part of the natural world as a tool. Either way, our role as a maker of tools is straightforward. Contrast this with our mysterious role as an alleged maker of the phenomenal world. (16) E.g. Whorf 1956: 55, 162, 213, 253; Kuhn 1970: 114, 117; Feyerabend 1978: 70; Latour and Woolgar 1986: 183; Hawkes 1977: 28. (17) I am here seeking rational explanations of Constructivism; and I have done so at greater length elsewhere (1991b: 13.4–13.7). But the popularity of a doctrine that is so bizarre and mysterious cries out for a different sort of explanation. For some learned, and very entertaining, suggestions, see Stove Page 20 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics * 1991. Stove thinks that antirealism, like religion, stems from our need to have a congenial world. For some suggestions by Georges Rey along similar lines, see my 1991b: 257 n. (18) [2009 addition] Ch. 6 of the present volume considers another example of an argument from semantics to an antirealist metaphysics that should equally give us pause: it is an argument from the global response‐dependency of concepts to “Worldmaking”. Michael Lynch (1998) may be proposing a further example in arguing for his constructivistic doctrine “pluralism”. Inspired by Wittgenstein's discussion of the concept of a game, Lynch argues that concepts are “fluid” in that they are subject to possible extension in the face of unforeseen circumstances” (p. 61). These extensions can yield “inconsistent applications [that] are equally correct” (p. 62). Lynch sees this conceptual pluralism as supporting his metaphysical pluralism: “there is a connection between pluralism at the conceptual level and pluralism at the metaphysical level” (p. 5). See the Postscript to this paper for more discussion of Lynch's view. The discussion that follows in this section is similar to that of underdetermination and extreme skepticism in Ch. 3, secs. 4–6, and of semantic ascent in Ch. 15, sec. 4, of the present volume. (19) Steven Hales drew my attention to the Moorean nature of this point. Note that the point is not that Realism is indubitable, to be held “come what may” in experience: that would be contrary to naturalism. The point is that, prima facie, there is a much stronger case for Realism than for the speculations. (Thanks to Paul Boghossian.) (20) In this respect, Kuhn and Feyerabend are very much part of the Establishment, despite the radical nature of their philosophy of science. They are part of a semantic tradition, one that includes the positivists before them and is still dominant to this day, that proceeds as if semantics is, at bottom, rather easy. At the level of terms (or concepts) we can rely on a priori intuitions about which features of a term (or concept) constitute its meaning. So to determine the meaning all we have to do is describe how the term (concept) is learned and used and we can simply “see” what its meaning is. Feyerabend remarks that “conversations about meaning belong in the gossip columns” (1981a: 113). Since his own writings are riddled with such conversations, we must see his remark as characteristic waggishness. Of course, one might wonder how an empirical theory of meaning should proceed. I think that it is very difficult to say. My attempt is 1996a: ch. 2. (21) A particularly important consideration against the a priori, in my view (1996a: 2.2), is the lack of anything close to a satisfactory explanation of a nonempirical way of knowing. We are told what this way of knowing is not—it is not the empirical way of deriving knowledge from experience—but we are not told what it is. Rey 1998 and Field 1998 have a more tolerant view of the a priori. Page 21 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics * My 1998, which is Ch. 12 in the present volume, is a response. [2009 addition] See also Ch. 13. (22) [2009 addition] Alexander Miller (2003: 215–16, n. 19) finds some confusion in my embracing the Quinean image of Neurath's boat as well as (i) “the claim that metaphysics is prior to semantics”; and (ii) the view, which Miller construes as an a priori constraint, that constructivism is something we should “never countenance for a moment” (1991a: 52; Ch. 2 in the present volume). But there is really no confusion. Concerning (i), I certainly accept the Quinean view that evidence on the metaphysical issue of realism could come from anywhere, including from semantics. This is quite consistent with a priority claim resting on the view that the empirical case for realism is much stronger than that for any epistemological or semantic thesis. Concerning (ii), the point is not that constructivism should be rejected a priori but rather that it should be rejected because of the overwhelming considerations against it of the sort adduced in the present chapter; see also n. 10 above. (23) See e.g. Putnam 1975; Kripke 1980; Dretske 1981; Millikan 1984. (24) [2009 addition] See also Ch. 4, sec. 4.2, and postscript, sec. 3, in the present volume. (25) Not to be confused with another thesis that they have also sometimes seemed to hold: that “observation” terms are semantically theory‐laden. This thesis amounts to a description theory of those terms. One could believe the epistemological thesis that one's judgment about the application of, say, ‘rabbit’ in a certain situation depends on all sorts of background assumptions about rabbits, whilst holding the semantic thesis that the meaning of ‘rabbit’ depends not on its relation to any other term but entirely on its direct causal relation to rabbits. (26) This is not the same, of course, as rejecting that meta‐incommensurability is semantic (as the characterization that opens this section shows it to be). However, they sometimes write as if the confusion of the argument and nature questions that I criticized in their discussion of incommensurability (n. 1) may also be present in their discussion of meta‐incommensurability (pp. 453, 461). (27) [2009 addition] See e.g. Ch. 2 in the present volume, part I. (28) A draft of this chapter was delivered at the conference, “Incommensurability (and other related matters)”, held in Hanover in June 1999. Eric Oberheim was my commentator. I am indebted to his commentary for several improvements in my presentation. (29) All unidentified references in this postscript are to this book. My discussion draws on an unpublished paper, “An Unmade World”, delivered at an author‐ Page 22 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Incommensurability and the Priority of Metaphysics * meets‐critics session on Lynch's book held by the Society for Realist/Antirealist Discussion at the American Philosophical Association Convention in New York, Dec. 2000. I am indebted to Michael Lynch for comments. (30) As he made clear in his response at the author‐meets‐critic session. (31) In sec. 3 of the present chapter we noted a similar move in the later writings of Kuhn (1983: 670–1).

Access brought to you by:

Page 23 of 23

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Global Response Dependency and Worldmaking *

Putting Metaphysics First: Essays on Metaphysics and Epistemology Michael Devitt

Print publication date: 2009 Print ISBN-13: 9780199280803 Published to Oxford Scholarship Online: February 2010 DOI: 10.1093/acprof:oso/9780199280803.001.0001

Global Response Dependency and Worldmaking * Michael Devitt (Contributor Webpage)

DOI:10.1093/acprof:oso/9780199280803.003.0007

Abstract and Keywords ‘Worldmaking’ is the antirealist doctrine that we make the known world with our concepts. It is Constructivism without a commitment to relativism. This chapter considers the relation between Worldmaking and the global responsedependency theory of concepts urged by Pettit. According to this theory all concepts are of dispositions to produce a certain sort of response in normal humans in normal conditions. Pettit denies that this theory leads to Worldmaking. The chapter argues that he is wrong. The theory leads to the view that all properties are response-dependent and this leads to Worldmaking. For that reason alone Pettit's theory of concepts should be rejected. Keywords:   worldmaking, realism, constructivism, relativism, global response-dependency, Pettit

1. Introduction In Chapters 2 to 4 I have argued for realist doctrines that we can sum up well enough as follows: Realism: Tokens of most commonsense, and scientific, physical types objectively exist independently of the mental. At the observable level, the tokens in question are of stones, trees, cats, and the like; at the unobservable level, they are of electrons, muons, curved space‐time, and the like. For the purposes of this chapter we can attend only to Realism about observables.

In Chapter 5 (secs. 5–7), from this Realist perspective, I have argued against an idealist doctrine usually called “Constructivism”. Constructivism starts from two Page 1 of 17

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Global Response Dependency and Worldmaking * Kantian ideas. The first of these is that the knowable “phenomenal” world of “appearances”—stones, trees, cats, and the like—is partly constituted by the cognitive activities of the mind. The second idea distinguishes objects as we know them from objects as they are independent of our knowledge. The latter objects make up “the noumenal world” of “things‐in‐themselves”. Only the noumenal world, forever inaccessible and beyond our ken, has the objectivity and independence required by Realism. The phenomenal world of familiar objects does not, as it is partly our construction. Constructivism then adds a third idea to this Kantian pair: relativism. But, suppose that we drop (p.122) the commitment to relativism. We are then left with the following doctrine which I will name “Worldmaking”: Worldmaking: The only independent reality is beyond the reach of our knowledge and language. A known world is partly constructed by our imposition of concepts. The commitment to relativism makes Constructionism worse than Worldmaking but my Moorean and naturalistic arguments in Chapter 5 count just as much against Worldmaking. It is another bizarre antirealism.

My aim in this chapter is to consider the relation between Worldmaking and the response dependency theory of concepts. Philip Pettit (1990) has proposed a global version of that theory: all our concepts are response‐dependent. I think, although Pettit does not, that his global theory entails Worldmaking and for that reason alone should be rejected. The global theory is another example of the cost of not putting metaphysics first. For, it yields an argument from a semantic premise to anti‐Realism.

2. The Response‐Dependent Argument The argument is one that we can construct from contemporary discussions of response‐dependent concepts. It starts from a theory of our concepts of secondary properties (“qualities”). Let us follow custom and take our concept REDNESS as the example. A popular and plausible Lockean theory is that REDNESS is the concept of a disposition to produce a certain sort of response in normal humans in normal conditions. Mark Johnston calls such concepts “response‐dependent” (1989: 144–5) and that usage has now been widely adopted.1 The argument for Worldmaking then runs as follows: (1) Global Response Dependency of Concepts: All concepts, not just those of secondary properties, are response‐dependent. So, (2) Global Response Dependency of Properties: All properties are response‐dependent: they are dispositions to produce certain sorts of responses in normal humans in normal conditions. So, (3) Worldmaking.

Page 2 of 17

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Global Response Dependency and Worldmaking * (p.123) A theory of the nature of concepts is a semantic theory. So this argument derives an antirealist metaphysics from semantic speculations, just the sort of argument criticized in Chapter 5. I say that we can “construct” this argument from contemporary discussions because, so far as I know, it has not been explicitly endorsed by anyone. Hilary Putnam proposed (2): everything we say about an object is of the form: it is such as to affect us in such‐and‐such a way. Nothing at all we say about any object describes the object as it is ‘in itself’, independently of its effect on us. (1981: 61) Putnam does not seem to infer (2) from (1) or from any theory of concepts. He proposes (2) as a way of understanding Kant and clearly sees it as the basis for (3). I have argued that it is indeed a basis (1991b: 13.7). I shall summarize the argument in section 3.

Philip Pettit (1990) proposes (1) as a solution to the problem of rule‐following. He sums up his proposal: if we are to resolve Kripke's version of the Wittgenstein problem of rule‐ following, then we must acknowledge a global form of response‐ dependence (1991: 588).2 Perhaps Pettit would accept the inference from (2) to (3).3 However, surprisingly, he explicitly rejects the inference from (1) to (2) (1991: 609), the inference from the response dependence of concepts to the response dependence of the properties that they are concepts of.4 In section 4, I shall argue that he is wrong to reject this and hence wrong to claim that “response‐dependence does not compromise realism in a serious manner” (1991: 558). Global Response Dependency of Concepts leads to Worldmaking and so compromises realism in the most serious way possible.

If I am right about this, the Moorean and naturalistic arguments in Chapter 5 supply the very best of reasons for rejecting Global Response Dependency of Concepts. We should be sufficiently confident of our Realist metaphysics to reject any semantic theory that fails to fit it. In particular, we should reject a theory of concepts that leads to Worldmaking. At this point of time, we do not know anywhere near enough about semantics to challenge Realism. If this leaves us without a solution to Kripke's skeptical problem about rule‐following (p.124) and meaning, so be it. Better to give up, accepting the skeptical position about rule‐following and meaning, than to entertain a solution that involves Worldmaking. Better still to keep looking for a solution compatible with Realism.5 A point of clarification. The argument talks of response‐dependent “properties”. This should be seen as a convenience and not as a commitment to realism about properties and rejection of nominalism. The argument can be rephrased in a way that has no commitment to the existence of properties and is quite neutral on the realism‐nominalism issue. Thus the response‐dependence theory of the Page 3 of 17

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Global Response Dependency and Worldmaking * property redness can be taken as a theory of the nature of red objects: it is the theory that something is a red object in virtue of being disposed to produce a certain sort of response in normal humans in normal conditions. And Global Response Dependency of Properties generalizes this story: all instances of the schema something is an F object (or an F) in virtue of being disposed to produce a certain sort of response in normal humans in normal conditions are true.6

In the discussion of response dependence I shall sometimes talk of theories of F‐ness because it is convenient to do so, but it should be remembered that what are really in question are theories of what it is to be an F object.

3. From Global Response Dependency of Properties to Worldmaking My aim in this section is to show how Global Response Dependency of Properties leads to Worldmaking, to show how (3) follows from (2). I shall start by considering the consequences of a response‐dependent view restricted to secondary properties. We see first how objects classified under secondary properties lack the thoroughgoing independence of the paradigm realist objects. Then we see how they are, nonetheless, different from the objects posited by Worldmakers. Finally we see how this difference is lost when response dependency goes global. (i) What are the paradigm realist objects? My Realism is committed to tokens of “physical types”. What tokens are they? I said that they are “stones, (p.125) trees, cats, and the like”. This is vague, of course. What is covered by “the like”? The intention is to cover objects the existence and nature of which are, near enough,7 totally independent of us. Stones, trees, and cats qualify. So do stars and dinosaurs. These are the paradigms. Artifacts and tools—hammers, paperweights, chairs, and the like—are not, for their natures have an interesting sort of dependence on us: something would not be a hammer if we had not designed it to hammer or commonly used it to hammer.8 Social entities— bachelors, money, votes, and the like—are not paradigms either: something would not be a vote if it did not stand in some complicated relation to us and our intentions (1991b: 13.5). In a sense we do make something a hammer or a vote, but in no sense do we make something a cat. This is not to say that one cannot be realist about the non‐paradigm entities. The point is rather that realism about them must accept a more qualified form of independence than realism about the paradigms: they are independent except in the ways just noted. And realism about them is rather pointless in the struggle with antirealism. Every object that is a non‐paradigm under one specification is a paradigm under another and so is covered by Realism anyway; thus every paperweight is also a stone, a shell, or whatever. It is important to realism that Page 4 of 17

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Global Response Dependency and Worldmaking * there be stones having natures that are not at all dependent on us. It is not important that these stones may have other natures—for example, as paperweights—that are only a little dependent on us. You do not get a realism worth fighting for by claiming merely that something exists objectively and independently of the mental. Indeed, Worldmakers and many other antirealists would accept that claim. The realist needs to be specific about the independent entities he is fighting for: they are stones, trees, cats, and the like (1984/1991b: 2.3; Ch. 2, part I, in the present volume). But it is enough to have an entity under one specification. Adding more specifications does not help in the struggle with the antirealist. Return to the secondary properties and red objects. The point to be made is that on the response‐dependence view, red objects are like the other non‐paradigms in that they have an interesting dependence on us: an object is red in virtue of its effects on us. And, once again, this dependence is not important in the struggle with antirealism because each red object is also a paradigm object. At least, it is also a paradigm object provided response dependency has not spread to stones, trees, cats, and the like. However, we shall see that (p.126) the dependence becomes very important when response dependency does go global. (ii) Assuming that response dependency is restricted to the secondary properties, how do red objects differ from the objects posited by Worldmakers? In two respects: the way they depend on us and, more importantly, the way they depend on the independent world. Concerning the first, according to the Worldmaker, we are assuming, if there had been no people there would have been no stars or dinosaurs. In contrast, according to our restricted response‐dependency view, if there had been no people there would still have been red things; there would still have been things that would cause a certain response in humans were humans around. There is a further difference if the response‐dependent view requires an “experiential” response rather than a “judgmental” one. According to the Worldmaker, things have their natures in virtue of the concepts we impose on them, in virtue of how we think about them. So there is not a further difference between Worldmaking and a response‐dependence view that claims that red things must cause humans to judge that they are red. However, there is a further difference with a view that claims—more plausibly, in my opinion—that red things must cause humans to have a certain sort of non‐cognitive experience— they must “sense redly”: whereas the nature of a Worldmaker object depends on how we think about it, that of a red object depends on how we experience it. So far as I can see, this difference is of little significance to the argument. For the sake of that argument, I shall take response dependence to require an experiential response. Page 5 of 17

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Global Response Dependency and Worldmaking * The most important difference between red objects on the response‐dependent view and the objects posited by Worldmakers is in the way that they depend on the independent world. On the response‐dependent view, there is no limit in principle to what we can say about how the world independent of us determines that red objects are red. We simply have to discover scientifically what it is that causes us to have the appropriate response. The contrast with Worldmaking is sharp. According to Worldmaking, the known world is determined by our thought which is constrained by a noumenal world in ways that are in principle unknowable. This is mysterious to the point of incoherence. Worldmaking might be modified as we noted Putnam and Kuhn seem to modify Constructivism, either denying any constraint or refusing to say anything about it (Ch. 5, sec. 5). But the modification does not improve Worldmaking. Why do we think that these things are trees and those things are cats, hence making them trees and cats? Why could we not make the known world in any way we liked? Neither version of Worldmaking can give (p.127) satisfactory answers. The independent world has a real role in making red things not its pseudo role in making the world of Worldmakers. So although, on the restricted response‐dependence view, red objects lack the thoroughgoing independence of the paradigm realist objects, they nonetheless differ sharply from the objects posited by Worldmakers.9 This difference is lost when we remove the restriction and go global. (iii) Global Response Dependency of Properties makes the nature of everything depend on our experiences: briefly, all instances of ‘x is F in virtue of causing us to experience F‐ly’ are true. So we can no longer tell the scientific story of an independent world's causal influence on our sensing redly, or whatever. We cannot explain that influence by appeal to the properties of objects that have natures that are independent of our experiences of their natures. There is no longer any basis for claiming that red things, or whatever, would have existed if there had been no people. The situation is worse. If we generalize response dependence to all properties then surely it should be generalized to all relations. So it should cover causality. So not only have we lost the independent world that influences our experiences, we have lost the independent influence (as Kant did; Ch. 5, sec. 5 of the present volume).10 The known world is all the result of our imposition. Global Response Dependency of Properties leads to Worldmaking. I have demonstrated earlier that red objects, though partly dependent for their redness on us, differ from the posits of Worldmaking. Why then does this difference not carry over when we generalize? Because that demonstration was against a Realist background, a world of stones, trees, cats, etc. that was quite independent of our experiences. The background supplied the crucial contrast between the way red objects and the entities of Worldmaking depended on an independent world. The way red objects depended on that world for their (p. 128) redness was a causal relation open to any amount of scientific Page 6 of 17

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Global Response Dependency and Worldmaking * investigation. So talk of this dependence yielded a genuine explanation of constraints on our thinking and hence on our worldmaking. The global view removes the Realist background. It leaves the dependence of red things, and everything else, on an independent world mysterious. Talk of this dependence yields only the empty facade of an explanation of constraints on our thinking and worldmaking. The contrast with the entities of Worldmaking is lost. The global view can, of course, talk of causal relations between objects. However, the relations and objects talked of are both created by our experiences. Set aside worries occasioned by our creating the relations and consider those occasioned by our creating the objects. Given that what cause our experiences are objects created by those very experiences what is left of the idea that our experiences are constrained by an independent world? The problem for the global view is with the causation of experiences by an independent world not causation in general. The natures of objects referred to in explanations of experience may themselves depend on experience without leading to Worldmaking, provided that, in the end, we get explanations that refer to natures that are not so dependent. Consider, for example, the following explanation: x made me think that it was like the sound of a trumpet because x was red. On the earlier response‐dependent view, x's being red is dependent on an experience. Despite this, we can use the fact that x is red to explain another experience without collapsing into Worldmaking because the experience involved in that fact can itself be explained by reference to an experience‐independent nature; we can explain it in terms of the reflective properties of x's surface, for example. On the global view however, this explanatory procedure never reaches an experience‐independent nature: the explanation of the experience that is constitutive of x's being red refers to another experience‐dependent nature; and so on. Unless we finally get to natures independent of our experiences, all these explanations of our experiences in terms of objects with certain natures collapse into explanations of one experience by another. For the nature of every object collapses into an experience. So Global Response Dependency of Properties does indeed lead to Worldmaking: (2) leads to (3). And for that very reason (2) should be rejected.11 The earlier response‐dependent view of secondary properties is plausible enough,12 but its (p.129) plausibility rests heavily on its implicit commitment to the independent nature of the world that is responsible for the relevant effects on us. There is no puzzle about typing objects by their effects on us provided that we can appeal to typings that are not dependent on us to explain those effects. A little bit of worldmaking is all right against a background of a world that we did not make and that influences our little effort. Generalizing response‐dependence removes what is essential to the plausibility of the account.

Page 7 of 17

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Global Response Dependency and Worldmaking * 4. From Global Response Dependency of Concepts to Global Response Dependency of Properties A striking feature of the contemporary debate about secondary properties is its focus on concepts. Thus, even though the concern is the traditional one about the properties themselves, most of the discussion is about our concepts of those properties.13 This is an example, of course, of the “linguistic turn” in philosophy: the view that insofar as a traditional metaphysical problem is philosophical rather than scientific, it should be approached by a priori “conceptual analysis”.14 So if analysis shows that our concept of a secondary property is response‐dependent, we expect to conclude that the property itself is response‐ dependent. At first glance, Pettit's discussion is a generalization of this sort of approach to secondary properties. So we expect him to move from his view that all our concepts are response‐dependent to the conclusion that all properties are too; we expect him to move from (1) to (2). Yet he rejects this move. I shall try to see why. (a) Why do we expect to move from the response dependence of, say, the concept REDNESS15 to the response dependence of the property redness? First, the idea of a response‐dependent property has been familiar since Locke. Of course, it was not familiar under the term ‘response‐dependent’, but once we have the term, extending it to cover a concept because the concept refers to a response‐ dependent property seems natural. So it is natural to think REDNESS is response‐dependent because it is about redness and redness is response‐ dependent. Second, this natural thought is encouraged by Johnston (1989), who introduced the term ‘response‐dependent’. Pettit rightly describes (p.130) Johnston's view as follows: “the properties to which [secondary property concepts] direct us are dispositions which are manifested in certain familiar responses” (1991: 597). In other words, for Johnston, the properties themselves are response‐dependent. Because of this view, Johnston believes a biconditional along the following lines: x is red iff and only if x is such as to look red to normal observers in normal circumstances (1989: 145; cf. Pettit 1991: 599). Johnston also believes that this biconditional is known a priori. This belief is important to him and to Pettit. From my naturalistic perspective (Chs. 12 and 13) it must be rejected, of course, but, so far as I can see, this disagreement is irrelevant to the present concern about realism. (b) A generalization of Johnston's view of the secondary properties yields Worldmaking, as we have seen (sec. 3). Pettit rejects Worldmaking and so his view should differ from Johnston's. How? Pettit focuses on the “response‐privileging” nature of secondary property concepts: “it is a priori that the responses which correspond to them leave no room for ignorance and error, at least under the appropriate conditions” (1991: 597). However, in comparing his view with Johnston's, he sees this focus as only Page 8 of 17

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Global Response Dependency and Worldmaking * “a difference of interest, not a difference of doctrine” (1991: 599); his sense of ‘response‐dependent’ is “something close to Johnston's sense” (1998b: 113). And he seems to hold the response‐dependent view of the nature of redness: “something is red because it looks red to normal observers: the capacity to look red to such observers is what marks off red things” (1991: 615; see also 1998a: 55–6; 1998b: 114). So, at first sight, he seems to hold just the view that leads to the disaster of Worldmaking when generalized. Despite this, Pettit insists that his theory does not involve any view of the properties in question: it is “not an assertion about the property or object or operation in question” (1993: 202–3). A fortiori, he is not committed to a response‐dependent view of these properties. Where Johnston's response‐ dependent concept “represents [its referent] as having a property that involves human beings”, his does not have “such anthropocentric content”(1998b: 114). Pettit sees his theory of REDNESS as being “studiously neutral . . . on precisely which property should be identified with redness” (1998a: 65) The biconditional for redness can be explained even if redness is not response‐dependent (1998a: 56–9). His theory allows us “to think of the colors of things in a variety of ways” (1991: 613–14): if to be red is to be such as to look red in suitable circumstances we may take the redness of a thing to be the higher‐level state of having a lower‐ level state that produces the required effect on observers; we may take it as the lower‐level state operative in (p.131) the thing; or we may take it as the disjunction of the lower‐level states that do the job required. (1991: 614 n.) We may take the property to be a “role‐property” or a “realiser‐property” (1998b: 116).

This is very puzzling. It seems to follow from the claim that to be red is to be such as to look red that the nature of redness is to play a certain role, the role of looking red. So redness has to be a response‐dependent role‐property. Suppose that, as a matter of fact, everything red in the actual world has a lower‐level property, P‐ness‐or‐Q‐ness. On Pettit's account, we could then take that property as the realizer‐property for redness in the actual world. Fair enough. But surely we could not identify that property with redness because there could be something “in another possible world” that was neither P nor Q and yet would look red to us and so would be red. And there could be something that was P or Q that did not look red to us and so would notbe red. This argument rests on what I would like to think are points made familiar by the rise of functionalism in the philosophy of mind. Suppose that everything that is F in the actual world A has a certain causal property C, a “causal power”; for example, everything that is fragile would break if struck sharply; everything that is red would look red to normal humans in normal circumstances. Suppose that Page 9 of 17

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Global Response Dependency and Worldmaking * the explanation of their having C in A is that they are P, where being P is some intrinsic physical property, perhaps a disjunctive one. Here are two very distinct views we might have about being F: 1. To be F is to have C or, expressing this in property talk,

The property of being F is the property of having C; 2. To be F is to be P or, expressing this in property talk,

The property of being F is the property of being P. Clearly view 1 identifies being F with a role‐property, and view 2, with a realizer‐ property. Many (including me) think that view 1 is right for fragility and redness, but many (including me) think that view 2 is right for many other properties; for example, being gold is an intrinsic physical property, the property of having atomic number 79, whatever the causal powers of gold.

Now return to what Pettit says. The sentence, “to be red is to be such as to look red in suitable circumstances” seems like a clear statement of view 1 (p.132) about redness. And since the causal role that is thereby identified with redness involves human responses, this is a clear statement of a response‐dependent view of redness. Now redness cannot be both this view 1 response‐dependent property and the view 2 lower‐level property that, as a matter of fact, causes objects to look red in suitable circumstances. So, given what Pettit says, redness is not that lower‐level property. He cannot remain neutral about whether redness is a role‐property or a realizer‐property: it is a role‐property. So, despite Pettit's claims to neutrality, some of his remarks about redness seem to commit him to a response‐dependent view of it. Why does Pettit think otherwise? I am sorry to say that I don't know. Perhaps the theory of the reference of REDNESS plays a role in the answer. If so, it shouldn't. The claim that to be red is to be such as to look red is about the property redness not the concept REDNESS. Unless the theory of reference yields something about redness that contradicts this claim, it is hard to see how it can be relevant: the claim, with its commitment to a response‐dependent property, will still stand. In any case, I shall argue in a moment that Pettit's theory of reference supports the claim. Aside from this, it is a mistake for Pettit to be aiming for a theory that is neutral on the nature of redness. For, he goes on to generalize this theory, and to be neutral about the generalized theory is to be neutral about Worldmaking. That is not something that anyone should be neutral about. Despite his claims to neutrality, Pettit favors the view that redness is a realizer‐property (1998a: 65).

Page 10 of 17

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Global Response Dependency and Worldmaking * My present point is that when he goes global, he must insist on some properties being realizer‐properties on pain of Worldmaking. Let us take stock. In point (a) I have argued that, given the history of response‐ dependency doctrines, we should expect (1) to be accompanied by (2). In point (b), I have argued that, given what Pettit says about the property of redness, he cannot be neutral, as he claims to be, about the nature of that property: he is committed to it being response‐dependent just as he thinks the concept REDNESS is. Generalizing this view of redness to all properties amounts to (2). I shall now argue that (2) follows from (1) anyway. So Pettit is doubly committed to (2) and hence to Worldmaking, once by what he says about redness and once by his commitment to (1). (c) In insisting that his assertion about the concept REDNESS is not about the property redness, Pettit claims that the assertion says “that the reference of the concept is determined in such a way that our responses are privileged under certain conditions”; it is “a point in the theory of reference” (1991: 609). And his neutrality about the nature of redness is revealed in his being prepared to “take the predicate ‘red’ to direct us to the realizer‐property that makes things look red, not to the role‐property” (1998b: 122). (p.133) The difficulty in achieving this metaphysical neutrality arises from what we might call a “principle of reference”: (P) Necessarily (x)(K‐NESS applies to x iff x is K). The theory of reference for any concept must comply with (P).

1. Pettit is concerned with “basic” concepts but it helps to start with the problem for a “defined” concept, a concept covered by a description theory of reference. Take REDNESS as our example and suppose that it were covered by the following description theory: Necessarily (x)(REDNESS applies to x iff LOOKS RED TO NORMAL HUMANS IN NORMAL CONDITIONS applies to x). (Compare: Necessarily (x)(BACHELOR applies to x iff ADULT UNMARRIED MALE applies to x). Applying (P) to each concept in turn yields:

Necessarily (x)(REDNESS applies to x iff x is red); Necessarily (x)(LOOKS RED TO NORMAL HUMANS IN NORMAL CONDITIONS applies to x iff x looks red to normal humans in normal conditions). From these we get:

Necessarily (x)(x is red iff x looks red to normal humans in normal conditions).

Page 11 of 17

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Global Response Dependency and Worldmaking * How could this conclusion hold? An obvious answer is that the very nature of redness is to look red to normal humans in normal conditions: that is its essence, that is what it is to be red. But that answer is, of course, the response‐dependent view of redness. If the theory of REDNESS is really to be metaphysically neutral, there has to be another possible answer: we need a response‐independent account of the nature of redness that still yields the necessity of our conclusion. How can we get that? Suppose that we were to identify redness with the earlier‐mentioned “realizer‐property” P‐ness‐or‐Q‐ness. How could we explain a necessary connection between having this property and the property of looking red? We have no reason to expect a lawlike connection between the two properties: in another possible world with the same laws of nature, there might be red things that are not P‐or‐Q and vice versa. And even if there was a law linking the properties, that would not suffice. The necessity we need is not mere “natural” necessity, it is “metaphysical”, a necessity of the strongest sort arising out of the very nature of things. In another possible world where (p.134) the laws of nature were different, P‐ness‐or‐Q‐ness might not be linked to looking red.16 Similarly, the view that the reference of WATER is determined by a description theory associating it with LOOKS WATERY TO NORMAL HUMANS IN NORMAL CONDITIONS is not metaphysically neutral for it cannot be combined with the view that the nature of water is H2O.

2. Turn now to Pettit's view: the response‐dependent concept REDNESS is basic and undefined. Since REDNESS is basic its theory of reference must connect it in some direct causal way to the objects it applies to: it is not connected via its association with other concepts. Since REDNESS is response‐dependent this causal link must be via certain responses we have to the objects. So we get a theory along the following lines, reminiscent of indicator theories (see e.g. Dretske 1981): (p.135) Necessarily (x)(REDNESS applies to x iff for normal humans in normal conditions x looks red and thus causes REDNESS tokens).17 Applying (P) yields:

Necessarily (x)(x is red iff for normal humans in normal conditions x looks red and thus causes REDNESS tokens). Simplifying, we get:

Necessarily (x)(x is red iff x looks red to normal humans in normal conditions). This is the same necessity we arrived at earlier when we took REDNESS to be a defined concept. So, once again, a response‐dependent view of the concept is not metaphysically neutral as Pettit claims: it is committed to a response‐dependent view of redness.

Page 12 of 17

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Global Response Dependency and Worldmaking * It is important to note that Pettit has this commitment simply because he holds that REDNESS is response‐dependent. Thus, suppose that we were to drop that and hence drop the idea that the particular responses we have to red things determine the reference of REDNESS. Then we could adopt a straightforward indicator theory: Necessarily (x)(REDNESS applies to x iff x causes REDNESS tokens in normal humans in normal conditions). With (P) this yields:

Necessarily (x)(x is red iff x causes REDNESS tokens in normal humans in normal conditions). This really is metaphysically neutral: it is quite compatible with redness being the response‐independent property P‐ness‐or‐Q‐ness. It is no threat to the independence of redness that objects that have it are necessarily causally related to whatever applies to them. Even if red objects cause REDNESS tokens via the response of looking red, that fact alone would not involve that response in the nature of redness. For, so far as this theory of reference is concerned, it might be simply a contingent matter of fact that this response does the referential job: in another possible world another response might do so.18

(p.136) I conclude that Pettit's Global Response Dependency of Concepts commits him to Global Response Dependency of Properties and hence to Worldmaking: (1) does indeed imply (2). I suspect that the core of my disagreement with Pettit comes out in my response to the following passage: Even if redness is identified with a mentally independent, perfectly objective property, there remains a question, in David Lewis's phrasing, as to why this property and not some other gets to deserve the name of ‘red’. Any story that sustains the a priori biconditional for redness . . . can be seen as providing an answer to that question. (1998a: 61) Two comments follow from my discussion. First, if redness is identified with a certain property then the remaining question is immediately answered by (P). For according to (P), something deserves the name ‘red’ iff it is redness. So, it follows that the property identified as redness deserves the name. Second, if that property is indeed a response‐ independent one then, I have argued, the necessary biconditional becomes unsustainable. (I say no more on the allegedly a priori nature of the biconditional.)19

In conclusion, Pettit proposes Global Response Dependency of Concepts as a solution to the problem of rule‐following. Contrary to what Pettit claims, this doctrine leads to Worldmaking. So it is an example of a semantic doctrine of the sort that Moore and naturalism give us the best of reasons to reject. It is possible that someday we could be sufficiently confident of a semantic doctrine to overthrow Realism but that day is surely very distant.20

Page 13 of 17

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Global Response Dependency and Worldmaking * Notes:

(*) This chapter draws heavily on sections II.3 to II.5 of “Worldmaking Made Hard: Rejecting Global Response Dependency” (Devitt 2004) and appears with the kind permission from Kruzak. It also incorporates some improvements from my response (2006a) to Jacob Busch's criticism (2006a) of that paper. (1) The idea of response‐dependency is related to Crispin Wright's idea of “order‐of‐determination”. See Johnston 1993 and Wright 1993 for more discussion of these ideas. (2) Pettit's thesis is that all basic concepts are response‐dependent (1998b: 113). The basic ones are the ones not defined in terms of other concepts. Clearly the defined concepts will inherit the response dependency of the basic ones in terms of which they are defined. (3) But see n. 10 below. (4) So my earlier attribution of (2) to him (1991b: 257 n.) was too swift, at least; but see sec. 4 below. I made this attribution in 1990 on the basis of a draft of his 1991. I took the inference from (1) to (2) to be obvious and his draft, unlike the published version, did not explicitly reject it. (5) My own attempt at this is Devitt and Sterelny 1999: 9.5. (6) I am simplifying by ignoring the fact that the world is composed of stuff (like water) as well as objects. (7) The qualification is because of a boring exception: Realism does not deny certain familiar causal relations between its paradigm objects and us: we throw stones, plant trees, kick cats, and so on. For more on this, see Ch. 2, part I, of the present volume. (8) So the common practice of treating chairs as an example in the realism debate is inappropriate. (9) Rosen 1994 emphasizes that the response‐dependent view of colors does not threaten their objectivity. (10) In a note (1991: 608 n.) Pettit responds briefly to an earlier version of this criticism (a draft of my 1991b: ch. 13). There are two problems with the response. First, he overlooks my point about the lost independence of the world, responding only to the point about the lost independence of the causal influence. And his discussion in the text does not address the former point either: he emphasizes the independence of redness but does not note that this independence is lost when response‐dependence goes global (pp. 608–14). Second, his response seems to miss the point of the criticism about causal influence. The response‐dependent view of causality is along the following lines: Page 14 of 17

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Global Response Dependency and Worldmaking * xs cause ys in virtue of the fact that under conditions in which we wanted to bring about ys we would be disposed to manipulate xs as a means of bringing about ys (Johnston 1993: 105). The criticism is that once response‐dependence goes global we can no longer appeal to an experience‐independent world to explain why we would be disposed to manipulate xs as a means of bringing about ys and hence why x does cause y; we can no longer explain why we would be disposed to do this rather than anything else. I suspect that underlying Pettit's response is the view discussed in the nextsection. (11) Huw Price offers another reason for rejecting the view: if it is really global then it leads to a vicious regress because it has to apply also to the fact that x causes us to experience F‐ly, then to the fact that explains that, and so on (1998: 123). (12) But see Johnston's rejection of the response‐dependency view of colors because “response‐dependent properties cannot be sensed or perceived” (1998: 18). (13) See e.g. Boghossian and Velleman 1989, a critical examination of contemporary views of color. Wedgwood 1998 complains about the focus on concepts in discussions of response‐dependency. (14) In Ch. 13 of the present volume I reject the view that there is any a priori conceptual analysis. (15) Instead of talking of the concept REDNESS we might talk of the term ‘redness' (which expresses the concept); see e.g. Pettit 1998a. (16) What if the theory of reference for REDNESS had appealed to a “rigidified” concept, LOOKS‐RED‐IN‐THE‐ACTUAL‐WORLD . . . ? This would have yielded the conclusion: Necessarily(x)(x is red iff x looks‐red‐in‐the‐actual‐world . . . ). Given the identification of redness with P‐ness‐or‐Q‐ness, this would yield:

Necessarily(x)(x is P‐or‐Q iff x looks‐red‐in‐the‐actual‐world . . . ). How could this be? Certainly something that looks‐red‐in‐the‐actual‐world has the property of looking‐red‐in‐the‐actual‐world in any possible world. That is a trivial: the property of being K‐in‐the‐actual‐world is an essential property of anything that is actually K. But being P‐or‐Q could not be an essential property of anything. After all, it is being identified with redness and something that is red might not have been. So some things that are P‐or‐Q in the actual world are not in another possible world. Yet they still look‐red‐in‐the‐actual‐world. Even on the rigidified theory of reference, the proposed identification must be rejected and Pettit's metaphysical neutrality remains unattainable.

Page 15 of 17

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Global Response Dependency and Worldmaking * Aside from that, the trivial essentiality of the property of looking‐red‐in‐the‐ actual‐world raises the idea that the argument should be strengthened by replacing its talk of “necessarily” with talk of “in virtue of”. Consider the contrast between the following: Necessarily(x)(x is human iff x is a featherless‐biped‐in‐the‐actual‐ world; (x)(x is human in virtue of x being a featherless‐biped‐in‐the‐actual‐ world. The former is true because being human, unlike being P‐or‐Q, is an essential property. Yet the latter is false: being a featherless‐biped‐in‐the‐actual‐world does not have the required “explanatory” link to being human. Similarly, if being P‐or‐Q were an essential property,

Necessarily(x)(x is P‐or‐Q iff x looks‐red‐in‐the‐actual‐world . . . ) would be true, but

(x)(x is P‐or‐Q in virtue of x looking‐red‐in‐the‐actual‐world . . . ) would be false. And it seems that it would have to be true on the rigidified theory of reference because, starting with the following replacement of (P),

(P*) (x)(K‐ness applies to x in virtue of x being K), we would be able to derive it (given the proposed identification). Metaphysical neutrality seems even more distant.

(17) This is rough in various ways that do not matter to the argument. (18) It is, of course, obvious that red objects cause REDNESS tokens via some experiential response or other (how else?). But this alone does not make REDNESS response‐dependent (on pain of it alone making all concepts response‐ dependent). What does make it response‐dependent is that the particular response of looking red, a response with a certain qualitative character, determines its reference. And the latter is clearly Pettit's view (1991: 597, 604– 5, 620; 1998b: 113, 122). (19) Just before this passage, Pettit acknowledges the influence of Frank Jackson and refers to Jackson 1998. In what seems to be the most pertinent part of that book, Jackson “identifies colours with physical properties” and then claims that “colours per se are observer‐independent properties, but which observer‐ independent properties they are is not observer‐independent”: which they are turns on whether the properties “have the right kinds of causal effects in the right kinds of ways on normal observers in normal circumstances” (p. 100). To explain he offers an analogy of colors with plutonium. Despite this analogy, and several emails from Jackson, I don't get it. Redness is a color. So, on Jackson's view it is some observer‐independent physical property like my P‐ness‐or‐Q‐ness. What, then, could it possibly mean to say that it is not observer‐independent which observer‐independent property redness—i.e. P‐ness‐or‐Q‐ness—is? We can Page 16 of 17

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Global Response Dependency and Worldmaking * all see that the following is observer‐dependent: things that have P‐ness‐or‐Q‐ ness look red to humans (analogy: plutonium is the most dangerous chemical structure for humans). But how could the (alleged) identity, P‐ness‐or‐Q‐ness is redness, be observer‐dependent? How could this turn on which property causes things to look red? (20) A draft of my 2004 on which this chapter is based was delivered at the conference, “La Structure du Monde”, held in Grenoble in Dec. 1999. Paul Boghossian was my commentator. I am indebted to him and to other members of the conference for several improvements. I am also indebted to Philip Pettit for several attempts over the years to talk me out of my thesis about global response‐dependency; and to Frank Jackson for an attempt to explain his position.

Access brought to you by:

Page 17 of 17

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Nonfactualism *

Putting Metaphysics First: Essays on Metaphysics and Epistemology Michael Devitt

Print publication date: 2009 Print ISBN-13: 9780199280803 Published to Oxford Scholarship Online: February 2010 DOI: 10.1093/acprof:oso/9780199280803.001.0001

The Metaphysics of Nonfactualism * Michael Devitt (Contributor Webpage)

DOI:10.1093/acprof:oso/9780199280803.003.0008

Abstract and Keywords The usual characterizations of nonfactualism are unsatisfactory. The problem partly comes from focusing on nonfactualism's special semantics instead of on the antirealist metaphysics that must motivate that semantics. The problem also comes from the genuine difficulty in characterizing this metaphysics. This chapter rejects the usual implicit characterizations: that there are no properties or facts in the problematic area. Using the examples of instrumentalism, noncognitivism, and deflationary truth, it argues for the characterization that, in the problematic area, there is no reality with a nature to be explained and with a causal-explanatory role. Finally, it rejects accounts of the special semantics of nonfactualism in terms of properties, facts, and truth conditions, but accepts ones that contrast the apparently descriptive or factual function of indicative sentences in the problematic area with their alleged function as expressive, prescriptive, or whatever. Keywords:   nonfactualism, antirealism, properties, facts, instrumentalism, noncognitivism, deflationary truth, truth conditions, expressive, prescriptive

1. Usual Characterizations of Nonfactualism I am concerned with a doctrine that is often called “nonfactualism” and that encompasses “noncognitivism”, “emotivism”, “projectivism”, and Simon Blackburn's “quasi‐realism” (1984, 1993a, b). Nonfactualism in an area is obviously a sort of antirealism or eliminativism about that area. But what sort exactly? The usual answers in the literature are along the following lines. Nonfactualism about some area of language—for example, moral language, causal language, or the theoretical language of science—is the view that the Page 1 of 19

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Nonfactualism * predicates in that area do not denote, correspond to, etc., properties.1 Or it is the view that the indicative sentences in that area are not assertions or statements,2 are not factual or descriptive,3 are not truth‐conditional,4 and do not correspond to facts.5 Rather, those sentences have other functions like expressing attitudes or emotions, or prescribing norms or rules.6 These answers make it seem as if nonfactualism is, primarily at least, a semantic doctrine, a doctrine about what sentences mean and predicates refer (p.138) to.7 Yet implicit in the answers is a certain metaphysical doctrine, a doctrine about the way the world is or is not. The answers suggest that, metaphysically, nonfactualism about some area is the view that there are no properties8 or facts9 appropriate to that area; for example, there are no moral properties or facts. Of course, it is not surprising that nonfactualism should have an implicit metaphysics. Intuitively, the central underlying idea of nonfactualism in an area is that the putative reality in that area is problematic or defective. This idea supplies the motivation for giving nonfactualism's special treatment to the predicates and sentences of the language in the area. Thus, it is because some philosophers think that there are no moral properties that they deny that moral predicates refer. Yet, given the impact of the “linguistic turn” in philosophy, it is also not surprising that language, not the implicit metaphysics, is the focus of discussion.10 I think that the failure to focus on the metaphysical doctrine is a mistake.11 For, that doctrine is distinct from the semantic one and, as just indicated, importantly prior to it. Indeed, concerning realism issues in general, I have argued that we should distinguish metaphysical from semantic doctrines. (I shall return to this idea in section 3.) Furthermore, we should always “put metaphysics first” by establishing a metaphysical base with near enough no appeal to semantics and by arguing from that base for a semantics (1984/1991b; 1996, particularly section 4.12; Chs. 2, 3, and 5 in the present volume).12 (p.139) Finally, in the case of nonfactualism, I shall argue, the failure to focus on the metaphysics has left the doctrine unclear. And the unclarity is not only in the metaphysics: it affects the semantics as well. It is my aim to remove the unclarity, so far as that is possible. I shall start with the metaphysics, leaving conclusions about the semantics until the end.

2. The Failure of the Metaphysical Characterizations Immediately we do focus on the metaphysics of nonfactualism, we see that the implicit characterizations of this in the literature are unsatisfactory.13 The problem is that the characterizations overlook the extent to which a philosopher's attitude to the metaphysics characterized might reflect a position on the general issues of realism about properties and facts rather than on the particular problematic area of reality that is the concern of the nonfactualist; for example, rather than a position on morality. Thus, consider a nominalist. She will Page 2 of 19

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Nonfactualism * agree that there are no moral properties because she thinks that there are no properties at all! Yet, manifestly, this alone does not commit her to nonfactualism; to thinking that there is something especially defective about moral reality, something that motivates a special nonfactual semantics for moral language. She might be as realist as could be about morality. Or, consider someone like David Armstrong (1978b) who is a selective realist about properties. Armstrong thinks that empty predicates, disjunctive predicates, and negative predicates have no corresponding properties. He thinks that some predicates apply to the world in virtue of many properties. Most importantly, he looks to science to tell us which properties there are. Such a person might well be a reductive realist about morality, thinking that a moral predicate may apply to an object in virtue of many properties none of which are moral properties; perhaps they are social and psychological properties. So he also agrees with the metaphysics implicitly attributed to nonfactualism and yet his metaphysics of morality is quite contrary to the antirealist one that we are attempting to characterize. Finally, consider the unselective realist who thinks that there is a property for each predicate. A nonfactualist might accept, as indeed Blackburn does (1993a: 206), that moral terms are predicates. If such a nonfactualist (p. 140) is an unselective realist she will think that there are moral properties, thus disagreeing with the implicit characterization of moral nonfactualism. And even if the nonfactualist denies that moral terms are predicates and hence that there are moral properties, the implicit characterization of her nonfactualism is problematic: it “runs the wrong way”. It finds a defect in moral reality because of something special about moral language where we need to find a defect in moral reality to motivate the view that moral language is special. The general issue of realism about properties is independent of the issue of nonfactualism. Similarly, the general issue of realism about facts. If nonfactualism has a coherent metaphysics it should be possible for someone to embrace it or reject it whatever her position on these general issues. There should be a way of stating that metaphysics that is appropriate whatever the truth of the matter about the reality of properties and facts. The problem spills over into the characterization of another sort of antirealism, usually called the “error” doctrine. The metaphysical contrast between the two sorts is often brought out by saying that whereas nonfactualism about morality claims that there are no moral properties, the error doctrine claims that there are moral properties but they are not instantiated.14 This characterization of the error doctrine is unsuitable for anyone but an unselective realist about properties. It is easy enough to remedy the situation for the error doctrine because that doctrine does not propose a special semantics. Let ‘F’ be any predicate in the problematic area. The metaphysics of the error doctrine is: there are no Fs (or F things); for example, there are no good people, right actions, and so on. And this Page 3 of 19

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Nonfactualism * brings out nicely an important contrast with nonfactualism. For, the nonfactualist is happy to say that there are some Fs (or F things). Indeed, it is both a mark and an advantage of nonfactualism—as Blackburn (1993a) emphasizes in his ingenious defense of quasi‐realism—that nonfactualism goes along with realism in this respect. Whereas an error doctrine claims that all moral utterances are false and so the practice of making them is mistaken, the nonfactualist is likely to think that many of these utterances express appropriate emotions or prescriptions and so the practice of making them is fine. It is precisely this mark of nonfactualism that makes the problem of characterizing its metaphysics seem puzzling. The nonfactualist talks like a realist while giving that talk a special interpretation. How then can we describe nonfactualism in a way that distinguishes it from realism? (p.141) A feeling of vertigo may set in at this point. We are attempting a characterization of the metaphysics that must motivate the special semantic treatment that nonfactualism gives to a certain area of language. Yet our attempts seem doomed to vitiation by that very semantic treatment. Attempts in the “ordinary” language of metaphysical commitment will fail because that language is interpreted so that it has no such commitment. So attempts are made in a “philosophical” language that talks of properties and facts, apparently on the assumption that this language is spared the nonfactualist interpretation. But we have seen that these characterizations are unsatisfactory. (And we may wonder why the philosophical language is spared the interpretation.) Nonfactualism is supposed to be a sort of antirealism and yet it seems impossible to state its antirealism. Realism issues begin to evaporate. Indeed, Blackburn sometimes comes very close to claiming that they have evaporated (1993a: 4, 15–34, 55–9; 1993b: 368).

3. Can We Really Do Metaphysics without Doing Semantics? Before attempting to cure this vertigo, I pause to address a more general problem suggested by the discussion. The discussion may seem to threaten the whole enterprise of seeking nonsemantic metaphysical characterizations of antirealist doctrines. For it casts doubt on my earlier assumption, argued for elsewhere, that we can sharply distinguish metaphysical issues about realism from semantic issues. The discussion suggests rather that metaphysical claims must be accompanied by a semantic theory about their interpretation. This suggestion is a common one and is an important reason why many philosophers insist on characterizing realism in semantic terms. Thus, objecting to my nonsemantic and objectual characterization of realism about commonsense and scientific physical entities (1991b), Michael Williams claims that any such attempt to identify realism with commitment to a certain body of truths, rather than a view about truth, is bound to misfire. For we have to add the proviso that these truths be accepted at “face value” and Page 4 of 19

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Nonfactualism * explaining how and why this is so will inevitably reinvolve us with questions about what the truth of propositions of common sense and science should be understood to consist in. (1993: 212 n.) Here is one way to develop this objection. Merely stating such truths as ‘There exist stones', ‘There exist trees', ‘There exist cats', and so on, does not ontologically commit you to stones, trees, and cats, and so on. That (p.142) commitment depends on accepting the statements at face value so that they have certain truth conditions; for, on that interpretation, those entities must exist for the statements to be true. The ontological question becomes clear only when we move into the metalanguage and consider this semantic question.15 The disagreement between the realist and the antirealist is not over statements like the one above but over how such statements are to be understood. So the disagreement is a semantic one.

It is indeed right that if object‐language sentences like ‘There exists cats' are to yield a commitment then they must, in some sense, be accepted at face value: we must simply rely on our ordinary understanding of them.16 But it is wrong that this motivates any move to a metalinguistic semantic statement of truth conditions in order to establish a commitment; for example, a move to ‘ ‘Cats exist’ is true if and only if there exists things that ‘cat’ applies to’. For, the very same problem arises for the semantic statement: if that statement is to establish commitment, we must rely on our ordinary understanding of it.17 I don't claim that there is no problem in establishing a commitment in the object‐language, but simply that any problem that there is arises as much in the metalanguage. If any language is to establish a commitment to anything, we have to rely on our ordinary understanding of some language. But it would be preposterous to claim that the language we rely on must be semantic; that commitment might come from ‘There exist things that ‘cat’ applies to’ but could not come from ‘There exist cats'. Language does not suddenly become kosher when you start doing semantics. The idea that talk about the world is unclear and in special need of interpretation, yet talk about language and its relation to the world is straightforward on the face of it, reflects the damage of years of living under the linguistic turn. It is a truism that a theory must be presented to us in language. So to draw any conclusions at all from the theory, whether about ontological commitment (p. 143) or the price of eggs, we have to understand the language in which it is presented. But this mundane fact supplies no reason for supposing that we must move to a semantic theory to determine the ontological commitment of our object theory, because the fact covers the semantic theory too: even semantics requires language. A semantic theory of a sentence could clearly help us to understand that sentence but the theory is not necessary for the understanding (else we would understand very little). And, equally clearly, even when the theory does help, it does not make the issue that concerns the sentence semantic.18

Page 5 of 19

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Nonfactualism * I conclude that the enterprise of seeking nonsemantic metaphysical characterizations of antirealist doctrines is not threatened; we do not have to retreat into semantics to do metaphysics. I return now to our problems with nonfactualism in particular.

4. Rejecting Global Nonfactualism Consider, first, the idea—perhaps endorsed by nobody—that realism issues evaporate entirely because of the possibility of “global” nonfactualism, the possibility that all of our language that seems to have realist commitments—in effect, all apparently factual or descriptive language—does not really do so. That is surely a possibility we need not take seriously. There certainly could be a language that was entirely nonfactual, making no claims about how the world is; for example, consider a segment of English that includes only certain commands. A realist doctrine could not be stated in such a language. Neither, of course, could any other doctrine. But we have the best reason in the world for thinking that English as a whole is not like this; that parts of it are factual. In this respect, it is noteworthy that the nonfactualist, like everyone else, seems to think that semantic claims are factual (and if they were not, nonfactualism itself would be nonfactual!). She also seems to think that the unsatisfactory characterization of her metaphysics in terms of properties and facts is factual. To dismiss global nonfactualism is not to claim that all apparently factual language must be taken as really being factual, thus making nonfactualism impossible. We can adapt Quine's favorite image from Neurath: rebuilding a boat whilst staying afloat on it. We can rebuild any part of the boat but in so doing we must take a stand on some other part. So we cannot rebuild it all at (p.144) once. Analogously, we can reject any apparently factual sentence as not really being so but in so doing we must take some other apparently factual sentences for granted. So we cannot reject all apparently factual sentences at once. There must always be some factual language we take as really being factual in order to stay afloat.19 To dismiss global nonfactualism is not to claim either that some parts of our language must be truth‐conditional. It is common to assume that the right semantics for factual language is truth‐conditional. Combine this assumption with the rejection of global nonfactualism and it obviously follows that some parts of our language are indeed truth‐conditional. I think that the assumption is right, but I have done nothing here to show that it is right. Non‐truth‐conditional semantics for factual language are possible. We shall return to this matter in section 9. In sum, we need not worry that realism issues will evaporate because all of our language might be nonfactual. Yet, taking it for granted that some of our language is indeed factual, we still have a worry. The worry is that realism issues will evaporate because the language of any issue might be entirely nonfactual. Page 6 of 19

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Nonfactualism * Perhaps language is so “compartmentalized” that the language of a realism issue in any given area can be interpreted nonfactually and there is no way to use the factual language “from elsewhere” to distinguish the realist from the nonfactualist in the given area. We cannot go nonfactualist everywhere at once but perhaps we can go entirely nonfactualist anywhere. I shall argue that we cannot. Roughly, language is not entirely compartmentalized because reality is not.

5. Characterizing the Metaphysics of Nonfactualism To avoid the evaporation of the realism issue in some area, and to characterize the metaphysics of nonfactualism in that area, we must first find some language that is not just apparently factual but is treated by the nonfactualist as really factual. We must then examine her statements using that language to find ones that disagree with realist statements about the area. We have a reason to be optimistic that we will find these disagreements. I have noted that the motivation for a nonfactualist semantics in some area comes from dissatisfaction with the putative reality in that area, a reality that the (p.145) realist embraces (sec. 1). A nonfactualist's statement of this dissatisfaction must be in the uncontroversially factual language if the dissatisfaction is really to play the motivational role. This statement should provide a genuine metaphysical disagreement with the realist. There is thus something fundamentally misguided about attempting to defend antirealism in an area by claiming that all the language in that area has a nonfactual semantics. Such a claim both undermines the realism issue and leaves the semantics unmotivated. If the claim were true there would be nothing to distinguish realism from antirealism in the area and hence there would be no reason for any special treatment of its apparently factual language. Two sorts of realist claim are the most promising candidates for denial by the nonfactualist. First, realists tend to offer some explanations of the nature of the problematic reality in language that the nonfactualist should agree is factual. For, the realist thinks that the problematic reality is constituted by, or supervenes on, a reality that should be unproblematic for the nonfactualist. Even though the nonfactualist claims to be able to accept many sentences that seem to describe the problematic reality, taking them as expressive, prescriptive, or whatever, she should not accept these explanations because, loosely, she does not accept that there is any such reality to be explained. Second, realists make claims in the uncontroversially factual language about the causal role of the problematic reality. For the realist thinks that the problematic reality is the cause or effect of some unproblematic reality.20 The nonfactualist should not accept these claims about the role of the problematic reality because, loosely again, on her view there is no such reality to play a role.21 Page 7 of 19

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Nonfactualism * Of course, this rejection of realist claims about the nature and role of the problematic reality may seem implausible. But that is the price that nonfactualism must pay for its motivation. It is interesting to compare these realist claims with others that have been the subject of much discussion in the nonfactualism debate.22 Take some simple sentence which, when affirmed on its own, is alleged to be nonfactual; for example, ‘Lying is wrong’. Now consider the role of this sentence in conditionals; for example, in ‘If lying is wrong, Alice should be punished’. How can the special nonfactualist semantics of the sentence when occurring (p.146) on its own be applied to the sentence when occurring in the conditional? The two occurrences must be given closely related meanings if the intuitive validity of certain inferences is to be captured; for example, the inference from these examples of a simple sentence and a conditional to ‘Alice should be punished’. The task of providing the required semantics may be impossible and is certainly difficult. I have suggested that the nonfactualist should reject realist claims about the nature and role of the problematic reality. Could she similarly reject these conditionals, and the arguments containing them, hence declining this difficult task? Both the friends and foes of nonfactualism clearly think not. They are surely right. However implausible it may be for nonfactualism to reject the claims about nature and role, it would surely be more so to reject the conditionals: it would make the nonfactualist view of the simple sentences very hard to accept. Furthermore, there is no pressing need to pay the price of this rejection, as there is the rejection of claims about nature and role, in order to motivate nonfactualism. For, important as these conditionals are in ordinary discourse, they are not central expressions of realist metaphysics. I have suggested nonsemantic ways to characterize metaphysical disagreements between realists and nonfactualists. Still, it must be acknowledged that the discovery of such a characterization in any particular case would be aided by some simple semantics: by a precise statement from the nonfactualist about the boundary between the language she takes to be factual and the language she takes to be nonfactual. But this semantics is only an aid to discovery. It is a preliminary to the characterization, not part of it. Furthermore, the semantic preliminary is not necessary: a discussion of the nature and role of the problematic reality will reveal the metaphysical disagreements. Where we lack both the semantic preliminary and the discussion, any attempt to characterize nonfactualism in an area, particularly by a realist, must be tentative. We can hope to indicate the general area of nonfactualist disagreement with the realist but cannot be confident about the exact place. And, given that nonfactualism deprives us of much of our language for describing reality, we should not be surprised that neat and simple Page 8 of 19

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Nonfactualism * characterizations of the disagreement are hard to come by. Although realism issues do not evaporate in the face of the possibility of nonfactualism, they do become much more difficult to characterize. What appeared to be relatively tidy issues (to me, at least; 1984/1991b; Ch. 2 in the present volume) become decidedly messy. I shall apply and develop these ideas for characterizing nonfactualism by considering three examples: instrumentalism, moral nonfactualism, and deflationary truth.

(p.147) 6. Instrumentalism I start with a doctrine that is no longer popular: the traditional scientific instrumentalism urged with such success by the positivists. According to this instrumentalism, a theory is a partially interpreted formal system. The vocabulary of the theory is divided into two parts, an “observational” part which is fully interpreted and a “theoretical” part which is uninterpreted. In our terminology, the observational part is factual whilst the theoretical part is a nonfactual “instrument” for generating factual claims. So far, all we have is semantics. What about metaphysics? The motivation for instrumentalism clearly comes from some general doubts about unobservable reality but the doubts are inexplicit (reflecting, of course, the positivist horror of all things metaphysical).23 It would be nice if we could take these doubts, straightforwardly, as being about whether the unobservables of science exist: the instrumentalist believes that they do not exist or that we cannot know that they exist. However, we will not find a simple expression of instrumentalist doubt about, for example, the statement ‘Atoms exist’. The problem is that, for the instrumentalist, ‘atom’ is part of the uninterpreted nonfactual vocabulary and the statement is an implicit part of a theory that she is as ready to endorse as the realist.24 To find the appropriate expressions of instrumentalist doubt we must look for realist statements that are “about unobservables” and yet are in the observational, hence factual, language. We can find these statements in realist views about the natures and roles of unobservables. Thus, consider what a realist influenced by the kinetic theory of gases might say “about molecules” without using ‘molecule’: that there are spherical, elastic, smooth entities constituting a gas; that their impact on the wall of a containing flask is responsible for the pressure exerted by the gas; that the temperatures of two gases are the same when the mean kinetic energy of those constituting entities of the two gases is the same; and so on. This language all seems observational and hence factual. Yet it posits entities that are unobservable. So the (p.148) instrumentalist should deny this realist claim or remain agnostic about it. There are many other such realist claims. So the

Page 9 of 19

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Nonfactualism * instrumentalist's antirealist metaphysics is characterized by her doubts about all these claims.25 Of course, this metaphysical position is not a comfortable one for the instrumentalist because the realist claims she is doubting are drawn from science. Yet she cannot accept the claims on pain of leaving her nonfactualist semantics unmotivated.26

7. Moral Nonfactualism Consider moral nonfactualism next. Speaking loosely and intuitively, the moral nonfactualist holds that the only reality underlying moral utterances is the realm of moral attitudes and/or emotions.27 The task is to specify precisely, in uncontroversially factual terms, the richer reality of the moral realist that is thus denied. We shall have no success trying to do this with existential statements. The nonfactualist thinks that she can join the realist in saying, “there are good people”, “there are right actions”, and so on. To find what the nonfactualist is most likely to deny, we must consider what the realist says about the natures and roles of good people, right actions, and so on. Realists claim that there are things about a person in virtue of which she is good, that make her good; for example, being kind, considerate, generous, honest, etc. Similarly, realists claim that there are things about an action in virtue of which it is wrong, that make it wrong; for example, leading to unhappiness, (p.149) being contrary to socially accepted rules, and so on. The language of these “in virtue of” clauses seems to be the sort that the nonfactualist will count as factual. (If not, she must help us find some other clauses along the same lines that she will count as factual). The nonfactualist must reject all such “in virtue of” claims as totally misconceived. Consider crude “Boo–Hooray” nonfactualism, for example. Suppose that Mark says, “Alice is good”. The nonfactualist, like anyone else, may explain in virtue of what Mark has the “hooray‐attitude” that he thus expresses: something about Mark, his disposition to behave in certain ways, a certain physiological state, or whatever, makes it the case that he has this attitude. But this is very different from explaining in virtue of what Alice is good (supposing that she is). The realist thinks that there is something about Alice that explains this: it is her kindness, generosity, disposition to behave in certain ways, disposition to cause hooray‐attitudes in others, or whatever, that make her good. The nonfactualist rejects any such explanation of Alice's goodness. Realists think that explanations may be given of how a person came to be good: because she had loving parents, such and such genes, and so on. Realists think that there are consequences of her being good: she is admired, is sought after as a friend, causes hooray‐attitudes in others, is taken advantage of, and so on. Realists think that it is because Hitler and his associates were depraved that we believe that they were depraved. And it is because they were depraved that they Page 10 of 19

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Nonfactualism * behaved as they did and that millions of people died in concentration camps. Realists think that there are consequences of an act being wrong: it causes boo‐ attitudes; the person committing it is condemned, avoided, and so on.28 The language in the causal clauses of these explanations seems to be factual.29 The nonfactualist must reject all such explanations. It is not the case that there are any causes or effects of things being good, depraved, wrong, and so on. The moral realist thinks that there is a moral reality which, like any other reality, has a nature and has relations to other realities; and that this nature and (p. 150) those relations need explanations.30 The nonfactualist reveals her antirealism by rejecting any such explanation. Even if she is right in thinking that she can join with the realist in accepting ordinary moral judgments, she cannot join with him in his explanation of the reality which he takes those judgments to describe.

8. Deflationary Truth Finally, I turn to deflationary truth. According to the deflationist, the function of ‘true’ is not to describe a sentence, at least not to describe it in the way that a normal predicate like ‘green’ describes an object. Rather, the function of ‘true’ is logical or expressive, a convenient device for making assertions about the largely nonlinguistic world. Thus, instead of repeating a person's sentence about a movie, I can make the same point about the movie by saying ‘That is true’. I can express general agreement with an article about the behavior of penguins by saying ‘Most of that article is true’. I can assert Goldbach's Conjecture even though I cannot remember it by saying ‘Goldbach's Conjecture is true’. ‘True’ is particularly convenient for the assertion of an infinite number of sentences. This is all about the semantics of ‘true’. What is the deflationist's metaphysics of truth? How does that metaphysics differ from the semantic realist's? (Here, the metaphysical issue between the nonfactualist and the realist is, confusingly, itself semantic.) They do not differ over whether there are true (false) sentences. They agree that there are. They disagree about the nature and role of truth. In virtue of what is a sentence, say ‘Schnee ist weiss', true? According to the realist the sentence is true because it is related in some way to the world. A substantial theory is then required to describe and explain this relationship. The theory might include causal theories of reference, claims about warranted assertability, or whatever. The deflationist rejects any such realist explanation of truth. (There is no controversy here about the factual nature of the language used in the explanation.) Furthermore, she has an interesting alternative along the following lines. Truth is basically “disquotational”. There is nothing (p.151) more to it than is captured by the infinite set of appropriate instances of the schema, s is true if and only if p.

Page 11 of 19

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Nonfactualism * An “appropriate” instance is one where what is substituted for ‘s’ names a “translation” of the sentence substituted for ‘p’. Given that ‘snow is white’ translates ‘Schnee ist weiss', ‘Schnee ist weiss' is true, according to the deflationist, simply in virtue of it being the case that snow is white. No deeper explanation is called for.31

Realists give truth important explanatory roles; for example, to explain the success of science or the success of people in meeting their goals. In my view, the most interesting realist role for truth is in a truth‐conditional explanation of meaning, where meaning itself plays a role in the explanation of behavior and in guiding us to reality (1996). The deflationist rejects all such explanatory roles for truth.32 In virtue of the logical role of ‘true’, if a sentence to which ‘true’ is applied plays an explanatory role, then so does ‘true’. But, for the deflationist, truth can have no explanatory role beyond this trivial one because, crudely, truth “isn't anything”. In this example, as in the previous two, I have claimed that the nonfactualist is distinguished from the realist by her denial of the realist's explanations of the nature and role of the problematic reality.33 But what about someone who claims to be a realist but does not see the need for any such explanations? For example, consider someone who simply insists that truth is “robust” or “substantive” but offers no explanation of it; and he thinks that truth is “epiphenomenal” not explanatory.34 It is hard to see how this position can be distinguished metaphysically from nonfactualism. And this surely adds to the implausibility of the position.

(p.152) 9. The Semantics of Nonfactualism I turn finally to the special semantics of nonfactualism. I began this paper with the usual characterizations of this semantics. Many of these are unsatisfactory. I have argued that the implicit characterization of the metaphysics of nonfactualism, talking of properties or facts, is unsatisfactory (sec. 2). It follows that characterizations of the special semantics of nonfactualism in those terms are also unsatisfactory. Thus, we cannot capture that semantics with the claim that predicates in the problematic area do not refer to properties; nor with the claim that the sentences in that area do not correspond to facts. A person might accept these claims because of general views about properties or facts, views that have nothing to do with nonfactualism about the problematic area. Another common characterization is also inappropriate: that the sentences in the problematic area are not truth‐conditional. This characterization is not general enough. It is suitable only for someone who believes that the right semantics for factual language is truth‐conditional. Many have this belief—and I am one of them—but it is inappropriate to presuppose it in characterizing nonfactualism. It should be possible for a verificationist who rejects truth‐ conditional semantics altogether to be a nonfactualist about, say, morality; indeed, Ayer, a famous moral nonfactualist, is presumably an actual example. Yet, if even factual sentences are not truth‐conditional, then the distinctive thing Page 12 of 19

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Nonfactualism * about moral sentences that makes them nonfactual cannot be that they are not truth‐conditional. And it should be possible for someone who is a deflationist, hence a nonfactualist, about truth to be a nonfactualist about something else like morality. Yet a deflationist is likely to think that all indicative sentences, including moral ones, are deflationarily truth‐conditional.35 The claim that the indicative sentences in the problematic area are not assertions or statements comes closer to a satisfactory characterization of nonfactualism. But the closest will usually be the claim that the problematic sentences are not factual or descriptive but rather expressive, prescriptive, or whatever. We take it for granted that many indicative sentences are factual (since global nonfactualism is not feasible). We take it for granted that many nonindicative sentences have other functions like expressing attitudes or emotions, or prescribing norms or rules. The nonfactualist is claiming that, (p. 153) despite appearances, the problematic sentences have a semantics like the latter not the former. If she is a truth‐conditionalist about the factual sentences, she will think that the problematic sentences have no truth conditions; if a verificationist, that they have no verification conditions. (Even this usually satisfactory characterization is unsuitable for deflationary truth, as Scott Soames has emphasized to me. Here we say that ‘true’ does not have the semantics of a normal descriptive predicate, perhaps not that of a predicate at all, but rather a certain logical or expressive role.36) This characterization of nonfactualism's special semantics makes vivid the need for an independent characterization of its metaphysics, a need that I have been at pains to emphasize. For, although this semantic characterization may seem a natural bedfellow for an antirealist metaphysics, it fairly obviously does not entail any such metaphysics; it tells us nothing at all about the nature of nonlinguistic reality.

10. Conclusion I have argued that the usual characterizations of nonfactualism are unsatisfactory. The problem partly comes from focusing on nonfactualism's special semantics instead of on the antirealist metaphysics that must motivate that semantics. The problem also comes from the genuine difficulty in characterizing this metaphysics because nonfactualism goes along with many realist utterances, claiming to be able to interpret them in a special way. I have rejected the usual implicit characterizations of the metaphysics: that there are no properties or facts in the problematic area. Using the examples of instrumentalism, moral nonfactualism, and deflationary truth, I have argued for a general method for characterizing this metaphysics: make precise the idea that, in the problematic area, there is no reality with a nature to be explained and with a causal‐explanatory role. There should always be some uncontroversially factual language in which to state this rejection of the problematic reality. Of course, the rejection may seem implausible, but that is Page 13 of 19

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Nonfactualism * the price that nonfactualism must pay for its motivation. Finally, I turn to the special semantics of nonfactualism, rejecting accounts of this in terms of properties, facts, and truth conditions, but accepting ones that contrast the apparently descriptive or factual function (p.154) of indicative sentences in the problematic area with their alleged function as expressive, prescriptive, or whatever.37 Notes:

(*) First published in James E. Tomberlin (ed.), Philosophical Perspectives, 10. Metaphysics (Devitt 1996b). Reprinted with kind permission from Wiley‐ Blackwell. (1) See e.g. Ayer 1952: 89; Sayre‐McCord 1988c: 7; Boghossian 1990a: 157–61, 166; 1990b: 266. (2) See e.g. Ayer 1952: 103; Haldane and Wright 1993b: 11; Hale 1993: 337. (3) See e.g. Ayer 1952: 107; Wright 1988: 29; Sayre‐McCord 1988c: 4; Blackburn 1993a: 3, 60; Haldane and Wright 1993b: 11–12. (4) See e.g. Ayer 1952: 103, 107; Sayre‐McCord 1988c: 5; Boghossian 1990a: 160–1, 164; 1990b: 266; Blackburn 1993a: 60; Hale 1993: 337, 340; Haldane and Wright 1993b: 11. (5) See e.g. Wright 1988: 29; Sayre‐McCord 1988b: pp. ix–x; Boghossian 1990a: 160; Haldane and Wright 1993b: 12. (6) See e.g. Ayer 1952: 103, 107; Sayre‐McCord 1988c: 4, 8; Boghossian 1990a: 160; Blackburn 1993a: 3, 60; 1993b: 365; Hale 1993: 337; Haldane and Wright 1993b: 11. Strictly speaking these accounts of nonfactualism need qualification because the sentences in question may be partly assertions, partly truth‐ conditional, and partly factual. We can ignore the qualification. (7) Caution is required in taking what appear to be semantic claims as really being so. The apparently semantic terms may be playing only a “disquotational” role (see sec. 8). So a claim that a predicate does not refer may be just a way of claiming that a property does not exist. (8) See e.g. Ayer 1952: 89; Boghossian 1990a: 157–9, 161–2; Blackburn 1993a: 3. (9) See particularly Wright 1988: 29–30; Sayre‐McCord 1988b: pp. ix–x, 4; Blackburn 1993a: 3, 52, 57; Hale 1993: 337; Railton 1993: 280. (10) [2009 addition] For more on why we must see nonfactualism as committed to an antirealist metaphysics, see Ch. 8, sec. 4, of the present volume.

Page 14 of 19

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Nonfactualism * (11) Some “no fact of the matter” doctrines do focus on metaphysics, of course; e.g. doctrines about absolute space‐time or inverted spectra. But these doctrines do not involve the semantic claims in my opening paragraph. Those semantic claims are an essential part of the nonfactualism I am concerned with. (12) In discussing realism about the external world, I capture these ideas in two maxims: Maxim 2: Distinguish the metaphysical (ontological) issue of realism from any semantic issue. (1984/1991b: 3) Maxim 3: Settle the realism issue before any epistemic or semantic issue. (p. 4) According to Maxim 2 no semantic doctrine about truth constitutes the metaphysical doctrine of realism. This is not to say that there are no connections between the two sorts of doctrine. I follow Duhem–Quine in thinking, roughly, that everything is evidentially connected to everything else. In particular, I agree with many that the metaphysical doctrine of realism is very hard to support if we argue from an epistemic view of truth (1984: 4.2; 1991b: 4.3; Ch. 2, Aberration 6, in the present volume). But, according to Maxim 3, that is the wrong way to argue: we should argue from metaphysics to semantics. So I reject John Haldane and Crispin Wright's implication— in arguing for Michael Dummett's close connection between metaphysics and semantics—that an evidence‐transcendent doctrine of truth is a necessary “semantical preparation” and “groundwork” for the metaphysical doctrine (1993b: 5–6). The metaphysical doctrine needs no such preparation or groundwork. Rather, the doctrine of truth needs the metaphysical groundwork.

(13) The unsatisfactory ones include my own in discussing quasi‐realism (1991b: 55). (14) E.g. see Boghossian 1990b: 265. A paradigm example of an error doctrine is John Mackie's view of morality (1977). (15) [2009 addition] This develops the objection so that it strikes at what I call, “the existence dimension” of the metaphysical doctrine of realism. Williams himself is more concerned with what I call “the independence dimension” of the doctrine: the view that the world is, as he says, “objective” and “independent of how we think”. The objection encourages him to continue thinking that “the obvious way to unpack” the independence dimension is in terms of “a radically non‐epistemic” notion of truth (p. 193). I argue that the dimension can and should be unpacked without any mention of notions of truth (1991b: particularly pp. 14–16, 246–56; in the present volume, Ch. 2, objection 3, and postscript, sec. 1; Ch. 6, sec. 3). (16) My discussion of this draws on my 1991b: 50–7.

Page 15 of 19

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Nonfactualism * (17) This point does not depend on any particular view of the nature of our ordinary understanding of a language. So the point survives the common, though I think mistaken, view that this understanding consists in (tacit) semantic propositional knowledge about the language. I argue (most recently in 1996: ch. 2), that this common view implies an unargued and implausible Cartesian access to semantic facts. We should take our linguistic competence to be simply a skill, a piece of knowledge‐how not knowledge‐that. [2009 addition] See also 2006d and Ch. 13, sec. 3.2, in the present volume. (18) [2009 addition] See Ch. 1, Postscript, secs. 1 and 2, of the present volume for a discussion of ontological commitment. I argue for the priority of an object‐ language criterion over a meta‐language criterion. (19) The argument against global nonfactualism has certain parallels with the argument against global “response dependency”. The conclusion of the latter argument is that global response dependency amounts to constructivist antirealism (1991b: 251–6; Ch. 6 in the present volume). (20) What if she is also a causal nonfactualist? Then the suggestions of this paragraph would have to be adjusted by replacing the allegedly nonfactual causal talk with talk that is descriptive of what the nonfactualist holds to be the only external reality underlying causal talk; say, constant conjunction. (21) It is usual to take realism to involve some claim about the objectivity and independence of the problematic reality's nature and role. We need not be concerned with such claims because nonfactualist disagreement comes “earlier”, with the claims that there is a reality with such a nature and role. (22) See e.g. Geach 1960; Dummett 1973: ch. 10; Blackburn 1984: 189–96; 1993a: ch. 10; 1993b; Wright 1988; Hale 1993. (23) One might be dubious of unobservable reality in general, or of some parts of it in particular, and hence be an instrumentalist in some sense, without embracing the described semantics. But that semantics is essential to traditional instrumentalism and makes it an example of the sort of nonfactualism I am discussing (cf. n. 11). (24) My earlier discussion of this is, therefore, mistaken (1991b: 129). (25) [2009 addition] This characterization of the metaphysics of instrumentalism may force a modification of my standard definitions of scientific realism; e.g. “Tokens of most current unobservable common‐sense and scientific physical types objectively exist independently of the mental” (1991b: 24; see also Ch. 2, part I, and Ch. 4, sec. 2, of the present volume). For we have noted that the instrumentalist may well go along with such language, interpreting it nonfactually. So the language will do the job of stating scientific realism only on Page 16 of 19

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Nonfactualism * the presumption that the language is factual. If we are not entitled to that presumption, which we may not be in arguing with the instrumentalist, we must add something along the following lines: ‘and have natures in virtue of which they causally interact with observable entities'. (26) Can she accept the claims but resist the realist interpretation of them by treating uses of ‘there are’ in “talk about unobservables” as uninterpreted? She could (although I don't think that she would), but doing so will not help her. ‘There are’ would still have its other, interpreted, use and the instrumentalist cannot prevent the realist from employing that use in these claims. The instrumentalist must still have doubts about the claims, so interpreted. (27) One sort of moral antirealist thinks that a moral judgment is implicitly relative to some norm; e.g. the utilitarian norm. Thus a person ought to do such and such only in that she ought to do it relative to some implicit norm; there is no “absolute” respect in which she ought to do it. This, combined with the view that no one norm is objectively better than any other, yields a fairly straightforward nonfactualist metaphysics, as Hartry Field (1994: secs. 3–4) shows in discussing Allan Gibbard (1990). This antirealism does not involve a semantics of the sort that is definitive of the nonfactualism I am discussing. (28) Peter Railton (1986) and Nicholas Sturgeon (1988) are realists who emphasize the explanatory role of moral reality. Similarly, Michael Slote (1971) emphasizes the explanatory role of aesthetic reality. John Mackie (1977) and Gilbert Harman (1977) are antirealists who deny the explanatory role of moral reality. (29) Blackburn may disagree: “If you tell me that injustice caused the revolution, I understand that there is some property that you take to give rise to injustice, and that caused the revolution. I must, in my own assessment, separate the truth of the causal story you are pointing toward, from my own verdict on whether it amounts to injustice. . . . Perhaps I would not myself call [the causal feature indicated] unjust, but I can assent to the explanation without endorsing the verdict on the feature” (1993a: 205–6). He can but he surely should not. If he does not think that the feature amounts to injustice then he does not think that injustice caused the revolution. (30) [2009 replacement] Note that this is true even of a subjectivist realist who thinks e.g. that for x to be good is for people to approve of x or for x to cause people to approve of her (although the subjectivist will have a hard time agreeing with the causal role just assigned to Hitler's depravity). The subjectivist agrees with the nonfactualist that only certain attitudes or emotions underlie moral talk, but disagrees in thinking that moral utterances describe rather than express that underlying reality. Related to this, the subjective realist,

Page 17 of 19

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Nonfactualism * unlike the nonfactualist, thinks that moral reality is constituted in some way out of that underlying reality and thus moral reality plays a causal role. (31) [2009 addition]. This account of the deflationary theory of truth is incorrect; see Ch. 8, particularly sec. 7, in the present volume. According to deflationism, properly conceived, it is the truth term not truth that is disquotational; and the equivalence schema explains the meaning of that term not the nature of truth. Indeed, truth has no nature to explain. (32) Or should do, at least. Sometimes deflationists wrongly claim an explanatory role for deflationary truth; see e.g. Horwich 1990: 45, and my 1991b: 278–80. (33) [2009 addition]. Another example would be causal nonfactualism. My account has an interesting problem here. The account suggests that the causal realist will think that causality itself plays a causal role. At first sight this alone may seem problematic, but it is not really: the causal realist may say that Bill smoked pot because pot got him high; and that I believe that Hitler's depravity caused millions to die because his depravity did cause this. The problem is that the causal nonfactualist can join the realist in these sayings, interpreting them in her special way. So, maybe the causal realist can be distinguished metaphysically from the nonfactualist only in what he says about the nature of causality; e.g. that it is constant conjunction + necessary connection. (The nonfactualist is, of course, likely to be a realist about constant conjunction but he cannot accept that it is causality on pain of turning into a realist.) (34) Thomas Nagel (1980: 114 n.) thinks that moral reality need not be explanatory. (35) Ayer muddies the water somewhat by denying that ethical statements are true or false (1952: 103, 107). Given his view that truth is merely deflationary (pp. 87–90), this denial must be a mistake unless he believes that ethical statements should not be asserted. For, on his view, saying that a statement is true is simply asserting it. So saying that it is true does not tell us that it is factual, nor anything else about it. (36) [2009 addition] See Ch. 8, secs. 3–4, in the present volume for a much more detailed discussion of the semantics of the truth term. This discussion brings out that although deflationism has important similarities to the nonfactualism discussed in this chapter, there are differences that may make it inappropriate to call it nonfactualism. (37) My views on nonfactualism benefited greatly from discussions with Georges Rey while writing Devitt and Rey 1991. Some of the present chapter builds on and modifies a brief discussion of the issue in that paper. (That paper was a response to Boghossian 1990b, which was a response to Devitt 1990, which was Page 18 of 19

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Nonfactualism * a response to Boghossian 1990a.) I thank the following for helpful comments on a draft of this chapter: David Armstrong, Lisa Busch, Keith Campbell, Hartry Field, Judith Lichtenberg, William Lycan, Georges Rey, and Michael Slote. The chapter also benefited from the discussion when a version was delivered at Princeton University in Oct. 1995.

Access brought to you by:

Page 19 of 19

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth *

Putting Metaphysics First: Essays on Metaphysics and Epistemology Michael Devitt

Print publication date: 2009 Print ISBN-13: 9780199280803 Published to Oxford Scholarship Online: February 2010 DOI: 10.1093/acprof:oso/9780199280803.001.0001

The Metaphysics of Truth * Michael Devitt (Contributor Webpage)

DOI:10.1093/acprof:oso/9780199280803.003.0009

Abstract and Keywords The difference between deflationary and correspondence truth remains unclear largely because of insufficient attention to the distinction between the metaphysics of truth and the linguistics of the truth term, and hence to what deflationary theories say, or should say, about that metaphysics. Emphasizing the similarity between deflationism and a sort of ‘nonfactualism’, this chapter argues that the metaphysics of deflationism should reject the need for and possibility of explaining the nature or causal role of truth. It is largely because of this metaphysics that deflationism rejects a standard semantics and a descriptive role for the truth term. Finally, the chapter argues that the case for correspondence truth over deflationism is strong provided we can explain reference. If we cannot then we should adopt deflationism. The heavy price for this would be eliminativism about meaning. Keywords:   deflationary truth, correspondence truth, truth term, nonfactualism, descriptive, reference, meaning eliminativism

1. Introduction The most popular theory of truth has probably been the correspondence theory. According to this theory, the truth of a statement or belief consists in some sort of correspondence between the statement or belief and the world.1 And truth is usually thought to play an important role in metaphysics. The correspondence theory has received several challenges. (1) Karl Popper and some logical positivists once thought that truth talk was unacceptably metaphysical and should be banished. They had the eliminativist Page 1 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * view that statements are not, from a properly scientific perspective, really true or false at all. But, since Tarski at least,2 this challenge to correspondence truth has disappeared. (2) Another challenge has come from the verificationist theory of truth, a theory that has received a boost from the work of Michael Dummett. On this view, the truth of a statement consists in its being warrantedly assertable, verifiable, or something similarly epistemic. And this is thought to be metaphysically important because of its association with antirealism about the external world. I think that verificationist truth does indeed lead to this antirealism and that, from a naturalistic perspective, this is sufficient reason to reject it (1991b: 4.3, 14.9; Ch. 2, part II, in the present volume). I shall not discuss it here. (3) In my view, the most interesting challenge has come from the deflationary theory of truth. The contrast between this influential theory and the (p.156) correspondence theory has been much discussed. Yet, I shall argue, the contrast remains unclear. That unclarity arises from insufficient attention to the distinction between the metaphysics of truth and the linguistics of the truth term; hence from insufficient attention to what the theories say, or should say, about the metaphysical issue. In arguing this in part I, I shall emphasize that deflationism is similar to a sort of “nonfactualism”. Then, against this background, in the much briefer part II, I shall summarize the case for the correspondence theory.

I: Distinguishing the Deflationary from the Correspondence Theory 2. Four Problems What exactly is the difference between the deflationary theory and the correspondence theory? The answer is not as easy to find as one might have expected. There are four related problems in finding it. The first problem is that the two theories have opposite focuses. Whereas the focus of the correspondence theory is on the nature and role of truth, the focus of the deflationary theory is on the nature and role of the truth term; for example, of ‘true’. The former focus is metaphysical, the latter, linguistic. So, an awful lot of what deflationists say does not bear directly on what the correspondence theorists say, and vice versa. A simple explanation of this difference in focus—too simple as we shall soon see —is as follows. Deflationism is really a sort of eliminativism, or antirealism, about truth: it deflates truth itself. We might say, very roughly, that according to deflationism, there is no reality to truth.3 Since there is no reality to truth there is nothing positive to be said about the nature of truth. However, unlike the eliminativists of challenge (1), deflationists have no objection to the use of the truth term. Indeed, they are enthusiastic about the term and have a great (p. 157) deal to say about its linguistic role and semantics.4 In contrast, correspondence theorists are realists about truth and therefore struggle to Page 2 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * explain its nature. But, for them, the truth term is just another one‐place relational predicate—like, say, ‘warranted’ or ‘patriotic’—with the standard sort of semantics of such predicates. This semantics is likely to start from the assumption that the term denotes the property truth or applies to all true things. This is so unexciting as to be hardly worth saying and the theorists are not usually inclined to say anything more exciting. In sum, the deflationist has little to say about the metaphysics of truth but much to say about the linguistic role of ‘true’, whereas the correspondence theorist has a lot to say about the metaphysics of truth but little to say about the linguistics of ‘true’. I am here describing a real difference in focus between the two theories. The second problem in distinguishing the theories is that this difference is often not apparent. Discussions of deflationism tend to blur the distinction between the linguistic and the metaphysical. In particular, remarks that should be about the truth term are often presented as being about truth: there is use/mention sloppiness, even confusion. So it can often seem that discussions are talking about truth when they are not really. As a result of the “linguistic turn” in philosophy, it has become common to slip casually back and forth between talking of words (or concepts, or notions) and talking of the world. Probably this is harmless in most cases, but in some it is not. One of these is realism about the external world. Another is truth. But, in my view, whereas it is relatively easy to avoid confusion in the case of realism, it is rather hard to avoid it in the case of truth. The third problem is that when discussions of deflationism do address the metaphysical issue, rather than merely appearing to when addressing the linguistic issue, what is said is often unsatisfactory. And this is not surprising because it turns out to be rather hard to capture the deflationary metaphysics of truth. That is the fourth problem. A sign of this problem is that my characterization of the metaphysics of deflationism in describing the first problem really is very rough. To appreciate the third and fourth problems it helps to notice that deflationism is similar to “nonfactualism” in ways to be explained (sec. 4). Despite these four problems, the difference between deflationary and correspondence theories on the linguistic issue of the truth term is relatively clear. Not so, the difference on the metaphysical issue of truth. As a result of the (p.158) four problems there is a good deal of uncertainty, if not confusion, over the difference between deflationary and correspondence views of the nature of truth. This is serious because this metaphysical difference is deeper than the linguistic one: it is explanatorily prior.

Page 3 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * Section 3 will be concerned with the linguistics of the truth term. Sections 4 to 6 will be concerned with the deflationary metaphysics of truth and the third and fourth problems. A standard characterization of the deflationary metaphysics will be criticized and a better one attempted. These accounts of the linguistic and metaphysical issues are necessary background for appreciating evidence in section 7 of the very tricky second problem: a use/mention sloppiness that obscures the real metaphysics of deflationism. The uncertainties and confusions arising from these four problems will emerge as we go along but will be particularly prominent in section 7. In talking of uncertainty and confusion I speak from bitter experience, for I am only too well aware that my own writings on truth have provided some examples.5

3. The Truth Term The deflationists have some very interesting things to say about the truth term. They have persuasively demonstrated that the term has an extremely useful “logical” or “expressive” role. Thus, suppose that Jack says, “We all lie about our sex lives”, and Jill replies, “That is true”. Intuitively, the role of the truth term here is to enable Jill to “say the same thing” as Jack without repeating his words (and whilst admitting his priority). Attention to such examples encourages the simplest deflationary theory, the “redundancy” theory, for they make it seem as if we could dispense with the truth term altogether. However, other examples show that the term is very useful. It enables us to assert briefly something that may otherwise be tedious, if not impossible, to assert. Suppose that Imogen wishes to express general but qualified agreement with a certain article. She can say simply, “Most of what that article says is true”. Consider what would be required to say this without using ‘true’. Her claim entails that at least half the claims in the article are true, but is not specific about which half. So her claim is equivalent to a long disjunction of conjuncts, each conjunct consisting of a different set of more than half the claims in the article. If she could remember all the claims, she could, in time, manage to express (p.159) this disjunction. If not, she needs the truth term. So does a person who has forgotten Goldbach's Conjecture but nevertheless wants to express agreement with it. He can say, “Goldbach's Conjecture is true”. A person who has lost track of all the utterances of the Great Helmsman can nevertheless express her commitment, ‘Everything Chairman Mao said was true’. Without the truth term, she faces the impossible task of asserting an infinite conjunction. So also does a logician in asserting each instance of a schema that has an infinite number of instances.6 The truth term can play its logical role because it yields equivalences like the classic one between ‘‘Snow is white’ is true’ and ‘Snow is white’. When the term is attached to the quotation name of a statement it yields a statement that is Page 4 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * equivalent to that statement: it undoes the effect of quotation marks. (Attention to this led to the name, ‘the disquotational theory of truth’.) Indeed, when the truth term is attached to any device for referring to a statement it yields a similar equivalence; it is a “denominalizing” device. Thus Jill's remark, “That is true”, is equivalent to Jack's, “We all lie about our sex lives”. If I were to say, “Jack's remark is true”, my remark would be equivalent to Jack's. The person who said, “Goldbach's Conjecture is true”, said something equivalent to the Conjecture. In general, the deflationary view supports “the equivalence thesis”: all appropriate instances of the “equivalence schema” s is true iff p hold, where an appropriate instance substitutes for ‘p’ a translation of the statement referred to by what is substituted for ‘s’.7

What is the “meaning” of the truth term? The deflationists have offered a variety of answers. Thus, Paul Horwich (1990) proposes a “minimal” theory according to which ‘true’ is an unusual “logical” predicate implicitly defined by its use in the appropriate instances of the equivalence schema. Dorothy Grover (1992) urges a “prosentential” theory according to which ‘true’ is not a predicate at all. Rather it is a syncategorematic part of an anaphoric “prosentence”, where prosentences are to sentences as pronouns are to nouns. This ingenious theory has the unhappy consequence that ‘that’ in ‘that is true’ does not refer to some statement, as one would naturally suppose (and as I supposed in introducing the equivalence thesis). This led Robert Brandom to propose a variation on the prosentential theory that avoided this consequence: (p.160) the truth term should be treated as a prosentence forming operator (1988: 88–90). So much for deflationary views of the meaning and role of the truth term. What are correspondence theorists to make of this? The important thing to notice is that they can, and should, go along with most it. Certainly, they cannot go along with a deflationary theory of meaning of the truth term, whether of the Horwich, Grover, Brandom, or any other variety. They think that the term has the standard semantics of a one‐place relational predicate, very likely explained in terms of reference to the truth property or to true statements, as I noted. Still they can and should accommodate the rest of the deflationary story. In particular, they should accept the equivalence thesis: that is a constraint on any theory of the truth term. And if the correspondence theory meets that constraint it can account for the logical role of ‘true’ that the deflationists have so persuasively demonstrated.8 So although the correspondence theorist disagrees with the deflationist over the meaning of the truth term, he should agree that the term has the logical role explained by the deflationist. There will probably be one other important disagreement. The deflationist will insist that the truth term does not have any role other than the logical one; in particular it does not have the “descriptive” Page 5 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * role of a normal predicate. The correspondence theorist is likely to think that the term has a substantial descriptive role in some theory of the world. These linguistic differences between a deflationary and a correspondence theory over the truth term are striking and obvious. The metaphysical differences between the two theories over the nature of truth are much less so. Yet the metaphysical differences are explanatorily prior because they largely motivate the linguistic ones.

4. The Deflationary Metaphysics of Truth: The Problem We have seen what the deflationists say about the truth term, but what is their view of truth? Where do they stand on the metaphysical issue? I have said that deflationism is a sort of eliminativism or antirealism, and roughly characterized it as denying that there is any reality to truth (sec. 2). But the inadequacy of this is apparent when we note that deflationists are as ready to talk about statements being true as correspondence theorists; thus many deflationists will (p.161) say that Jack's remark is true, because to say this is just to express the common belief that we all lie about our sex lives; and they will all agree that ‘Snow is white’ is true, because to say this is just to say that snow is white. So what does their antirealism consist in? The focus of deflationist literature is not on answering this question and the little the literature says is often unsatisfactory; that was our third problem in section 2. The question turns out to be rather hard to answer; that was our fourth problem. To appreciate these two problems it helps to realize that analogous problems arise elsewhere. For, deflationism about truth is similar to the “nonfactualism” exemplified by “noncognitivism” about morals, “projectivism” about causality, positivistic instrumentalism about science, and Simon Blackburn's “quasi‐ realism” (1984, 1993a, b).9 Characterizations of the metaphysics of nonfactualism also tend to be unsatisfactory and it is difficult to give a satisfactory one. I have discussed these problems for nonfactualism elsewhere and will draw on those discussions in what follows.10 Deflationism has two defining features of this kind of nonfactualism. The first is at the linguistic level and is very explicit in the literature. Nonfactualism in an area has a revisionist view of the language in that area: the language is not “descriptive” as we would naturally take it to be. This view is expressed in a variety of ways, some rather unsatisfactory, but the key idea is clear: terms that appear to be predicates in the area do not have the standard semantics of a normal predicate; perhaps they are not predicates at all. Because these terms are in this way “nondescriptive” they are not like a normal predicate in purporting to “describe reality”; they have some other role. Thus, the most famous nonfactualism, noncognitivism about morals, has a revisionist view of the semantics of ‘good’ as a result of which indicative sentences containing it are Page 6 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * not assertions or statements. Rather, those sentences express attitudes or emotions, or prescribe norms or rules.11 The deflationist view of the truth term, discussed in the last section, is a similar sort of revisionism. The truth term does not have the standard semantics of a normal predicate. And its role is not to describe sentences; its only role is logical or expressive. What is meant by “the standard semantics”? Typically a philosopher's standard semantics will be truth‐referential but it need not be: it might be (p.162) verificationist, for example. And the standard semantics of a deflationist cannot be truth‐referential because, for her, truth is not explanatory (sec. 5). So the non‐truth‐referential meaning that she attributes to a normal predicate—for example, a certain sort of use condition—she does not attribute to the truth term. Despite the linguistic similarities between deflationism and nonfactualism, there are important differences which should make one reluctant to treat deflationism as a species of nonfactualism. First, the deflationist's “expressive” role for ‘true’ is nothing like the noncognitivist's “expressive” role for ‘good’: the former is logical, the latter emotive. Second, the noncognitivist holds that, because ‘good’ is nondescriptive, sentences of the form ‘x is good’ are not factual. In contrast, the deflationist does not hold that because ‘true’ is nondescriptive, sentences of the form ‘S is true’ are not factual. For her, whether these sentences are factual depends on whether S is factual. So, if S is ‘x is good’, it is factual for a deflationist who is a cognitivist but not factual for one who is a noncognitivist. The second defining feature of nonfactualism is at the metaphysical level and is often more implicit than explicit in the literature. Nonfactualism in an area is antirealist about that area. Thus noncognitivists are antirealist about goodness. Deflationists are similarly antirealist about truth. Consider the problem of characterizing the antirealism of nonfactualism. The most straightforward way of characterizing antirealism in general, using the ordinary language for denying ontic commitment, obviously does not capture the metaphysics of nonfactualism. Thus the noncognitivist does not claim that there are no good people, right actions, and so on. She is as ready as the realist to say, “This person is good” and “That action is right”, for she thinks that these utterances express appropriate emotions or prescriptions. We have already made the analogous point about deflationism: the deflationist does not claim that there are no true statements. The nonfactualist and the deflationist talk like a realist but give that talk a revisionist interpretation.12 This is what poses the problem of distinguishing these doctrines from realism at the metaphysical level; cf. our fourth problem. But perhaps the problem is illusory. Maybe my confident claim that nonfactualism and deflationism are antirealist is mistaken. Perhaps the focus of Page 7 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * these doctrines is so linguistic because they really have no commitment one way or the other on the metaphysical issue. If this were so our enterprise of attempting to characterize their antirealism would be misguided. There are (p. 163) two reasons why this dissolution of the problem must be rejected. The first is that the doctrines are presented in opposition to realist views; thus, deflationists oppose correspondence truth. And, despite the linguistic focus, the doctrines are accompanied by claims that are clearly intended to be antirealist even if, as we shall see, the claims are often not adequate to the intention. The second reason is that an antirealist metaphysics is needed to motivate the revisionist view of language urged by these doctrines. If there were not something problematic or defective about the area of reality that ‘true’ or ‘good’ appear to concern why suppose that they do not have the standard semantics of a descriptive predicate? Of course, the semantic revisionism is typically supported by some purely linguistic considerations: evidence of a nondescriptive role for the language in question. Thus, deflationists are motivated by the logical role of the truth term and noncognitivists by the action‐guiding role of moral language. But what is to stop language covered by the standard semantics from playing these roles? Indeed, we have already suggested that a truth term with the standard semantics could play the logical role (sec. 3). So the antirealist metaphysics is still needed to make the standard semantics for this language unattractive. It is needed to show that the language does not have a descriptive role as well as the role emphasized by nonfactualism and deflationism. Behind the linguistic facade of these doctrines must lie an antirealist metaphysics. It is because deflationism's metaphysics is needed to motivate its semantics that the metaphysical difference between deflationism and the correspondence theory is explanatorily prior to the linguistic difference.13 So the problem of characterizing the antirealism of nonfactualism and deflationism remains. The most straightforward characterization is obviously hopeless and may have no adherents. However, another simple characterization is popular: the doctrines are said to deny that there are any properties in the area in question.14 Thus noncognitivism denies that there is a property of goodness and deflationism that there is one of truth. As soon as we look carefully at this popular characterization, we see that it cannot be satisfactory. This is our third problem with deflationism and the (p. 164) analogous problem with nonfactualism. The characterization is unsatisfactory because it overlooks the extent to which a philosopher's attitude to the metaphysics characterized might reflect a position on the general issue of realism about properties rather than on the particular problematic area of reality that is the concern of the nonfactualist or deflationist; for example, rather than a position on morality or truth. Thus, consider a nominalist. She will agree that there are no properties of goodness and truth because she thinks that there are no properties at all! Yet, manifestly, this alone does not commit her to Page 8 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * nonfactualism and deflationism; to thinking that there is something especially defective about the realms of morals and truth, something that motivates a revisionist semantics. She might be as realist as could be about morality and truth. Or, consider someone like David Armstrong (1978b) who is a selective realist about properties. Armstrong thinks that empty predicates, disjunctive predicates, and negative predicates have no corresponding properties. He thinks that some predicates apply to the world in virtue of many properties. Most importantly, he looks to science to tell us which properties there are. Such a person might well be a reductive realist, thinking that ‘good’ or ‘true’ apply to an object in virtue of properties none of which are goodness or truth; they apply in virtue of scientifically acceptable properties. So the popular characterization fits his views even though his metaphysics of goodness and truth is quite contrary to the antirealist one that we are attempting to characterize. Finally, consider the unselective realist who thinks that there is a property for each predicate. A nonfactualist might accept that ‘good’ is a predicate, as indeed Blackburn does (1993a: 206), and a deflationist might accept that ‘true’ is, as indeed Horwich does. If such a person is an unselective realist she will think that there is a property of goodness or truth, thus disagreeing with the popular characterization. And even if the nonfactualist denies that ‘good’ is a predicate, and the deflationist that ‘true’ is, hence that there are properties of goodness and truth, the popular characterization of their antirealism is dubious: it ‘runs the wrong way’. It finds a defect in reality because of something special about language where we need to find a defect in reality to motivate the view that the language is special.15 The general issue of realism about properties is independent of the issues of nonfactualism and deflationism. It should be possible for someone to embrace or reject the metaphysics of these doctrines whatever her position on this (p.165) general issue. There should be a way of stating that metaphysics that is appropriate whatever the truth of the matter about the reality of properties. So far, then, we have made no progress characterizing the antirealism of nonfactualism and deflationism. The most straightforward statements of realism, using the ordinary language of ontic commitment, are not denied by these doctrines because they are reinterpreted so that they have no such commitment. We have just seen the failure of a characterization using more “philosophical” talk of properties. In general, the nonfactualist/deflationist practice of talking like a realist while giving that talk a revisionist interpretation makes progress hard. We are attempting a characterization of the metaphysics that must motivate the special semantic treatment that the doctrines give to a certain area of language. Yet our attempts seem doomed to vitiation by that very semantic treatment. Nonfactualism and deflationism are supposed to be a sort of antirealism and yet it seems impossible to give a metaphysical statement of their antirealism. Realism issues begin to evaporate. Indeed, Blackburn sometimes

Page 9 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * comes very close to claiming that they have evaporated (1993a: 4, 15–34, 55–9; 1993b: 368).

5. The Deflationary Metaphysics of Truth: The Solution To avoid the evaporation of realism as a metaphysical issue and to characterize the metaphysics of nonfactualism in an area, we must first find some language in that area that is not just apparently descriptive but is treated by the nonfactualist as really descriptive. We must then examine her statements using that language to find ones that disagree with realist statements about the area. I have argued elsewhere (1996b, which is Ch. 7 in the present volume: 165–70; 1997a: 313–18) that two sorts of realist claim are the most promising candidates for denial by the nonfactualist. First, the typical realist offers explanations of the nature of the problematic reality in language that the nonfactualist should agree is factual. For, the realist thinks that the problematic reality is constituted by, or supervenes on, a reality that should be unproblematic for the nonfactualist. Even though the nonfactualist claims to be able to accept many sentences that seem to describe the problematic reality, taking them as expressive, prescriptive, or whatever, she does not accept the need for, or possibility of, these substantial “broadly reductive” explanations. Thus, moral realists claim that there are things about a person in virtue of which she is good, that make her good; for example, being kind, considerate, generous, honest, (p.166) etc. And there are things about an action in virtue of which it is wrong, that make it wrong; for example, leading to unhappiness, being contrary to socially accepted rules, and so on. The noncognitivist must reject all such “in virtue of” claims as totally misconceived. The deflationist has a similar disagreement with the typical realist about truth. The realist will claim that there is something common and peculiar to true statements: a statement is true in virtue of some sort of correspondence relation to the world; this relation makes it true. A substantial theory is then needed to describe and explain this correspondence, a theory that may include, for example, causal theories of reference. Deflationists should reject any such reductive explanation of truth. Horwich does so in denying that truth has an “underlying nature” or some “hidden structure awaiting our discovery” (1990: 2): “being true is insusceptible to . . . scientific analysis” (p. 6). Grover claims that “truth talk . . . can be explained without appeal to any kind of analysis of the nature of truth” (1992: 3). This is not to say that the deflationist rejects all statements of the form ‘p explains that S is true’. The deflationist, like everyone else, accepts the need for, and possibility of, explanations of “worldly facts” such as that snow is white, explanations that appeal to laws of nature. Suppose that E explains that snow is white. So, given the deflationary theory, E explains that ‘snow is white’ is true. But this sort of explanation, varying from truth to truth, is not what the Page 10 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * correspondence theorist seeks. He seeks an explanation of what all true statements have in common, an account of “correspondence to the world”. That is the sort of explanation that the deflationist must reject. The second sort of realist claim that the nonfactualist should deny concerns causal role. The typical realist thinks that the problematic reality is the cause or effect of some unproblematic reality. The nonfactualist should not accept these claims about the role of the problematic reality because on her view there is no reality that could play such a role. Thus, the typical moral realist thinks that there are causes and effects of a person being good or bad. He thinks that it is because Hitler and his associates were depraved that we believe that they were depraved. And it is because they were depraved that they behaved as they did and that millions of people died in concentration camps. The noncognitivist must reject all such explanations. Once again, the deflationist has a similar disagreement with the realist about truth. The typical realist will give truth important explanatory roles; for example, to explain the success of science or the success of people in meeting their goals; or to explain meaning, where meaning itself plays a role in the (p.167) explanation of behavior. A deflationist must reject all such explanations and Brandom clearly does reject them all (1988: 91–2).16 This is not to say that the deflationist cannot use the truth term in explanatory statements. The logical role of the truth term makes an explanation of the form ‘p because it is true that q’ equivalent to one of the form ‘p because q’. But the appearance of ‘true’ in the former sort of explanation does not make truth explanatory of p. Consider an example: ‘Clinton was impeached because he is hated by the religious right’ can be rewritten as ‘Clinton was impeached because it is true that he is hated by the religious right’. Manifestly, what is explanatory here is hatred not truth. Even where the expressibility provided by the truth predicate is essential to an explanation—because without it the explanation would be infinite—it is not truth that is explanatory. In sum, the typical realist thinks that there is a reality to truth which, like any other reality, has a nature and causal role; and that this nature and role need explanations. The deflationist reveals her antirealism by rejecting the need for and possibility of such explanations. Although she can join with the realist in accepting ordinary truth claims, she cannot join with him in his explanation of the reality which he takes those claims to describe. The deflationist should have nothing that is positive and substantial to say about truth. Sadly, this account of the distinction between realism and nonfactualism/ deflationism has a flaw, reflected in the frequent uses of ‘typical’. There are doubtless some philosophers who claim to be moral realists and yet join the noncognitivists in denying the need for an explanation of moral reality and in Page 11 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * denying that this reality has any causal role: it is inexplicable and epiphenomenal. One can imagine an analogous claim from someone who sees himself as a realist about truth. Such positions are deeply antinaturalist, of course. They are also hard to motivate: Why believe in a truth or goodness that can do nothing and cannot be explained? Still, the positions are possible. And if they have a standard semantics for ‘true’ and ‘good’ they surely are realist, for they accept the straightforward statements of realism without interpreting away the ontic commitment of the statements. So, the flaw in my account is that it does not distinguish this atypical realism from nonfactualism and deflationism at the strictly metaphysical level. I suspect that this realism cannot be so distinguished. If not, we must conclude, disappointingly, that to fully capture the antirealism of nonfactualism/deflationism we have to add a little semantics: what makes these doctrines antirealist is not only their denial of explicable nature and (p.168) causal role but also their adoption of a nonstandard semantics that removes the commitment from apparently straightforward statements of realism. The nonfactualist/deflationist and the atypical realist agree that in a certain area there is no reality with an explicable nature and a causal role. Despite this failure, the atypical realist holds that there is a reality in that area: the reality is simply inexplicable and epiphenomenal. In contrast, the failure motivates the nonfactualist/deflationist to reject the reality altogether by revising the semantics for what would otherwise be straightforward statements of realism.

6. The Equivalence Thesis The difference between deflationism and the correspondence theory should emerge in their responses to a demand for an explanation of the equivalence thesis.17 Let us take the most famous instance of the equivalence schema as our example: ‘Snow is white’ is true iff snow is white. In virtue of what is this so?18 The core of the correspondence theorist's answer is a reductive theory of the nature of true statements, of what is common and peculiar to these statements. This will be an account of the relation that true statements stand in to the world. When this theory is applied to ‘Snow is white’ it shows that this statement is related to the world in such a way that the statement is true iff snow is white. So the theory of truth, together with facts about the statement ‘Snow is white’, explain why ‘Snow is white’ is true iff snow is white. The deflationist, in contrast, cannot accept any appeal to a theory of the nature of truth in her explanation because she dismisses the possibility of saying anything substantial about that nature. So, what explanation does she offer? Basically, none. She thinks the demand for an explanation here is (p.169) misguided: that ‘Snow is white’ is true iff snow is white is a “brute fact” needing no explanation. However, she has something further to say to make this provocative claim palatable: a diagnosis of the error of thinking that we need an explanation here. The diagnosis moves up to “the semantic level”, considering Page 12 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * the way the brute fact is expressed. Although the deflationist denies the need to explain why ‘snow is white’ is true iff snow is white, she accepts the need to explain why people wrongly think that the statement ‘‘snow is white’ is true iff snow is white’ expresses something that needs explaining. The error arises from treating ‘true’ as if it were a normal descriptive relational predicate, thus taking the truth of ‘Snow is white’ to depend on some relation that statement has to snow being white. Once the nondescriptive meaning of ‘true’ is appreciated, we see that to say that ‘Snow is white’ is true is not to relate the statement in some way to the world but simply to say that snow is white. So, of course ‘Snow is white’ is true iff snow is white, just as snow is white iff snow is white.19 No more needs to be said (unless “the logical structure of the world” is to be explained). Similarly, it might be claimed that the following are brute facts needing no explanation: that Schnee ist weiss iff snow is white; that all bachelors are unmarried; and that Hesperus is Phosphorus. However, someone might think otherwise because she failed to appreciate the relevant semantic facts: that ‘Schnee is weiss' is synonymous with ‘snow is white’; that the meaning of ‘bachelor’ includes the meaning of ‘unmarried’; and that ‘Hesperus' and ‘Phosphorus' rigidly designate the same object. The contrast between the two theories should not be that the correspondence theory must offer a substantial explanation of why ‘Snow is white’ is true iff snow is white where the deflationary theory offers a trivial one appealing to the meaning of ‘true’.20 The contrast should be that the correspondence theory must offer an explanation where the deflationary theory appeals to the meaning of ‘true’ to explain why no explanation is necessary. The position I am attributing to the deflationist on this matter is undoubtedly hard to grasp. The position is developed and modified a little in the next section. I started this part of the paper by mentioning four related problems in distinguishing deflationism from the correspondence theory. The first problem was a difference in focus: the focus of deflationism is on the linguistics of the truth term, the focus of the correspondence theory on the metaphysics of truth. The third problem was the unsatisfactory nature of attempts to characterize (p. 170) the metaphysics of deflationism and the fourth was the difficulty of such a characterization. I have said a lot about these three but nothing yet about the second problem. We now have the background to discuss it.

7. Use/Mention Sloppiness The second problem was use/mention sloppiness, even confusion, in the literature: deflationist remarks that should concern the linguistics of the truth term are often misrepresented as being about the metaphysics of truth, thus obscuring the real metaphysics of deflationism. In giving examples of this sloppiness I do not mean to suggest that all of them amount to real confusions in thinking. Some surely are just insignificant carelessness or convenient rhetoric. Page 13 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * Still I want to show, first, how pervasive the sloppiness is. Second, I want to make it plausible that there are some cases of real confusion: that what should be a theory of the truth term is really being taken as a theory of truth, not simply carelessly expressed. This shows, it seems to me, how very difficult it is to handle the use/mention distinction in discussing truth. In the light of our discussion so far, it is easy to spot the sloppiness. The deflationist is talking about truth itself, and saying something appropriate, when she denies that truth has a nature or causal role that needs or can have an explanation (sec. 5). And she is talking about truth itself, but saying something inappropriate, when she denies that truth is a property (sec. 4). Anything else she says, particularly anything positive, that is represented as being about truth should very likely be about the truth term. 1. The problem starts with the very names of some deflationary theories. The name ‘the redundancy theory of truth’ implies that truth is redundant yet what is really redundant according to the theory is the truth term. Similarly, what is really disquotational according to ‘the disquotational theory of truth’ is the truth term not truth. What is prosentential according to the ‘the prosentential theory of truth’ is not truth but a linguistic expression including the truth term. The generic name, ‘the deflationary theory of truth’, does refer to theories that deflate truth not the truth term, and so the name does not confuse use and mention. Still, the name is a bit misleading because only a small part of what deflationary theories actually say concerns truth. What they say mostly concerns the truth term. They deny a descriptive role for the term but emphasize other roles that were largely unnoticed or ignored by correspondence theories. On balance, deflationary theories inflate the truth term. 2. Consider next an historically important but notoriously difficult case of deflationism: Alfred Tarski. A special difficulty is that Tarski does not see (p. 171) himself as a deflationist but rather, it seems, as a correspondence theorist (1956: 153, 404). On the opening page of “The Concept of Truth in Formalized Languages”, Tarski variously describes his enterprise as the definition of truth, of the term ‘true sentence’, and of the concept of truth (1956: 152). The last two can be taken to be the same21 but, prima facie, they are different from the first. Defining truth is a matter of explaining its nature, a metaphysical matter,22 whereas defining the term and the concept are linguistic matters. We have use/mention sloppiness. What does Tarski actually do? He defines the meaning of ‘true‐in‐L’, where L is any of a certain range of formal languages. Does this have anything to do with explaining the nature of truth? Set aside worries arising from the fact that he has defined ‘true‐in‐L’ not ‘true’ and suppose that he had defined ‘true’. Could that have shown anything about truth? It depends on the definition. In certain Page 14 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * cases we can move straight from a definition of a word's meaning to an explanation of the nature of the reality that the word concerns. For example, we can move straight from defining ‘bachelor’ as ‘adult unmarried male’ to the explanation: to be a bachelor is to be an adult unmarried male. So moving back and forth between talk of defining ‘bachelor’ and defining bachelorhood would be an insignificant use/mention sloppiness, of interest only to pedants. But a linguistic definition licences this move to metaphysics only if it treats the word in question as a normal descriptive predicate. Where the definition amounts to a revisionist view of the word's meaning, the definition cannot yield a substantial explanation of the nature of the reality that the word appears to concern.23 Indeed, an antirealist view of that reality is necessary to motivate the revisionist semantics (sec. 4). Consider an example: we could not move from a noncognitivist definition of ‘x is good’ as ‘Hooray for x!’ to an explanation of goodness as “Hoorayness”; and noncognitivism is partly motivated by an antirealist view of goodness. One lesson I think that we should draw from Hartry Field's classic article, “Tarski's Theory of Truth” (1972), is that Tarski's (p.172) definition of ‘true’ is of the revisionist sort and so, as it stands, does not show us anything substantial about truth. Tarski's use/mention sloppiness is of more than pedantic interest. Tarski's definition of ‘true‐in‐L’ rests on list‐like definitions of various referential words along the lines of the following definition of ‘designate’: ‘N designates x’ = df ‘N is ‘France’ and x is France or . . . or N is ‘Germany’ and x is Germany’ By comparing such definitions with a similar one for ‘valence’, Field brings out dramatically that the definitions do not yield satisfactory reductive explanations of the nature of reference.24 So, the definition of ‘true’ in terms of the referential words does not yield a satisfactory reductive explanation of the nature of truth. In the light of subsequent discussions, we can see why: the list‐like definitions are essentially deflationary25 and so could not yield anything substantial about reference. Indeed, in offering these definitions Tarski is implicitly committed to antirealism about reference: only if there were something problematic about reference would there be adequate justification for not treating the referential terms as ordinary two‐place relational predicates; for not saying, for example, that ‘designate’ designates the relation designation, or applies pair‐wise to all ordered pairs where the first member designates the second. Tarski shared the physicalism of the positivists and clearly did think that there was something problematic about both reference and truth. And that was the thought that drove his enterprise. Although Tarski seemed to view himself as a correspondence theorist about truth, the theory he actually presented is deflationary, as I think is now generally agreed. So there is a far from innocent use/mention confusion in representing Tarski's definition as a theory of truth, as Tarski and others do.26 Tarski's definition tells us a lot about ‘true‐in‐L’. It tells us nothing about truth‐ in‐L27 because it is implicitly committed to the view that there is nothing to tell.

Page 15 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * 3. Scott Soames begins “What is a Theory of Truth?” (1984), an important defense of Tarski from Field's criticisms, with the report: “Alfred Tarski's (p. 173) theory of truth and its successors . . . are commonly believed by philosophers to provide analyses of the nature of truth” (p. 411). If Soames is right that this belief is common—and I think he is—the use/mention confusion I have just noted is widespread. Soames does not share my view that this belief misrepresents Tarski's achievement but he notes that “there is considerable doubt about whether, or in what sense, [Tarksi's theory] is a theory of truth”. He goes on: One main reason for this uncertainty is the difficulty of determining what a theory of truth ought to be. Generally, theories of truth have tried to do one or the other of three main things: (i) to give the meaning of natural‐language truth predicates; (ii) to replace such predicates with substitutes, often formerly defined, designed to further some reductionist program; or (iii) to use some antecedently understood notion of truth for broader philosophical purposes . . . (p. 411) This is striking. Suppose that we wondered what a theory of, say, genes tries to do. Two things occur: (a) it tries to describe the role of genes—state the laws about genes —which is what Mendelian genetics does; (b) it tries to say what genes are—explain their nature—which is what molecular genetics does. Now explaining the role of genes is, near enough, analogous to Soames's (iii). But explaining what genes are has no analogue on Soames's list! The metaphysical task of explaining what truth is, which is surely what correspondence theorists and many others were trying to do, has become one or other of the two linguistic tasks, (i) and (ii). Use has become mention.

4. The theory Horwich proposes in his influential book, Truth (1990), is explicitly deflationist. Yet he talks positively of “the minimalist function” of truth (p. xii), of “the entire conceptual and theoretical role of truth” (p. 6), of “the properties of truth” (p. 26), and of “all the facts involving truth” (p. 7). Strictly speaking, on his antirealist theory, truth can have no function, role, or (nontrivial) property, and cannot be involved in any facts. The truth term is what has the function, role, properties and involvement. Horwich claims that his minimal “theory of truth . . . involves nothing more than the equivalence schema” (p. 12); it is “what is expressed by [the schema's] uncontroversial instances” (p. 7).28 But this is misleading at best for his theory of truth is (p.174) really to be found in various negative remarks about the nature of truth, some of which I have quoted (sec. 5). Of course, the word ‘true’ is used not mentioned in each instance of the equivalence schema and this might suggest that these instances explain the nature of truth. But that suggestion treats ‘true’ as a normal descriptive predicate which is precisely what Horwich, like all deflationists, denies: he has a revisionist theory of the truth term.29 This theory of the truth term is what really “involves nothing more than the equivalence schema”: it holds that every fact about the role of the term can be explained simply by taking the meaning of the Page 16 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * term to be implicitly defined by its use in the appropriate instances of the equivalence schema.30 The equivalence schema has nothing to do with his theory of truth, everything to do with his theory of the truth term.31 An analogy with goodness may help. Suppose that a noncognitivist were to talk positively of the function, role and properties of goodness, and of the facts involving goodness. Suppose that she claimed to be giving a theory of goodness that was quite clearly based on her views about the nondescriptive meaning and expressive role of ‘good’; for example, she claimed that her theory of goodness involves nothing more than the view that to say that x is good is just to express a pro‐attitude toward x. It would be obvious that she was misdescribing her position: on her view, such remarks should really apply only to ‘good’ not goodness. For her theory of goodness is, roughly, that there isn't any; more precisely, it is the view suggested in section 5. 5. Stephen Leeds's “Theories of Reference and Truth” (1978) and Robert Brandom's “Pragmatism, Phenomenalism, and Truth Talk” (1988)32 are two of the best brief presentations of deflationary truth. Leeds sketchs a disquotational (p.175) theory of the sort famously suggested by Quine (1970). He comments: “What we have sketched is not a theory of truth . . . but a theory of the concept of truth” (1978: 122). But then he spoils this assessment by claiming that his account explains “facts about truth‐in‐English” and “what we ordinarily say about truth” (p. 123). It doesn't. But it might explain facts about ‘true’ in English and ordinary uses of ‘true’. Brandom takes the “central theoretical focus” of deflationism to be “on what one is doing when one takes something to be true, that is, our use of ‘true’ ”. He goes on: “It is then denied that there is more to the phenomenon of truth than the proprieties of such takings” (1988: 77). But, strictly speaking, on the deflationary view the proprieties are not any part of the phenomenon of truth because, roughly, there is no such phenomenon. The only phenomena are truth takings. 6. Finally, consider Marian David's Correspondence and Disquotation (1994), the most detailed and informed critique of the disquotational theory of truth available. David starts his description of the disquotational theory by claiming that it unlike, say, the correspondence theory, is “an antitheory of truth”: its view is, “Truth has no nature”. So far, so good. But then he continues: “The correct explanation of truth . . . requires less extravagant resources”. The correct explanation is that “truth is disquotation” (pp. 3–4). But the disquotational view does not require less extravagant resources to explain truth, it does not require any because, properly understood, it is the view that truth does not need and cannot have an explanation. That is the respect in which it “has no nature”. And disquotation does not explain truth, it explains the truth term.

Page 17 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * Not surprisingly, when David sets out to find the unextravagant disquotational theory of what it is for a sentence to be true, he finds the theory “a bit elusive” (p. 62). The core of the disquotational theory is, of course, the equivalence schema. David worries away at the schema trying unsuccessfully to find in it a theory of truth other than a correspondence theory. Sometimes he comes close to realizing that he is seeking something that the disquotationalists think is not there to be found: they think that “sentence‐truth is in a sense ‘nothing’ ” (p. 65); Strictly speaking, the question [about truth] will not even receive a response with the right logical form to count as an answer to this question, for the grammatical truth predicate does not function like an ordinary predicate . . . Given that the standard way of answering “What is F?”‐ questions does not work when it comes to truth all one can do is describe the linguistic role that the term ‘true’ plays in our language. (pp. 68–9) Just so. Still he remains puzzled: “where does the deflationary idea that truth is nothing but disquotation come from?” (p. 69). Deflationists have given him (p.176) reason to be puzzled, as we have seen. Despite what is often suggested, the disquotational view should not be that truth is nothing but disquotation. The view should be that truth is nothing.

In this section, I have indicated how pervasive use/mention sloppiness is in the discussion of deflationary truth.33 Some of this sloppiness is surely insignificant. Yet I hope to have shown that some of it is not: a theory of the truth term is really being taken as a theory of truth. This helps to obscure the metaphysics of deflationary truth and hence the difference between deflationism and the correspondence theory described in sections 5 and 6.

8. Summary In this part, I have attempted to bring out the real difference between deflationism and the correspondence theory by emphasizing the similarity between deflationism and nonfactualism. At the linguistic level, the real difference is fairly apparent. The correspondence theorist can, and should, grant that the truth term has the logical role emphasized by the deflationist. But the correspondence theorist does not accept the deflationary view that the term has no other role: he holds that it has a descriptive role. Furthermore, he thinks that the term has the standard semantics of a one‐place descriptive predicate, a view that the deflationist rejects. At the metaphysical level, the real difference between the two theories is much harder to discern. The typical correspondence theorist thinks that truth has a nature and causal role that need explaining. The deflationist should reveal her antirealism in the characteristic nonfactualist way by rejecting the need for and possibility of any such explanation. Finally, the metaphysical difference motivates the linguistic one, implicitly if not explicitly: it is largely because of her antirealism that the deflationist rejects a standard semantics and a descriptive role for the truth term. Page 18 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * I have located the difficulty in discerning the metaphysical difference in four problems. The first problem is a difference in focus: the focus of deflationism is on the linguistics of the truth term, the focus of the correspondence theory on the metaphysics of truth. The second problem, just illustrated in some detail, is that use/mention sloppiness in discussions of deflationism tends to obscure the real metaphysics of deflationism. The third problem is that when discussions do address the metaphysical issue, rather than merely appearing to when addressing the linguistic issue, what is said is often unsatisfactory. (p.177) And this is not surprising because it turns out to be rather hard to capture the deflationary metaphysics of truth, as it is to capture the metaphysics of nonfactualism. That is the fourth problem. Having clarified the difference between the deflationary theory and the correspondence theory, I shall now summarize the case for the correspondence theory.

II. The Case for the Correspondence Theory 9. Knowledge of Equivalences I start my case by acknowledging something that may seem to count against the correspondence theory in favor of the deflationary theory. It is a striking fact that people competent with a truth term tend to believe instances of the equivalence schema; for example, that ‘snow is white’ is true iff snow is white. This fact is easy for a deflationary theory to explain because according to that theory such beliefs are obvious once one has mastered the truth term: what is believed is then as obvious as that snow is white iff snow is white (sec. 6). But how can a correspondence theory explain the striking fact?34 According to that theory the beliefs hold in virtue of a certain complicated relation that statements stand in to the world, a relation that the best correspondence theorists have yet to explain satisfactorily. How come so many people innocent of the correspondence theory nonetheless believe instances of the equivalence schema? A deflationist might reasonably claim that there is a strong argument for deflationism here.35 I do not have a response to this claim that fully satisfies me. Still I am dubious about the claim. The correspondence theory holds that the nature of truth (p. 178) is such that the equivalences hold. Certainly, the theory cannot pretend that mastering the truth term teaches you the theory of that nature, any more than it teaches you how to recognize the worldly situations that make statements true. But why could it not be the case that mastering the truth term does teach you the equivalences? For, surely, mastering the term teaches you that statements are true if the world is as the statements describe. And this, together with some obvious background knowledge will yield belief in the equivalences. The equivalences do not capture the nature of truth, they are simply a readily apparent manifestation of that nature.36

Page 19 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * 10. Explaining Meaning In light of our part I discussion, the basic form that a case for correspondence truth must take is clear: truth plays a causal‐explanatory role in the world and has a nature that can be explained along correspondence lines. The core of any case for the rival deflationary theory must be the denial of this. For, the denial is necessary to motivate the revisionist semantics that the deflationary theory proposes for the truth term. Without the denial, the correspondence theorist can accept that the truth term has the logical role emphasized by the deflationist whilst insisting that the term has the semantics of a normal descriptive predicate. Do we need correspondence truth to explain anything? It has been usual to think that correspondence truth has one or more important roles in metaphysics. Correspondence truth is thought to explain individual success, species success, the observational success of scientific theories, or the convergent view of scientific progress. Related to this correspondence truth is often thought to be essential to realism about the external world. I am skeptical about all this. In particular, I have argued that the issues of realism about the external world and truth have almost nothing to do with each other (1984/1991b; 1997a; Ch. 2 in the present volume; but see 2 below for the interesting exception). In my view, correspondence truth is important to semantics not metaphysics. We need correspondence truth to explain meanings; more accurately, we need correspondence truth conditions to explain meanings. Truth conditions, hence truth, then play a causal role because meanings do. The correspondence truth that is needed is not the traditional one of “correspondence to the facts” but a contemporary one explained in terms of syntactic structures and the reference of words that go into structures. (p.179) I argued for this representational view of meaning in Coming to Our Senses (1996a). Here is a summary of the argument: 1. The argument starts by asking the following. What are meanings (contents) supposed to do? What theoretical purpose do we serve by ascribing them? Meanings are supposed to play at least two very important roles. In virtue of having meanings, beliefs contribute to the explanation of behavior and guide us to the reality that the beliefs concern. In virtue of the fact that statements express beliefs (“language expresses thought”) the meanings of statements have, derivatively, the same two roles. 2. Next consider what folk and social scientists, rightly or wrongly, ascribe to beliefs and statements for these purposes of explaining behavior and being guided to reality; consider what these people, in effect, take to be meanings. The properties they ascribe, using ordinary attitude ascriptions like ‘X believes that . . .’ and ‘Y says that . . .’, are, as Page 20 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * a matter of fact, entirely constituted by properties that go into determining truth conditions and reference. This is as true when the ascriptions are opaque as when they are transparent. So no “stereotype” or non‐reference‐determining functional role is ever ascribed. So the semantic status quo is truth‐referential; and it is not holistic. Furthermore, these truth‐referential properties cannot be explained in epistemic terms on pain of slipping into antirealism (sec. 1, item (2)): so the properties involve correspondence not verificationist truth. 3. But is the status quo right? We have good reason to think so because these ordinary ascriptions, by folk and social scientists, are mostly successful. So probably the properties they ascribe really do explain behavior and guide us to reality. So probably those properties really are meanings. 4. Finally, there is no convincing argument for revising this status quo, for thinking that some properties other than the nonepistemic truth‐ referential ones we ordinarily ascribe would better explain behavior and guide us to reality. There is no good reason to believe in two‐factor theories, conceptual‐role theories, use theories, and the like. Indeed, these are not so much theories as hand wavings. At the time I presented this argument I had not read Brandom's Making it Explicit (1994). It is as far from hand waving as one could get. Brandom has always accepted the onus on the deflationist to provide an alternative to the usual truth‐referential approach to meaning.37 In this massive book, he offers (p.180) a very detailed use theory, drawing on ideas from Kant, Wittgenstein, Sellars, Dummett, and others.38 The theory gives a “broadly inferential” account, explaining meanings (contents) holistically in terms of inferences, language entries in perception, and language exits in actions. These various “conceptual‐ role” processes are in turn explained in terms of the social practices of undertaking and attributing commitments to think and act in appropriate ways. The practice of attributing commitments—interpreting—is more basic. These social practices are implicitly normative in that they implicitly acknowledge the correctness of certain performances. Finally, this implicit normativity is explained in terms of sanctions. However, that does not naturalize the normativity because the sanctions themselves are “internal” to the normative system. Brandom is totally up‐front about abandoning naturalism. In the preface he notes that his theory “makes essential use of normative vocabulary” in specifying use. “No attempt is made to eliminate [this vocabulary], in favor of nonnormative or naturalistic vocabulary” (p. xiii). Later he describes his order of explanation as the reverse of the traditional naturalistic one that treats representation as fundamental and hopes to explain “the normative character of the practice in

Page 21 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * which intentional states are significant . . . in ultimately naturalistic terms” (p. 149).39 My own representationalism in Coming to Our Senses is naturalistic in both the epistemological sense (rejecting a priori knowledge) and the metaphysical sense (physicalism).40 And my criticisms of holism, conceptual‐role theories, and so on, presuppose naturalism. Perhaps criticisms of this sort could be shown to bear against Brandom without the presupposition. In particular, perhaps he is susceptible to the charge that we do not as a matter of fact ever ascribe holistic conceptual‐role meanings to explain behavior and guide us to the world. Yet I suspect that, from his “interpretative” perspective, he has answers to such criticisms: his book conveys the sense that he has “thought of everything”. So, it may be that the objection to Brandom's theory of meaning is simply that it is not naturalistic. Still, this is a very serious objection because naturalism is worth dying for. (p.181) If we hold to naturalism, have we then decided the case for correspondence truth against deflationism? Sadly not.41 That case depends not only on correspondence truth having an explanatory role in the theory of meaning but also on its having a satisfactory explanation itself. My claim was that it can be explained in terms of syntax and reference. This explanation in turn requires naturalistic explanations of syntax and reference. The latter, at least, has proved very hard to come by. I think that we should be optimistic that some combination of ideas from historical‐causal, indicator, and teleological theories will do the trick (Devitt and Sterelny 1999: 156–62). But suppose that this optimism is misplaced, what then? Something appalling: we would have to do what Quine has already done: abandon meaning altogether. But we would not have to abandon truth: we could accept the deflationary theory. In sum, from a naturalistic perspective, the case for correspondence truth over deflationism is strong provided we can explain reference. If we cannot explain reference then we should adopt deflationism. The very heavy price for this would be eliminativism about meaning. This price seems so heavy that surely we should be optimistic about explaining reference.42 Notes:

(*) First published in Michael Lynch (ed.), The Nature of Truth (Devitt 2001b). Reprinted with kind permission from MIT Press. “The Metaphysics of Deflationary Truth” (Devitt 2002b) is a shortened version. (1) For convenience I shall mostly just talk of statements. By ‘statement’ and ‘belief’ I refer to meaningful (contentful) tokens. Some prefer to talk of the truth of propositions. Nothing I say hinges on which preference is right, so far as I can see.

Page 22 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * (2) Popper credits Tarksi with making it respectable to engage in truth talk (1968: 274). (3) Realism about truth is not to be confused with realism about the external world (1984/1991b; Ch. 2 in the present volume). Still, realism about truth is like the other realism in having two dimensions. The first of these is an existence dimension, committed, very roughly, to the reality of truth. That is what eliminativists, including deflationists, deny. The second dimension is an independence dimension, committed to the reality of truth being appropriately independent of our minds. The verificationist theory of truth is antirealist in virtue of denying the independence dimension. The correspondence theory, as usually understood and as I am understanding it here, is committed to both dimensions and so is realist. Still, there could be versions of the theory without the independence dimension and, according to Richard Kirkham, have been (1992: 133–4). (4) The “semantics” of a term concerns its meaning. So also does its “linguistics” but the latter may also concern other aspects of the term's nature and role. (5) E.g. 1991b: 3.4; 1991c. (6) The expressive power that we get from the truth term could also be obtained by introducing sentence variables into our language; see Horwich 1990: 4–5 n. for an interesting discussion. (7) This is rough; e.g. we need to guard against the semantic paradoxes and allow for indexicals. The formulation talks of translation because an appropriate instance might refer to a statement that is not in the language of the instance; e.g. ‘Schnee ist weiss' is true iff snow is white. (8) As Mark Lance in effect, points out (1997: 184–5). (9) Boghossian 1990a and b argue that deflationism is inconsistent with nonfactualism about an area and that deflationism itself is incoherent. For some responses, see Devitt 1990; Devitt and Rey 1991; Soames 1997. (10) 1996b, which is Ch. 7 in the present volume; 1997a: 307–20. The former includes a brief, and it now seems to me, somewhat mistaken discussion of deflationary truth (pp. 169–70). (11) See e.g. Ayer 1952: 103, 107; Sayre‐McCord 1988c: 4–5, 8; Boghossian 1990a: 160–1, 164; Blackburn 1993a: 3, 60; 1993b: 365; Haldane and Wright 1993b: 11–12; Hale 1993: 337, 340. (12) Note that the nonfactualist is not speaking a different language from the factualist. Rather, she has a different theory of the language that they both speak. Page 23 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * (13) Concerning realism issues in general, I have argued that, from a naturalistic perspective, we should always “put metaphysics first” by establishing a metaphysical base with near enough no appeal to semantics and by arguing from that base for a semantics. For we know far more about the world than we do about meanings (1991b; 1997a; Devitt and Sterelny 1999: 11.4, 12.4; Chs. 3 and 5 in the present volume). (14) Or that there are any facts in the area. That characterization has similar problems to the one about properties. For examples of these characterizations, see Ayer 1952: 89; Wright 1988: 29–30; Sayre‐McCord 1988: pp. ix–x, 4; Brandom 1988: 90–1; Boghossian 1990a: 157–9, 161–2; Grover 1992: 14; Blackburn 1993a: 3, 52, 57; Hale 1993: 337; Railton 1993: 280; Soames 1997: 4; Lynch 1998: 112. (15) Kirkham notes the problem that the realism issue about properties poses for the popular characterization. His solution is to characterize deflationism as the thesis “that ‘true’ is not a genuine predicate” (1992: 311). One objection to this is that some deflationists—e.g. Horwich—think that ‘true’ is a genuine predicate. A more serious objection is that it is a linguistic characterization and we need a metaphysical one. (16) See also Grover 1992: 14. However, Horwich does claim an explanatory role for truth (1990: 45). I have argued against this (1991c: 278–80). (17) [2009 addition] Douglas Patterson makes a good point against the view that a correspondence theory must explain the equivalence thesis (2002: 8–12). A correspondence theory of the sort I am proposing explains what it is for a sentence to be true in terms of its syntactic structure and referential relations— which each require in turn substantial theories—and, of course, the way the world is. So, given information about the syntactic and referential properties of a particular sentence, s, the correspondence theory would tell us how the world has to be for s to be true; it has to be, say, that p. But, as Patterson emphasizes, ‘s is true iff p’ is an instance of the equivalence schema only if ‘p’ is a translation of, is meaning equivalent to, s and that is not something that the correspondence theory needs to tell us: the theory is about truth not meaning. Still, a correspondence theory does have an explanatory obligation with respect to the equivalence thesis. Suppose that ‘p’ is meaning equivalent to s then the correspondence theory has to explain why, given that equivalence, ‘s is true iff p’. (18) Note that we are asking in virtue of what ‘Snow is white’ is true iff snow is white, not in virtue of what ‘‘Snow is white’ is true iff snow is white’ is true. Deflationism and the correspondence theory would give similarly different responses to the latter question, but the responses would be more complicated.

Page 24 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * (19) Compare: “All the anaphoric [prosentential] theory of truth tells us about what it is for ‘Snow is white’ to be true, is that it is for snow to be white” (Lance 1997: 188). (20) Which is what I have suggested in previous discussions (e.g. 1991b: 32–3; 1991c: 276–7). My mistake arose from a use/mention confusion of the sort discussed in the next section. (21) In general, I take it that our concept or notion of truth can near enough be identified with the meaning of the truth term. In “The Semantic Conception Conception of Truth”, after a similar variety of descriptions of his enterprise, Tarski has this to say about his usage: “The words ‘notion’ and ‘concept’ are used in this paper with all of the vagueness and ambiguity with which they occur in philosophical literature. Thus, sometimes they refer simply to a term, sometimes to what is meant by a term, and in other cases to what is denoted by a term” (1949: 80 n.). (22) As in scientifically defining water as H2O; cf. the linguistic definition of ‘vixen’ as ‘female fox’. (23) If a predicate is covered by a description theory, as our example takes ‘bachelor’ to be, it will have a definition. If it is covered by a causal theory, as words like ‘tiger’ very likely are, it will not have a definition. So an explanation of the nature of Fs can be derived from a theory of ‘F’ only if ‘F’ is both a normal descriptive predicate and covered by a description theory. (Even where the explanation of nature can be derived in this way, it should not be, in my view. We should start with metaphysics not semantics because we know more about the world than about meanings; see n. 13.) (24) The definition of ‘designate’ does yield: for N to designate x is for N to be ‘France’ and x to be France or . . . or N to be ‘Germany’ and x to be ‘Germany’. Perhaps we could count this as an explanation of the nature of reference: talk of “nature” is not clear enough to rule this out; cf. sec. 6. But we cannot count it as substantial reductive explanation of the sort indicated in sec. 5. (25) Also implausible. Brandom 1984 is a more appealing deflationary theory. (26) E.g., Rudolf Carnap talks of Tarski's “definition of truth” (1963: 60); Kirkham, in his impressively thorough introduction to theories of truth, nicely distinguishes the metaphysical project from the linguistic one (1992: 20–1) but then places Tarski “firmly with the metaphysical project” (p. 33). (27) As it stands, but if we revised it by dropping its list‐like definitions, then we could see it as yielding an explanation of truth in terms of reference, as Field points out. If this were then supplemented by a substantial theory of reference, we would have a correspondence theory of truth. Page 25 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * (28) Consider also: “The basic idea for deflationary theories of truth . . . is roughly that there is no more to truth than the equivalence thesis” (Devitt 1991b: 30; see also Ch. 6, sec. 8, of the present volume); “The deflationist tells us . . . : Truth's ‘nature’, such as it is, is (pretty much) exhausted by the equivalence of a claim p with the claim p is true” (Richard 1997: 57). (29) Horwich is happy to go along with the unselective realist about properties, holding that ‘true’ is a predicate referring to a “logical” property (p. 38). So, in that respect, instances of the schema are “about truth”. But it is still a mistake to think that the instances say anything substantial about the nature of truth. Truth as a logical property has no nature open to reductive explanation. Indeed, it has no properties except trivial ones like being logical and being a property. And although we might perhaps take the equivalence schema to yield an explanation of truth, it does not yield a substantial reductive one; cf. n. 24. (30) On this see pp. 34–7 (abstracting from the conflation of meaning with a speaker's knowledge of meaning; cf. Devitt and Sterelny 1999: ch. 8). (31) Kirkham takes Horwich at his word and so sees his remarks about the equivalence schema as an answer to the metaphysical question about the nature of truth (1992: 339). As a result of this, and Horwich's acceptance that truth is a property, Kirkham does not classify Horwich as a deflationist. Soames takes “the leading idea” of deflationism to be that the equivalence schema “is in some sense definitional of the notion of truth” (1997: 4). The talk of “notion” makes this appropriately linguistic. But the talk immediately follows the inappropriately metaphysical: “the equivalences . . . are crucial in explaining what truth consists in” (p. 3). And it is immediately followed by the claim that the statement, “there is no such property as truth”, which is straightforwardly metaphysical, is a variation of it (p. 4). (32) Brandom's excellent paper is sadly neglected; it gets no mention e.g. in Kirkham's encyclopedic discussion (1992). (33) [2009 addition] Patterson 2002 is a recent example; Båve 2009 is not. (34) [2009 addition] I noted earlier Patterson's objection to the idea that a correspondence theory must explain the equivalence thesis (n. 17). For a related reason, Patterson rejects my assumption here that the correspondence theory has to explain why people tend to accept instances of the equivalence schema (2002: 13–14). I think he is right again. But the explanation I give in the text does not in fact call on the correspondence theory. Clearly I should have simply pointed out that there has to be an explanation that is compatible with the correspondence theory. And the explanation I go on to give (for what it is worth) is compatible. Patterson's own explanation is that “people accept such biconditionals for languages they understand because meaning is truth conditional and understanding a language is knowing what its sentences Page 26 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

The Metaphysics of Truth * mean . . . what the truth‐conditions of its sentences are” (p. 14). This is not an explanation I can accept because I have persistently argued against the view that competence in a language consists in any sort of propositional knowledge about the language. In my view, linguistic competence is knowledge how not knowledge that (1981: 95–110; 1991b: 270–5; 2006d; Ch. 13, sec. 3.2, in the present volume). (35) E.g. see Horwich 1997: 95–6. (36) See also Field 1986: 71–5; Richard 1997: 69–73; Lance 1997: 185–7. (37) In general, Brandom views the truth issue very much the way I do even though we end up with opposite conclusions. Indeed, he might well endorse part I of this chapter. (38) Brandom 1997 is a helpful summary. (39) I have offered a brief example of such an explanation of norms in discussing the skeptical position that Kripke finds in Wittgenstein (Devitt and Sterelny 1999: 213–14). (40) See Ch. 12, part I, preliminary point (2), in the present volume for a discussion of these two senses of naturalism. (41) [2009 addition] Since this chapter was written, Paul Horwich has supported his deflationist theory of truth with a use theory of meaning that does not, like Brandom's, rest on unexplained normativity (1998, 2005). I have criticized the theory (2002d, 2009d). (42) Special thanks to Hartry Field whose doubts about the main theses of this chapter have led to many changes. I am grateful for comments at the University of Sydney and the Graduate Center of City University of New York when versions of the chapter were presented. My thanks also to Marian David, Paul Horwich, Mark Lance, Bill Lycan, Paul Pietroski, and Georges Rey for helpful comments.

Access brought to you by:

Page 27 of 27

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Moral Realism A Naturalistic Perspective *

Putting Metaphysics First: Essays on Metaphysics and Epistemology Michael Devitt

Print publication date: 2009 Print ISBN-13: 9780199280803 Published to Oxford Scholarship Online: February 2010 DOI: 10.1093/acprof:oso/9780199280803.001.0001

Moral Realism A Naturalistic Perspective * Michael Devitt (Contributor Webpage)

DOI:10.1093/acprof:oso/9780199280803.003.0010

Abstract and Keywords This chapter argues for moral realism. What is moral realism? Standard answers in terms of truth and meaning are rejected. These answers are partly motivated by the phenomenon of noncognitivism. Noncognitivism does indeed cause trouble for a straightforwardly metaphysical answer but still such an answer can be given. Moral realism should be accepted because it is prima facie plausible and its alternatives are not. But what about the arguments against moral realism? The chapter looks critically at the argument from ‘queerness’, the argument from relativity, the argument from explanation, and epistemological arguments. But there is a major worry for moral realism: How can it be accommodated in a naturalistic world view? The chapter concludes with some brief and inadequate remarks in response to this question. Keywords:   moral realism, truth, meaning, noncognitivism, queerness, relativism, naturalism

I propose to sketch a defense of moral realism from a naturalistic perspective. My naturalism has two quite distinct aspects. Its epistemological (and methodological) aspect is the view that there is only one way of knowing, the empirical way that is the basis of science. Its metaphysical aspect is that all facts must ultimately depend on physical facts. I shall say a little more about this naturalism later. I start by considering what moral realism is. Next I give some reasons for believing it. Then I consider four arguments against it. I conclude with some thoughts on fulfilling the naturalistic project.

Page 1 of 15

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Moral Realism A Naturalistic Perspective * 1. What is Moral Realism? What is moral realism? This question is more difficult and controversial than one might have expected. My first attempt at an answer is: MR1: There are objective moral facts. This answer has a problem which I will address soon. But many will think that my answer has another problem: it is entirely nonsemantic. Compare it with the following, more fashionable, answer from Geoffrey Sayre‐McCord:

Wherever it is found . . . realism involves embracing just two theses: (1) the claims in question, when literally construed, are literally true or false (cognitivism), and (2) some are literally true. Nothing more. (1988c: 5) What is it for something to be “literally construed”? To answer this, Sayre‐McCord thinks, we need a theory of meaning (1991: 157). So on his view the (p.183) realism debate is all about truth and meaning. Consider, next, the following from Peter Railton:

A chief cause of the changing character of the realism dispute over time has been changes in philosophical approaches to language and meaning . . . philosophers increasingly are clear that questions about meaning are intimately bound up with questions of metaphysics, epistemology, mind, empirical science, and even rationality and evaluation. (1996: 49–50)1 Once again, theories of meaning are taken to be central to realism debates. Sayre‐ McCord and Railton are particularly concerned with the debate over moral realism but their remarks are quite general. Their semantic view of realism debates is typical. Consider, for example, Jarrett Leplin's editorial introduction to a collection of papers on scientific realism. He lists ten “characteristic realist claims” (1984b: 1–2). Nearly all of these are about the truth and reference of theories. Not one is straightforwardly metaphysical.2

The popularity of these views of realism debates is surprising. One would have thought that these debates are metaphysical, concerned with what the world is like, not with our talk about the world. John Mackie, another unfashionable Australian, has it right: he describes his opposition to moral realism as “an ontological thesis, not a linguistic or conceptual one” (1988: 95). What he is opposed to is a thesis about moral reality, along the lines of MR1. The debate over this has almost nothing to do with any theses about meaning or truth. In Realism and Truth (1984/1991b: ch. 4; 1997a: 304–7; see also Ch. 2, part I, in the present volume), I have argued that semantic issues have almost nothing to do with the debate over realism about the external world. I shall briefly apply this argument to the moral realism debate. On the one hand, MR1 does not entail any theory of truth or meaning at all, as is obvious at a glance. In particular MR1 does not entail a correspondence theory of truth for moral statements. On the other hand, no theory of meaning or truth entails MR1.

Page 2 of 15

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Moral Realism A Naturalistic Perspective * Not only is MR1 independent of any doctrine of truth, it does not even use the truth term (e.g. ‘true’) to state moral realism (cf. Sayre‐McCord). This is not to say that there is anything “wrong” with using the truth term for this purpose. We can simply exploit the “disquotational” property of the term to redefine moral realism as: (p.184) MR1*: There are moral statements that are objectively true. But this exploitation of the disquotational property to “semantically ascend” does not make moral realism semantic (else every doctrine could be made semantic); it does not change the subject matter at all. It does not involve commitment to the correspondence theory of truth, nor to any other theory. Indeed, it is compatible with a deflationary view of truth according to which, roughly, there is nothing to truth.3

Why has the metaphysical issue been conflated with semantic issues? This is a difficult question but part of the answer is surely the “linguistic turn” in twentieth‐century philosophy. At its most extreme, this turn treats all philosophical issues as about language. Another part of the answer is the dominance of worries about noncognitivism. For noncognitivism seems to be a doctrine that is both antirealist and primarily semantic. I shall discuss noncognitivism in a moment. I claim that no semantic doctrine is in any way constitutive of moral realism. This is not to claim that there is no evidential connection between the two sorts of doctrines. Indeed, I favor the Duhem–Quine view that, roughly, everything is evidentially connected to everything else. So the possibility remains that semantic doctrines might be used as evidence against moral realism. But arguments that proceed in this direction should give us pause. Suppose, as I do (sec. 2), that moral realism is prima facie plausible. Should we have sufficient confidence in any semantic doctrine to allow it to undermine moral realism? Perhaps realism could be undermined in other ways—and we shall soon consider some possibilities—but could a semantic doctrine alone do the trick? A Moorean response is appropriate: moral realism is much more firmly based than any semantic doctrine that is thought to undermine it. We have started the argument in the wrong place: rather than using the semantic doctrine as evidence against realism, we should use realism as evidence against the doctrine. We should, as I like to say, “put metaphysics first”. Naturalism supports this Moorean response. The semantic doctrines that feature in arguments against realism are typically presented as if they were known a priori. But the epistemological aspect of naturalism rejects a priori knowledge altogether: philosophy becomes continuous with science.4 And (p.185) the troubling semantic doctrines have no special status: they are simply some among many empirical hypotheses about the world we live in. As such, they do not compare in evidential support with moral realism. Experience has taught us a great deal about what is good and right, but rather little about the language Page 3 of 15

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Moral Realism A Naturalistic Perspective * we use to talk of the good and the right. So semantics is just the wrong place to start the argument.5 I have been arguing about what moral realism is not, it is not a semantic doctrine. It is time to say more about what it is. This realism, like most others, has two dimensions, an existence dimension and an independence dimension (1984/1991b: ch. 2; also Ch. 2, part I, in the present volume). The rough idea of the independence dimension is that moral reality is mind independent. MR1 captures this in its talk of moral facts being “objective”. They are objective in that they are not constituted by our opinions, by our feelings, by our social conventions, by the synthesizing power of the mind, nor by our imposition of concepts, theories, or languages. All varieties of subjectivism and relativism reject the independence dimension of MR1. MR1 captures the existence dimension in its commitment to moral facts. And this is its problem. Must the moral realist really commit herself to facts, what Quine would regard as “creatures of darkness”? Surely not. Surely a person could be a moral realist whatever her position on the controversial matter of the ontological status of facts. Can we then take the talk of facts as a mere manner of speaking, convenient, but to be paraphrased away when the chips are down?6 We can indeed: MR2: There are people and acts that are objectively morally good, bad, honest, deceitful, kind, unkind, etc. (virtues and vices); acts that one objectively ought and ought not to perform (duties); people who are objectively morally entitled to privacy, to a say in their lives, etc. (rights). There is no relevant controversy over the existence of people and acts. So MR2 puts the controversy where it belongs: over whether some people are objectively honest, over whether some acts are objectively ones we ought to perform, and so on. Still, MR1’s talk of facts yields a very convenient (p.186) shorthand. So I shall continue to use that talk in the knowledge that it can always be paraphrased away along the lines of MR2 when necessary.

Mackie's famous “error theory” rejects the existence dimension of MR1 and MR2. He thinks that there are no moral facts. It would be nice if we could rest with MR2 as our account of moral realism. But we cannot: noncognitivism forces us to say more. I have argued that moral realism is not a semantic issue. This view seems to be threatened by noncognitivism which, as I have already remarked, seems to be both antirealist and semantic. An analogous threat comes from other forms of “nonfactualism”, for example, “projectivism”, and Simon Blackburn's “quasi‐ realism” (1984, 1993a, b). Indeed, Sayre‐McCord and many others clearly take the apparently semantic nature of noncognitivism as supporting their semantic Page 4 of 15

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Moral Realism A Naturalistic Perspective * view of the realism issue. I have discussed this matter at length elsewhere (1996b, which is Ch. 7 in the present volume; 1997a: 307–20). I summarize. The usual characterizations of noncognitivism are along the following lines. It is said to be the view that moral predicates do not denote, correspond to, etc., properties.7 Or it is the view that the indicative sentences in moral discourse are not assertions or statements,8 are not factual or descriptive,9 are not truth‐ conditional,10 and do not correspond to facts.11 Rather, those sentences have other functions like expressing attitudes or emotions, or prescribing norms or rules;12 for example, in the crudest version, to say “x is good” is to say “Hooray for x!” and to say “x is bad” is to say “Boo to x!” So, at first sight, noncognitivism does seem to be a semantic doctrine, a doctrine about what sentences mean and predicates refer to. Yet, when we look a little closer, we see that a certain metaphysical doctrine is implicit in the answers, the view that there are no moral properties13 or facts.14 Indeed, it would be surprising if noncognitivism did not have an implicit (p.187) metaphysics. For, the main motivation for noncognitivism's special treatment of moral language must come from the view that moral reality is somehow problematic or defective. Why else would you treat moral language differently, claiming that moral predicates do not refer and that what seem to be assertions are not really? Despite the focus on semantics in describing noncognitivism, at bottom it must be a metaphysical doctrine. What doctrine? Unfortunately, the lack of attention to its metaphysics has led to a very unsatisfactory answer: that there are no moral properties or facts. Our discussion of MR1 indicates the problem. The general issue of realism about properties or facts must be independent of the issue of moral realism. Thus it surely has to be possible for a nominalist who denies properties and facts altogether to be a moral realist. We solved this problem with MR2 by taking talk of facts as convenient shorthand for talk of people and acts. But this solution does not work here: noncognitivists are likely to accept MR2, reinterpreting its language so that it involves no commitment to a moral reality. They are likely to agree, say, that some people are objectively honest but interpret this as simply an expression of emotion, or whatever. Noncognitivists seem to talk like moral realists. MR2 does not take adequate notice of noncognitivism. We need a characterization of moral realism that makes the underlying metaphysics of noncognitivism antirealist. This characterization is not easy to find. We need (i) to find some language that is not just apparently factual but is treated by the noncognitivist as really factual; (ii) to examine noncognitivist statements using that language to find ones that disagree with the statements of moral realists. I argue that we should see noncognitivists as denying: (1) realist explanations of the nature of moral reality—for example, in virtue of what Alice is honest; and (2) realist claims about the causal role of that reality—for example, that it was because Hitler was Page 5 of 15

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Moral Realism A Naturalistic Perspective * depraved that millions died. This yields my final attempt at characterizing moral realism: MR3: There are people and acts that are objectively morally good, bad, honest, deceitful, kind, unkind etc (virtues and vices); acts that one objectively ought and ought not to perform (duties); people who are objectively morally entitled to privacy, to a say in their lives, etc. (rights). That this is so is open to explanation and plays a role in causal explanations.15 (p.188) Or, returning to our convenient manner of speaking, we can say briefly: there are objective moral facts that are open to explanation and play roles in causal explanations.

2. Why Believe it? Why should we believe moral realism? It is prima facie plausible. It is a central part of the folk view of the world and, until recently (before the postmodernist plague), the social science view. Moral explanations seem to work: explanations adverting to moral facts seem to be successful; the cruelty of one person is as explanatory of behavior as is the cleverness of another. Moral realism makes sense of moral argument and moral disagreement. In light of all this, we should give it up only in the face of powerful arguments. I shall look critically at several arguments against moral realism in the next section. The most worrying argument I shall consider in the last section. This is the argument that moral realism cannot be fitted into a naturalistic world view. How do moral facts relate to facts about people, society, and the world in general? The answer must be that moral facts are part of the natural world. And this amounts to claiming that these facts must ultimately depend on the facts of physics, just as do the facts of chemistry, biology, and psychology. The demand here is not for some crude reduction. The idea is rather that there is a hierarchy of “levels” of facts, each to a degree autonomous, and yet each supervening on a “lower” level until we reach physics. The naturalistic project is then to show that moral facts supervene somehow on psychological and social facts, particularly on the psychological and social facts exemplified in humans and their societies. There is no need to give a priori naturalistic “definitions” of moral terms, defining ‘x is good’ as ‘x is N’, thus committing “the naturalistic fallacy”. So there is no worry with Moore's open question: “Is being N good?” The thesis that a moral fact supervenes (p.189) on certain nonmoral facts will not be knowable a priori; it is an empirical thesis. So, the Naturalistic Project is not only metaphysically naturalistic but also epistemologically naturalistic. Finally, one very good reason for moral realism is the implausibility of all alternatives. The error theory has the problem of all eliminativism: denying a commonly accepted reality. Such a denial is surely sometimes right—for Page 6 of 15

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Moral Realism A Naturalistic Perspective * example, atheism—but it is usually difficult to support and is in the case of morals. All versions of subjectivism and relativism have well‐known difficulties in accounting for central features of moral life, moral disagreement, and argument. Noncognitivism cannot account for moral explanations and has not yet succeeded in giving a convincing semantics for complex sentences with moral parts, particularly for conditionals (“If breaking promises is wrong then Fred should have done A”) In sum, moral realism is the best theory in town.

3. Arguments Against The Argument from Queerness

Perhaps the most popular argument against naturalistic moral realism arises from the idea that moral judgments are “partly prescriptive, or directive or action‐guiding” “just knowing [moral realities] or ‘seeing’ them will not merely tell men what to do but will ensure that they do it, overruling any contrary inclinations” (Mackie 1988: 101). Mackie summarizes Plato's view as follows: “The Form of the Good is such that knowledge of it provides the knower with both a direction and an overriding motive” “the end has to‐be‐pursuedness built into it” (p. 112). Similar views are to be found in Kant and Sidgwick. The argument against moral realism is then that any fact tied in this way to actions is, as Mackie nicely puts it, “queer”. How could this sort of queer fact be accommodated by naturalism? In responding we need to distinguish two claims: (i) the claim that awareness of the fact necessarily gives a reason for action because the fact is prescriptive and normative; (ii) the claim that awareness of the fact necessarily motivates action, causing feelings and desires that are usually overriding, leading to an intention to act, so that the agent will normally act morally. Stephen Darwell has called this view judgment internalism. Railton describes it as follows: “a moral judgment J can sincerely be made or asserted only by someone who is (or believes himself to be) motivated to some extent either to act in accord with J or to feel guilty for failing to do so” (1996: 62). (p.190) Claim (i) is no problem for naturalism. What it describes is the categorical nature of moral reasons. “Why ought I to do what I ought to, do what is right, and advance the good?” is a question internal to morals, a moral question, and is answered within morals. Moral facts are intrinsically normative. There is no task of naturalizing their normativity over and above the task of naturalizing their nature. Claim (ii) ties moral facts essentially to moral feelings. If those facts were so tied, they would indeed be queer. But why believe they are? There is a correlation—an imperfect one—between awareness of a moral fact and having certain desires that move a person to act. But it is indeed bizarre to think that it is part of the very nature of the fact that it moves people to act. Rather, the correlation between awareness and desire to act is an independent Page 7 of 15

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Moral Realism A Naturalistic Perspective * psychological fact that we explain scientifically: presumably, this response to moral facts is partly innate, partly a result of education. Given the obvious advantages to the human species of this response it would not be surprising if it were partly an adaptation.16 And we are all familiar with the effort that good parents put into bringing up their children to be morally sensitive. So genetically normal well‐brought‐up people tend to respond in the appropriate way to moral facts. This independence of moral facts from moral feelings yields a much more plausible view of the phenomena of (a) immorality, (b) amorality, and (c) ordinary human moral fallibility, than the alternative view. Consider how the alternative view that awareness of moral facts is essentially motivating—call this “the queer view”—must deal with these phenomena. (a) It could say that the immoral are simply unaware of the facts that would motivate moral behavior. But the evidence of what immoral people say shows that this is not generally the case; think of many cases of dishonesty and theft. Where the immoral person is clearly aware of the moral facts, the queer view must hold that the person is emotionally torn, with an impulse to do good overwhelmed by some other desire. But this is not psychologically plausible as a generalization: much immorality is not accompanied by guilt or shame, for example. (b) The queer view does not have the option of seeing the amoral as emotionally torn, for the very essence of amorality is to be unconcerned with morals. So the view must be that the amoral are simply unaware of the moral facts. But again, the evidence of what people say goes against this. The main failing of the amoral is not that they are unaware of the moral but that they (p.191) are unmoved by it. (c) Finally, the phenomena of the immoral and amoral are simple extreme examples of the phenomena of human beings, aside from a few saints perhaps. Every now and then, most of us recognize something we ought to do whilst remaining sadly unmoved to do it: “I know I ought to . . . but . . .”. In contrast, the independence view can handle these cases with ease, looking to nature and nurture for psychological explanations of these various moral failings: “he was born evil”; “she wasn't properly brought up”; and so on. Richard Boyd has suggested, plausibly, that these failings are tied to a lack of sympathy (1988: 215). I would prefer to talk of empathy. A person who is unable to “put himself in another's shoes” is likely to lack moral feelings. Empathy makes “moral facts motivationally relevant” (p. 216). Some people, for whatever reason, lack this empathy and hence lack a moral sensibility. The Argument from Relativity

Another popular argument against moral realism attends to moral diversity and disagreement. Mackie talks of the “well‐known variation in moral codes from one society to another and from one period to another”; we think of slavery, cannibalism, and infanticide. There are differences within contemporary societies over sex, abortion, capital punishment, and vegetarianism. Mackie claims that these differences do not seem to come from “speculative inferences Page 8 of 15

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Moral Realism A Naturalistic Perspective * or explanatory hypotheses based on inadequate evidence”; rather they come from “adherence to and participation in different ways of life” (1988: 109). Before defending realism from this criticism, let me take the attack to its rivals. A question that has been insufficiently addressed is the following. How does antirealism explain the similarity in moral codes: think of the agreement about murder, theft, honesty, kindness, trustworthiness, incest, etc.? The realist thinks that this similarity arises from our responding to the objective moral facts.17 Antirealism must dismiss this. (a) It is tempting for the antirealist to appeal to the social efficaciousness of these codes. But this appeal is in danger of (p.192) collapsing into realism. (b) Presumably the antirealist will appeal to a common human nature, in particular, common moral feelings, leading to similar but not identical moral codes. That moral nature, not moral reality, is the cause of what is common in the codes. And the causes of differences are culturally explained differences in moral feelings. But this raises a deep question. How do we get from the feelings to the codes, from having certain feelings when observing theft to believing that theft is wrong? Why don't we just express our moral feelings by saying “Boo!” or “Hooray!”? Or perhaps state our feelings: “I feel pleased with you” or “What you are doing disgusts me”? Antirealists respond to this, drawing on Hume, by talking of “the projection or objectification of moral attitudes” (Mackie 1988: 114). But how convincing is this psychological story?18 In any case, several things can be said in defense of moral realism from the relativity argument. (i) Since moral facts supervene on other facts, a difference of moral opinion between two cultures may be explicable by a difference in those other facts; for example, by the difference between a hunter‐gatherer society and a capitalist one. (ii) Even where there are no differences in the facts on which morals supervene, there are several reasons why people might differ in their moral opinions. (a) The supervention is complicated and so moral facts may often be hard to discern. Discerning them is aided by education in theory and the arts, an education that may differ greatly between individuals and cultures. Thus it is helpful to be aware of social “experiments” with utopianism, communism, fascism, democracy, etc., which throw light on the human potential for good and evil. And it is helpful to be educated about the psychology of slaves, women, babies, workers, foreigners, animals, etc. (b) There may be socially induced distortions in views about the moral facts and the facts on which they supervene: ruling groups often have an interest in instilling false moral views about, for example, a neighboring culture. (c) Similarly, a person may have an interest in not acknowledging a moral fact leading to what Judith Thomson calls “walling off” (Harman and Thomson 1996: 205). (iii) There may be indeterminacy arising from clashes between morally relevant facts. Similarly,

Page 9 of 15

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Moral Realism A Naturalistic Perspective * there may be indeterminacy in epistemology arising from clashes between epistemically relevant facts. This sort of indeterminacy yields “hard cases”.19 (p.193) The Argument from Explanation

Gilbert Harman has pointed out that “you need to make assumptions about certain physical facts to explain the occurrence of the observations that support a scientific theory” (1988: 121). He claims that the situation is different in ethics: there is no “obvious reason to assume anything about ‘moral facts’”; “the truth or falsity of the moral observation seems to be completely irrelevant to any reasonable explanation of why that observation was made”; there is no way the “actual rightness or wrongness of a given situation can have any effect on your perceptual apparatus” (p. 122). Nicholas Sturgeon has responded that, on the contrary, “Hitler's moral depravity . . . forms part of a reasonable explanation of why we believe he was depraved” (1988: 234). Unless you assume for some other reason that there are no moral facts, this seems to be a plausible and acceptable explanation. And moral facts explain things other than our moral beliefs: history and biography are full of explanations in terms of moral character. Consider Harman's own example: the wrongness of the hoodlums setting fire to the cat is not irrelevant to the explanation of your judgment. If, as the naturalist thinks, wrongness supervenes on some more basic nonmoral properties and the act had not been wrong, then the more basic properties would have been absent and you might well not have made your judgment. Of course, one could deny the supervention, but that is a different objection. It is important to see that Harman is not entitled to assume that there is no supervention. He does seem to assume that there is none in the following contrast with science: “the physicist thinks he understands how the passage of a proton through the apparatus might explain a vapor trial” (1986: 64); we do not have the analogous understanding in moral case. I take his point to be that we understand the mechanisms by which the proton acts on the observer, but not the mechanisms by which the moral facts do. But if moral properties supervene on more basic nonmoral ones, then those ones will supply the mechanisms. Epistemological Arguments

1. Harman starts his book with the question: “Can moral principles be tested and confirmed in the way scientific principles can?” (1988: 119). He thinks not. (p.194)

2. A little later he asks: “Can you ever perceive the rightness or wrongness of what he does?” (p. 120); for example, setting light to a cat. Again he thinks not. Related to this, Mackie claims that our awareness of moral facts “would have to be by some special faculty of moral perception or intuition”. This view was embraced, of course, by the intuitionists who Page 10 of 15

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Moral Realism A Naturalistic Perspective * are now rather unfashionable. Mackie goes on: “intuitionism makes unpalatably plain what other forms of objectivism wrap up” (1988: 111). Sturgeon's response to point 1 is good. He appeals to the Duhem–Quine thesis: scientific principles cannot be tested in isolation either. In both cases we can derive empirical consequences if we include background assumptions (1988: 231). Harman himself shows how we could test, “It is wrong to cause unnecessary pain to animals”:

For Alfred to hit his cat with stick would be for Alfred to cause unnecessary pain to an animal; Alfred will not do anything wrong. (1986: 59) So we predict we will not observe Alfred hitting the cat with a stick. If we do, some assumption will have to go. Consider also how “We live in a morally decent neighborhood” is tested by observing Harman's hoodlums burning the cat. Consider the many falsifications of “Hitler is morally admirable”. Consider the standard procedure in applied ethics where principles are tested with examples.

Concerning 2, there is no special problem about observing moral facts unless one assumes that they do not supervene on more basic facts, an assumption that would beg the question against the realist. We may straightforwardly observe the basic facts (people dying, living in squalor, etc), or we may observe the symptoms of the more basic facts (of pain, of intentions, of character, etc.). We have a rough idea of the way moral facts supervene on these; for example, on the pain of the burning cat, on the intentions of the hoodlums. In these respects the situation is similar to many cases of supervening facts. Thus, we feel entitled to talk of observing social facts on the strength of observing more basic facts on which they supervene. Of course the observations of moral facts are theory‐laden. But so are the observations of any fact. Any observation we make is influenced by background theory and opinion to some degree or other; for example, by the opinion that we are not under the influence of an hallucinatory drug. This theory‐ladenness is thoroughly established in philosophy of science and psychology. As Boyd says, moral intuitions are just like the “theory‐determined intuitions in science, (p. 195) which the scientific realist takes to be examples of epistemically reliable trained judgments” (1988: 200).20

Page 11 of 15

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Moral Realism A Naturalistic Perspective * 4. Fulfilling the Naturalistic Project I will conclude with some brief and inadequate suggestions for the naturalistic project, drawing on Boyd and Railton. These suggestions do not amount to much more than gestures at the start of the project. But, first, how worried should the moral realist be about the incomplete state of the project? Is this a sufficient reason for withholding belief in moral realism? As the circumstances stand, I think not. Of course were circumstances different, it might well be rational to withhold belief: (i) If there were other persuasive reasons against moral realism. But I have just argued that there are not. (ii) If moral attributions were not a success, if moral explanations did not work. But the explanations are a success. (iii) If there were persuasive alternatives to moral realism. But I have claimed that there are not. (iv) Finally, and most pertinent, if we had no idea how the naturalistic project might be fulfilled or, worse, good reason to think it could not be. But this is very far from the case. We already have a good idea of the psychological and social facts that are salient for morality. We know where to look for the facts that underlie morality. Withholding belief in moral realism whilst awaiting its naturalization would be no more appropriate than would have been withholding belief in Mendelian genetics whilst awaiting molecular genetics.

All this having been said, it is worrying that we have made so little progress on the naturalistic project. Still, we do have some interesting ideas for advancing the project. Consider Boyd's views. Boyd sets out to explain moral goodness. He roughly identifies a number of “human goods”. He sees these goods as “homeostatically clustered” in that they are united by causal mechanisms. Something is good to the extent that it fosters them (1988: 203). Railton starts by appealing to people's interests to explain an instrumental rationality from an individual's point of view: rational behavior furthers those (p. 196) interests. Then he develops this to yield a theory of moral norms: “rationality not from the point of view of any particular individual, but from what might be called a social point of view” (1986: 190). So he offers “an explanatory theory that uses the notion of what is more or less rational from a social point of view” (p. 196): “what is morally best . . . [is] what is instrumentally rational from a social point of view” (p. 200). In brief, Boyd appeals to human goods and Railton to human interests. These approaches are naturalistically acceptable and seem promising. I have argued for a nonsemantic, purely metaphysical, view of the nature of moral realism. I have given some reasons for believing it. I have rejected four

Page 12 of 15

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Moral Realism A Naturalistic Perspective * arguments against it. Finally, I have offered some brief thoughts on fulfilling the naturalistic project.21 Notes:

(*) First published in the Croatian Journal of Philosophy, 4 (Devitt 2002c). Reprinted with kind permission from Kruzak. (1) See also Boyd 1988: 182; Mason 1988: 413; Nagel 1986: 139. (2) Some other examples: Hesse 1967: 407; Hooker 1974: 409; Papineau 1979: 126; Ellis 1979: 28; Boyd 1984: 41–2; R. Miller 1987; Fales 1988: 253–4; Jennings 1989: 240; Matheson 1989; Kitcher 1993; J. R. Brown 1994. (3) For a less rough account of deflationary truth, see my 2001b, which is Ch. 8 in the present volume. (4) A particularly important consideration against the a priori, in my view (1996a: 2.2), is the lack of anything close to a satisfactory explanation of a nonempirical way of knowing. We are told what this way of knowing is not—it is not the empirical way of deriving knowledge from experience—but we are not told what it is. Rey 1998 and Field 1998 have a more tolerant view of the a priori. My 1998, which is Ch. 12 in the present volume, is a response. [2009 addition] See also Ch. 13. (5) It follows from the present responses that the sweeping dismissal of naturalistic moral realism by Terence Horgan and Mark Timmons (1990/1, 1992a, b) on the basis of semantic arguments is deeply misguided. We do not know nearly enough about the semantics of moral terms on Earth, let alone on their Moral Twin Earth, to put this sort of trust in such arguments. The objection is not, of course, to semantic arguments per se, but to ones that are supposed to establish metaphysical conclusions. [2009 addition] The Moorean and naturalistic responses to such arguments here parallel those in Ch. 3, secs. 4–6, and Ch. 5, sec. 6, of the present volume to epistemic and semantic arguments against realisms about the entities of common sense and science. (6) [2009 addition] See Ch. 1 of the present volume on ontological commitment and paraphrase. (7) See e.g. Ayer 1952: 89; Sayre‐McCord 1988c: 7; Boghossian 1990a: 157–61, 166; 1990b: 266. (8) See e.g. Ayer 1952: 103; Haldane and Wright 1993b: 11; Hale 1993: 337. (9) See e.g. Ayer 1952: 107; Wright 1988: 29; Sayre‐McCord 1988c: 4; Blackburn 1993a: 3, 60; Haldane and Wright 1993b: 11–12.

Page 13 of 15

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Moral Realism A Naturalistic Perspective * (10) See e.g. Ayer 1952: 103, 107; Sayre‐McCord 1988c: 5; Boghossian 1990a: 160–1, 164; 1990b: 266; Blackburn 1993a: 60; Hale 1993: 337, 340; Haldane and Wright 1993b: 11. (11) See e.g. Wright 1988: 29; Sayre‐McCord 1988b: pp. ix–x; Boghossian 1990a: 160; Haldane and Wright 1993b: 12. (12) See e.g. Ayer 1952: 103, 107; Sayre‐McCord 1988c: 4, 8; Boghossian 1990a: 160; Blackburn 1993a: 3, 60; 1993b: 365; Hale 1993: 337; Haldane and Wright 1993b: 11. Strictly speaking these accounts of noncognitivism need qualification because the sentences in question may be partly assertions, partly truth‐ conditional, and partly factual. We can ignore the qualification. (13) See e.g. Ayer 1952: 89; Boghossian 1990a: 157–9, 161–2; Blackburn 1993a: 3. (14) See particularly Wright 1988: 29–30; Sayre‐McCord 1988b: pp. ix–x, 4; Blackburn 1993a: 3, 52, 57; Hale 1993: 337; Railton 1993: 280. (15) Sadly, there is a flaw in this characterization. There are doubtless some atypical moral realists who reject MR3 because they join the noncognitivists in denying the need for an explanation of moral reality and in denying that this reality has any causal role: moral reality is inexplicable and epiphenomenal. Such a position is deeply antinaturalist, of course. It is also hard to motivate: why believe in a goodness that can do nothing and cannot be explained? Still, the position is possible. So far as I can see this atypical moral realism cannot be distinguished from noncognitivism at the metaphysical level. It differs only at the semantic level in having a standard semantics for ‘good’. So it differs in accepting MR2 without reinterpreting it to remove its commitment to a moral reality. The noncognitivist and the atypical moral realist agree that there is no moral reality with an explicable nature and a causal role. Despite this failure, the atypical realist holds that there is a moral reality: it is simply inexplicable and epiphenomenal. In contrast, the failure motivates the noncognitivist to reject moral reality altogether by revising the semantics for what would otherwise be straightforward statements of moral realism. (16) [2009 addition] From the realist perspective, the interesting recent work on “the evolution of morality”—see e.g. the papers in Sinnott‐Armstrong 2008a— should be seen as attempting to explain the evolution of moral feelings and perhaps also the evolution of sensitivity to moral facts. Clearly it could not explain moral facts. (17) [2009 addition] Hauser et al. ask: “[W]hy are some moral judgments relatively immune to cross‐cultural variation? Are certain principles and Page 14 of 15

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Moral Realism A Naturalistic Perspective * parameters universally expressed because they represent statistical regularities of the environment, social problems that have recurred over the millennia and thus been selected for due to their consistent and positive effects on survival and reproduction?” (2008: 135). Their speculation here is, of course, very congenial to the realist. They make this speculation in the course of promoting the idea that there is a moral faculty akin to a Chomskian language faculty. (I take a somewhat skeptical view of the language faculty in my 2006d.) If there is a moral faculty then such innate sensitivity to moral facts as we may have (n. 16) will reside there. This sensitivity, together with experiences of the moral facts, would lead to our moral beliefs. Alternatively, it might be thought that the beliefs themselves are innate and residing in the moral faculty. I very much doubt this alternative. Indeed, I very much doubt that any beliefs are innate (Ch. 13, sec. 1, in the present volume). (18) [2009 addition] Haidt and Bjorklund, in their recent Humean account, talk of a “tight link” between the initial “gut feeling” emotional response to a situation and the moral “belief” about it (2008: 186–8). One wonders what this tight link could be: the feeling seems to be one sort of thing, the belief, which involves the application of a moral concept, a very different one. I emphasize that I am not denying that moral judgments are typically accompanied by moral feelings but only that the feelings are in any way constitutive of the judgments. (19) [2009 addition] I do accept that fundamental moral disagreement counts against realism. So I favor what Doris and Plakias call “convergentist” rather than “divergentist” realism (2008: 304–6). They claim that “there is . . . empirical reason to believe that fundamental disagreement is a substantial feature of morality” (p. 327). The realist must hope to explain away this apparent disagreement along the lines indicated. (20) [2009 addition] And they are just like other intuitions in being subject to “framing effects” (Sinnott‐Armstrong 2008c). (21) The paper of which this chapter is a version was given at the Universidad Ricardo Palma (Peru) in June 2000, the Victoria University of Wellington (New Zealand) in Dec. 2000, the University of Rijeka (Croatia) in Sept. 2001, and the University of Sydney (Australia), Nov. 2001.

Access brought to you by:

Page 15 of 15

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Natural Kinds and Biological Realisms *

Putting Metaphysics First: Essays on Metaphysics and Epistemology Michael Devitt

Print publication date: 2009 Print ISBN-13: 9780199280803 Published to Oxford Scholarship Online: February 2010 DOI: 10.1093/acprof:oso/9780199280803.001.0001

Natural Kinds and Biological Realisms * Michael Devitt (Contributor Webpage)

DOI:10.1093/acprof:oso/9780199280803.003.0011

Abstract and Keywords This chapter examines some interesting ‘realism’ issues in biology, arguing that these are best seen as about which kinds are explanatorily significant, which ones are ‘natural kinds’. Seeing them as ‘realism’ issues has caused unclarity and confusion. The main issue discussed is that between Ereshefsky's ‘pluralistic antirealism’ and Kitcher's ‘pluralistic realism’ about species. The chapter argues that so far as ‘realism’ is concerned, these views are actually the same. Concerning the higher categories it finds no good reason for the view of Eldredge and Cracraft that the higher taxa ‘do not exist in the same sense as do species’ and doubts the cladistic view that only monophyletic groups are ‘real’, that is to say, natural kinds. However, the signs are that the higher categories are not natural kinds and hence that the Linnaean hierarchy should be abandoned. Keywords:   natural kinds, biological realisms, Ereshefsky, Kitcher, species, Eldredge, Cracraft, cladism, monophyletic, Linnaean hierarchy

The species category does not exist. (Ereshefsky 1998: 113) . . . taxa of higher rank than species do not exist in the same sense as do species (Eldredge and Cracraft 1980: 327) To the cladist true believer, there is no such thing as a reptile. Page 1 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Natural Kinds and Biological Realisms * (Sterelny and Griffiths 1999: 197)

1. Introduction There are a number of “realism” issues in biology, issues about what “exists”, what is “real”, what is “objective”.1 In general, realism issues tend to be confused2 and the biological ones are no exception. We shall see that the interesting “realism” issues in biology are best seen as ones over which kinds “carve nature at its joints”, which ones are “natural kinds”. And that seeing them as “realism” issues has caused unclarity and confusion.3 I shall start with the issue that arises out of the debate between “species monists” who think that there is just one good (p.198) “species concept”—one good account of what it is to be a species—and “species pluralists” who think that there are many. To that end, in section 2, I shall summarize the various species concepts. In section 3, I shall summarize the motivation for pluralism. In section 4, I shall describe realism in general, particularly “realism about the external world”. Against this background, I turn, in section 5, to the issue in question, focusing on the apparent clash between Marc Ereshefsky's “pluralistic antirealism” (1998) and Philip Kitcher's “pluralistic realism” (1984). In section 6, I shall consider some “realism” issues about genera and higher categories in the Linnaean hierarchy.

2. Species Concepts Species pluralism arises out of the controversy over which species concept is correct. “The species problem is one of the oldest controversies in natural history” (O'Hara 1993: 231); it is “one of the thorniest issues in theoretical biology” (Kitcher 2003: p. xii).4 There are around two dozen species concepts and “at least seven well‐accepted ones” (Ereshefsky 1998: 103). Samir Okasha (2002) places them in “four broad categories”: 1. Phenetic concepts. On this sort of view, organisms are grouped into species on the basis of overall similarity of phenotypic traits. This is thought by its proponents to have the advantage of being fully “operational”. Okasha says that phenetic concepts are “the least popular” (2002: 199) and this is hardly surprising because they arise from the “philosophical attitude . . . of empiricism” (Sokal and Crovello 1970: 29). “Phenetic taxonomists have often wanted to segregate taxonomy from theory” (Sterelny and Griffiths 1999: 196). 2. Biological Species concepts (BSC). The most famous example of BSC is due to Ernst Mayr. He defined species as “groups of interbreeding natural populations that are reproductively isolated from other such groups” (Mayr 1969: 26). Kim Sterelny and Paul Griffiths remark that “If the received view has a received species concept” it is BSC (1999: 188). 3. Ecological Niche concepts. According to these concepts, a species occupies a certain ecological niche. “A species is a lineage . . . which Page 2 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Natural Kinds and Biological Realisms * occupies an adaptive zone minimally different from that of any other lineage in its range and evolves separately from all lineages outside its range” (van (p.199) Valen 1976: 70). Okasha puts the view succinctly: species “exploit the same set of environmental resources and habitats” (2002: 200). 4. Phylogenetic‐Cladistic concepts (P‐CC). On this view we “identify species in terms of evolutionary history . . . [with] particular chunks of the geneological nexus. . . . Species come into existence when an existing lineage splits into two . . . and go extinct when the lineage divides, or when all members of the species die” (Okasha 2002: 200). Sterelny and Griffiths claim that “something like a consensus has emerged in favor of a cladistic conception of systematics” (1999: 194). The various species concepts are answers to what Mayr (1982: 253–4) calls the “species category” problem: they are telling us what it is it for a kind to be a species rather than a subspecies (variety), genus, or whatever. In claiming this I am not being controversial. However, my claim about the following question is controversial: Do these species concepts tell us anything about what Mayr calls the “species taxon” problem? Do they tell us about what it is for an organism to be a member of a kind that happens to be a species? The answer to the “first level” taxon problem tells us why Fido is a dog. An answer to the “second level” category problem tells us why dogs are a species but poodles aren't. Now it is common to think that the species concepts not only answer the category problem but also the taxon problem.5 I have argued elsewhere (2008b, which is Ch. 11 in the present volume) that this common thought is wrong. But the main point to make here is that theories about the species category are one thing, theories about species taxa, another. In particular, realism about the category is one thing, realism about taxa, another.6

3. Species Pluralism Now, as I have noted, controversy has raged over which of the many species concepts is right. In the face of this controversy some have argued that we should abandon the monist idea that just one concept is right and hence that there is just one species category. Rather we should adopt the pluralist (p.200) idea that many of the concepts are right and hence there are many species categories. According to Kitcher many concepts “can be motivated by their utility for pursuing a particular type of biological inquiry” (1984: 118). Kyle Stanford puts the point thus: “certain explanatory demands are inextricably bound to certain species concepts” (1995: 72). And there are many different, but equally legitimate, types of biological inquiry and explanatory demands: “we have independent and legitimate explanatory interests in biology which require distinct concepts of species” (p. 76).7

Page 3 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Natural Kinds and Biological Realisms * In brief, there are a range of cases that force our attention to “the diversity of biological interests” (Kitcher 1984: 125), motivating different ways to classify organisms into species. And these ways will often lead to different classifications. Thus Ereshefsky claims that the BSC interbreeding and the P‐CC phylogenetic approaches to species “carve the tree of life in different ways. Many interbreeding species fail to be phylogenetic species, and many phylogenenetic species fail to be interbreeding species” (1998: 105). I have doubts about the extreme form of pluralism urged by Kitcher, doubts that I will air in section 5. Still, the case for some form of pluralism strikes me as strong. But what bearing does this monism‐pluralism issue have on biological “realism”? To assess this, we need to be clear about what biological realism is. And to get clear about that it will help to start by considering realism in general.

4. Realism in General The background issue that is most relevant is often known as “realism about the external world”, concerned initially with the observable entities of commonsense, but spreading to scientific entities, both observable and unobservable. Let us attend only to scientific entities. What is realism about these entities? It has two dimensions, one committed to the existence of entities, the other to their mind‐independence. We can capture the doctrine well enough as follows: Realism: Entities of most scientific kinds exist mind‐independently. I have argued the case for this doctrine at length in Realism and Truth (1984/1991b) and subsequent papers (1999a; 2001a, which is Ch. 5 in the present volume; 2002a; 2005c, which is Ch. 4 in the present volume).

(p.201) This doctrine, Realism, needs to be kept quite distinct from another traditionally called “realism”. This other realism is about “universals”, abstract entities like kinds, properties, or sets. Insofar as this realism arises from the “one‐over‐many problem” I follow Quine in thinking it arises from a pseudo problem (1980, which is Ch. 1 in the present volume; Devitt and Sterelny 1999: 277–9). Furthermore, I lean toward the view that there are no good reasons for realism about universals—I lean toward nominalism—but I shall not try to argue the matter. Indeed, it seems to me best to discuss the issues of realism in biology without any commitment on this vexed, millennia‐old, metaphysical issue.8 I shall continue to talk of kinds as if they existed but remain neutral on whether this is a mere manner of speaking that can be paraphrased away or whether it amounts to a real commitment. One further aspect of realism debates looms large in biology. Philosophers have been concerned not only with whether the posits of science exist mind‐ independently but also with whether these posits are “appropriately special” rather than somewhat arbitrary. In particular there has been a concern about whether our scientific posits “carve nature at its joints”, about whether there is Page 4 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Natural Kinds and Biological Realisms * something in the nature of the world that, in some sense, determines our categorization of it. I take this to be a concern about whether the kind of entity posited by a theory plays a causally significant role, whether it is partly because an entity is of that kind that it has the characteristics and behavior that it has. Theories need to posit such so‐called “natural” kinds if the theories are to be genuinely explanatory. And Realists are likely to take it for granted that their paradigm entities—for example, cats and planets—are indeed of natural kinds.9 This admittedly vague idea of naturalness is what Kitcher has in mind in saying that “natural kinds are the sets that one picks out in giving explanations” (1984: 132 n. 16). Similarly, Richard Boyd says that “the naturalness (and the ‘reality’) of natural kinds consists solely in the contribution that reference to them makes to [the accommodation between conceptual and classificatory practices and causal structures]” (1999: 141). There are two things we should note about definitions along these lines. First, they are insufficient: being a motor car, a hammer, or a paperweight are causal‐explanatory kinds. We (p.202) need somehow to capture the idea that natural kinds are causal‐explanatory “in science”. Second, explanatory significance, hence naturalness, comes in degrees: positing some kinds may be very explanatory, positing others, only a little bit explanatory, positing others still, not explanatory at all. This provides one reason for not using the term ‘realism’ to label the issue of naturalness: existence, unlike naturalness, does not come in degrees. But a more important reason is that the requirement that a kind be natural is quite distinct from the requirement that entities of that kind exist objectively and mind‐ independently. As we shall see in the next section, it is easy to name kinds of entities that are more arbitrary than causal‐explanatory and yet those entities still exist mind‐independently. And if there were kinds of entities that were mind‐ dependent in, say, a Kantian way, those kinds could still be causal‐explanatory. So, I shall keep the issues distinct. But they have not been kept distinct in biology, with some unfortunate consequences that will soon become apparent. I shall start with Ereshefsky's antirealism about species and his apparent disagreement with Kitcher.

5. Ereshefsky versus Kitcher In section 2, we noted the importance of distinguishing realism issues at two levels to do with species: at the first level we are concerned with species taxa— for example, tigers—and at the second level we are concerned with the species category itself. Ereshefsky is concerned with the latter. He claims that “the species category does not exist” (1998: 113) but he “does not call into question the reality of those lineages we call ‘species' ” (p. 104). Ereshefsky's argument, the “Heterogeneity Argument”, starts from species pluralism: “a number of species concepts should be accepted as legitimate” (p. 103). He takes this pluralism to imply antirealism and thus sees himself as being Page 5 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Natural Kinds and Biological Realisms * at odds with pluralistic realists like Kitcher (1984) and John Dupré (1993). Ereshefsky rightly points out that “species pluralism implies that the world contains different types of species” (p. 111). He goes on: What do these different types of species have in common that renders them species? If species taxa lack a common unifying feature, then we have reason to doubt the existence of the species category. (p. 111) Ereshefsky finds a suggestion in the literature for what the common feature might be, a suggestion with two components: (i) a similarity in the “process that (p.203) renders species taxa cohesive entities; (ii) a similarity of “structure” (p. 111). He argues that neither component can do the job (pp. 111–13). He concludes that “what is left as the common feature of species taxa is the term ‘species' ” (p. 113).10 Hence he draws his conclusion that the species category does not exist.11

What could he mean by this conclusion? He certainly does not mean that tigers and the like do not exist: we have noted already that he does not reject first‐level taxa realism. And he seems to accept that there are “different types of species”, each one captured by a different species concept; for example, “interbreeding species”, “phylogenetic species”, and “ecological species” (p. 115) or, more briefly, “biospecies”, “phylospecies”, and “ecospecies” (p. 117). So he accepts that there exist organisms that are members of a biospecies, organisms that are members of a phylospecies, organisms that are members of an ecospecies, and perhaps organisms that are members of other types of species. But, then, since biospecies, phylospecies, ecospecies, and perhaps others are all types of species, all of these organisms are members of a species. Indeed to be a member of a species simply is to be a member of a biospecies, phylospecies, ecospecies, or perhaps other types of species. So what entity could Ereshefsky be denying the existence of? His response to a “realist rejoinder” points to the answer. In this response, he denies that there is any “distinctive commonality” among biospecies, phylospecies and ecospecies. He is not satisfied with a disjunctive species category of the sort just illustrated. “A disjunctive definition of the species category would not tell us why various taxa are species . . . disjunctive definitions lack ontological import” (p. 115). What he finds lacking in the species category is explanatory significance, or anything close to that; the category is too close to being arbitrary. But how could this lack be a lack of existence? I take it that Ereshefsky must be thinking of the species category as an abstract entity, a “universal”, and must be presupposing a sort of selective realism about such entities. Whereas an unselective realist is committed, roughly, to there being a universal for every predicate, a selective realist is committed to there being one for some but not all predicates.12 And Ereshefsky is committed to there (p.204) being one for a predicate where the objects it applies to share a distinctive commonality, but not for any other predicates. According to his pluralism,

Page 6 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Natural Kinds and Biological Realisms * species lack that commonality. So the species category, a universal, does not exist. Now, it would be better if we could capture Ereshefsky's position on biology without commitment to a heavy‐duty, highly controversial, metaphysical thesis of selective realism about universals. And we can. We can capture it as the rejection of the view that the species category is a “natural” kind. And the rejection is not based on denying the existence of anything but on denying the explanatory significance of kinds being species. To see that nonnaturalness does not arise from nonexistence, it is helpful to note that entities of many kinds exist, even exist mind‐independently, where their being members of those kinds is of little or no explanatory significance. Consider cousins, for example. These include first cousins, second cousins, fifth cousins three times removed, and so on. The explanatory significance of being a cousin is surely close to zero. And we get even closer to zero if we consider step‐ cousins. And there is nothing to stop us naming a totally arbitrary and nonnatural kind: thus we could call anything that is either an acid, a river, or a bachelor a “grugru”. Yet despite nonnaturalness and explanatory insignificance, cousins, step‐cousins, and grugrus really exist, and do so mind‐independently. So, the best metaphysical message to take from biological pluralism is not that the species category does not exist or that it is not “real”. Nor is the best message that species do not correspond “to something in the objective structure of nature” (Kitcher 1984: 128), are not “an objective feature of the living world” (Sterelny and Griffiths 1999: 180). For, according to pluralism to be a species is to be a biospecies, phylospecies, ecospecies, or perhaps other type of species and there is nothing subjective or otherwise mind‐dependent about that. To think clearly about realism issues it is vital to distinguish sharply two sorts of freedom, a freedom we have and a freedom we don't (1991b: 245). The freedom we do have is to choose to name any kind we like, whether for explanatory reasons, for frivolous reasons, or for no reason at all; naming kinds is a subjective matter. The freedom we do not have is to choose whether something is a member of a kind, whatever our reason for naming that kind in the first place; kind membership is an objective matter. We have chosen to name cats for very good explanatory reasons and grugrus for no good reason at all. But grugrus exist as objectively and mind‐independently as cats. And they existed long before I named them “grugrus”. My naming them “grugrus” didn't make them grugrus any more than people naming cats “cats” made them cats. It is common to talk as if, in doing science, we impose our concepts to “carve up reality”. But this is not literally so: we choose our concepts (p.205) in an attempt to discover the causally significant features of a nature that is already “carved up”. The

Page 7 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Natural Kinds and Biological Realisms * importance of distinguishing theory making from worldmaking could hardly be exaggerated.13 In sum, Ereshefsky is best seen as rejecting the view that the species category is a natural kind because that category is not explanatory. He is best seen as rejecting this because that is what his argument from pluralism supports without the baggage of selective realism about universals. I noted that Ereshefsky sees himself in disagreement with Kitcher and Dupré who, he claims, “suggest a realistic interpretation of species pluralism: various species concepts provide equally real classifications of the organic world” (p. 103). Is there an actual disagreement and if so what is it? I shall only consider Kitcher and I shall set aside any possible disagreement arising solely from Ereshefsky's selective realism about universals. Kitcher does, of course, call his position “pluralistic realism” but what exactly is “realist” about his position? First of all he sets aside as “trivially true” a realism about the species category (and taxa) that is like the doctrine I have called simply “Realism”, requiring only mind‐independent existence. He is interested in a stronger realism: Pluralistic realism rests on the idea that our objective interests may be diverse, that we may be objectively correct in pursuing biological inquiries which demand different forms of explanation, so that the patterning of nature generated in different areas of biology may cross‐classify the constituents of nature. (1984: 128) Kitcher clearly thinks that various species categories—biospecies, phylospecies, ecospecies, and the like—are explanatory and hence natural kinds. Setting aside for a moment a disagreement about the variety of species categories, I assume that Ereshefsky would agree (1998: 117). But, of course, the view that the various species categories are natural kinds is not the view that the species category itself is. The latter view requires that it be explanatory significant that some kinds are species. Yet, according to Ereshefsky's argument, pluralism has the consequence that the species category is disjunctive and not explanatory. So if Kitcher's pluralistic realism is about the species category itself, and not merely (p.206) about various species categories, then Ereshefsky is indeed in disagreement with him. Is Kitcher's realism about the species category itself? He certainly does not argue that the species category is explanatory. And the message I take from “the moral for philosophy of science” that he draws at the end of his paper is that he thinks that the category is not explanatory: ‘species' refers not to a natural kind but to a “heterogeneous collection” of natural kinds (1984: 129). If I am right about this then Kitcher's pluralistic “realism” is not, in this respect, different from Ereshefsky's pluralistic “antirealism” and so they seem not to be in disagreement after all.

My suggestion is that the appearance of disagreement here between Ereshefsky and Kitcher has arisen from confusingly describing an issue about natural kinds as an issue about realism. Page 8 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Natural Kinds and Biological Realisms * However, Ereshefsky and Kitcher do differ in their pluralisms and hence in the variety of species categories that they are prepared to be “realist” about. Kitcher's pluralism is radical. He follows Mayr in drawing a distinction—a very important one in my view (2008b, which is Ch. 11 in the present volume)— between two types of explanation in biology, ones that Kitcher calls “structural” and “historical”. The structural seek to “explain the properties of organisms by means of underlying structures and mechanisms”. The historical seek to “identify the evolutionary forces that have shaped the morphology, behavior, ecology, and distribution of past and present organisms” (1984: 121). Kitcher claims that these two types “generate different schemes for classifying organisms” (p. 122). He finds variations within the two types leading him to posit nine different taxonomies and hence nine distinct species categories (p. 124). Ereshefsky's pluralism is more conservative. He is prepared to accept only species categories that are justified by historically motivated taxonomies, for example, the categories of biospecies, phylospecies, and ecospecies. His form of pluralism assumes that all species taxa are genealogical entities. To assume otherwise places species outside of the domain of evolutionary biology. The explanatory backbone of evolutionary theory is the assumption that organisms are connected by geneology. (p. 107) However, this is not really an argument against Kitcher's radicalism. For, Ereshefsky is attending only to the explanatory needs of evolutionary biology, whereas Kitcher is emphasizing that there are other explanatory concerns in biology: there are the concerns of structural explanations. And Kitcher thinks that these explanations motivate taxonomies that are different from those motivated by historical explanations and hence different species categories.

(p.207) Is there an argument against Kitcher's radicalism? I think so. Suppose that we go along with him about the nine different taxonomies. What about the inference to nine different categories? If the taxa picked out by a certain taxonomy are to justify a certain species category, S, it has to be explanatorily significant that the taxa are S. Now given the role of species in theories of evolution it is plausible to think that a historically motivated taxonomy will justify a species category: as Ereshefsky puts it, “species taxa are the paradigmatic units in which descent with modification occurs” (p. 107).14 But why suppose that a structurally motivated taxonomy will justify a category? That is, even if structural explanations demand certain taxonomies, what significance is there for such explanations that some taxon is a species? Consider Kitcher's example: A biologist may be concerned to understand how, in a particular group of bivalve mollusks, the hinge always comes to a particular form. The explanation that is sought will describe the developmental process of hinge formation, tracing the final morphology to a sequence of tissue or cellular Page 9 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Natural Kinds and Biological Realisms * interactions, perhaps even identifying the stages in ontogeny at which different genes are expressed. (1984: 121) It is hard to see how it makes any difference to the structural explanation we seek whether that group of mollusks is a species, subspecies, genus, or whatever. Our interest in structural explanations may demand that we group those mollusks together in our taxonomy but it does not seem to demand that we assign them to any particular category. So, I am inclined to think that only historically motivated taxonomies can justify species categories and that Ereshefsky is right to be conservative.

I shall return to this disconnect between taxonomies and categories in discussing the higher categories in the next section.

6. The Higher Categories The issues that come up for species can come up for any of the higher categories (“ranks”) in the Linnaean hierarchy, in “the tree of life”: genera, families, orders, classes, phyla, and kingdoms. Thus, consider the kinds we call “genera”. At the first level, there is the Realist issue of whether Canis and other genera exist mind‐independently. Then there is the further issue of whether it is explanatorily significant that they are Canis, the issue of whether Canis is a natural kind. At the second level, we are concerned with the analogous (p.208) pair of issues about the category of being a genus; in particular, with whether it is explanatorily significant that Canis and the like are genera and hence with whether being a genus is a natural kind. Let us start with the first level. No issue came up at this level for species and that is surely appropriate.15 We should all hold the Realist view that dogs and the like exist mind‐independently. And being a dog is surely a natural kind if anything is: it features in historical explanations because being a dog is part of the evolutionary story; it features in structural explanations because a lot of the morphology, physiology, and behavior of Fido is explained by his being a dog.16 And one would have thought that there should similarly be no first‐level issue with the higher taxa. Once again we should be Realist and we should take the higher taxa as natural kinds. Thus, whether or not being a genus is a natural kind surely being a Canis is: it features in historical evolutionary explanations and structural explanations, just as being a dog (Canis familiaris) does. And so too, say, being a mammal. This is not to say that membership in a higher taxon is as explanatorily significant as species membership: explanatory significance comes in degrees, as we noted (sec. 4). Still, surely some of the morphology, physiology and behavior of Fido are explained by his being a Canis. In light of this, one wonders what to make of the view that “taxa of higher rank than species do not exist in the same sense as do species” (Eldredge and Cracraft 1980: 327). According to Ereshefsky this sort of view is part of “The Modern Synthesis”. He describes that Synthesis as holding: “Higher taxa . . . are merely artifacts of evolution at the species level. So while species are real and the ‘units of evolution,’ higher taxa are merely aggregates and ‘historical Page 10 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Natural Kinds and Biological Realisms * entities' ” (2001: 229). Ereshefsky himself rejects this view (1991: 381) as does James Mallet: “Whether species do have a greater ‘objective reality’ than lower or higher taxa is either wrong or at least debatable” (1995: 296). Clearly I think that Ereshefsky and Mallet are right in their rejection. I suspect that the rejected view arises out of a confusion of a Realism issue with a naturalness issue and/or a confusion of a first‐level issue with a second‐level one. Another first‐level issue has arisen from the fierce controversy between those favoring “phenetic”, “evolutionary”, and “cladistic” taxonomies. Sterelny and Griffith favor the cladistic taxonomy and take it to be the emerging consensus (p.209) (1999: 194). Yet cladistics involves the following surprising “metaphysical” first‐level claim: real groups in nature are all, and only, monophyletic groups . . . groups that consist of a species and all, and only, its descendents. To the cladist true believer, there is no such thing as a reptile. ‘Reptile’ does not name a real group, for there is no species that is ancestral to all the reptiles that is not also ancestor of the birds. (p. 197) Now, we should note first of all that the claim that “there is no such thing as a reptile” is a most unhappy way of putting the cladistic point. Of course there are reptiles because there are crocodiles, snakes, and lizards and to be a reptile is simply to be one of those. Reptiles exist, and do so mind‐independently, just as much as cousins, step‐ cousins, and grugus. And it is not much more apt to say that reptiles are not a “real group” in nature: there is nothing interestingly unreal about crocodiles, snakes, and lizards. I take it that the best way to put the cladistic point is to say that being a reptile is not an explanatory significant kind. In general, the cladist thinks that nonmonophyletic are not natural kinds. Once again, the nonnatural is being confusingly described as the nonreal.

Is this restriction to monophyletic groups appropriate? Although Sterelny and Griffiths favor cladism they are inclined to think that the restriction is too extreme: there may well be sensible evolutionary hypotheses about all the nonmarine mammals. The group is not a monophyletic clade, because there is no species ancestral to all the land‐breeding mammals that is not also ancestral to the whales. (p. 198)17 This is an evolutionary reason for abandoning strict monophyly. There are other explanatory concerns in biology that seem to demand this too. There are the sort of structural concerns that partly motivate the phenetic and evolutionary taxonomies. There seems no reason to think that only monophyletic classifications can serve those other concerns. Perhaps being a reptile is explanatorily significant in many cases.18 Doubtless some nonmonophyletic groups that biologists have posited do not “carve nature at its joints” but the cladists case that none do seems inadequate at best, reflecting attention only to historical explanations.19

Page 11 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Natural Kinds and Biological Realisms * (p.210) Turn now to the second level. We might expect taxonomic disputes to play a role here in two ways. (A) The pluralism about the species category that we have been discussing may have repercussions for higher categories (ranks). (B) There is the possibility that the just‐mentioned dispute between phenetic, evolutionary, and cladistic classifications is relevant. I shall be very brief about (A). If anything like Kitcher's or Ereshefsky's species pluralism is right then it seems that we would have to have a distinct tree of life for each type of species; the species of each type are the base taxa for a distinct tree. If so then the best we could hope for with the higher categories is that each type of a category—for example, each type of genus based on a type of species— is a natural kind. But, turning to (B), we find that this hope seems not to be realized. Consider (B). In discussing Kitcher's pluralistic realism, I noted that if the taxa picked out by a certain taxonomy are to justify a certain species category, S, it has to be explanatorily significant that the taxa are S. And, using Kitcher's mollusk example, it is hard to see how it makes any difference to structural explanations whether a group is a species, subspecies, genus, or whatever. This carries over to the higher categories: taxonomies may not yield explanatory categories in general. So even if we could establish that a certain taxonomy— phenetic, evolutionary, cladistic, or whatever—was theoretically sound, so that classifying organisms in that way served our explanatory purposes in biology, that alone would not justify the view that any category posited by the taxonomy was a natural kind. Just because a taxonomy is right to classify a group of organisms as Canis and a subgroup as Canis familiaris does not show that there is any explanatory significance in treating the former as a genus and the latter as a species. We still need to show that the category itself does explanatory work. Do the higher categories do any explanatory work? Cladists think not: taxonomic ranks make little sense. . . . they do not think there will be any robust answer to the questions when should we call a monophyletic group of species a genus? a family? an order? Only monophyletic groups should be called anything, for only they are well‐defined chunks of the tree. But only silence greets the question are the chimps plus humans a genus? (Sterelny and Griffiths 1999: 201) So, on the cladistic picture, all categories (ranks) above species must be abandoned. Ereshefsky agrees although, as we have noted, he abandons the (p.211) species category as well. He rightly points out that if a certain category is to be acceptable, the taxa of that category must be “comparable” and draws attention to reasons for thinking that this condition is not met (1999: 299). Brent Mishler claims that “practicing systematists know that groups given the same rank across biology are not comparable in any way” (1999: 310–11). In a lengthier critique of the Linnaean Page 12 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Natural Kinds and Biological Realisms * hierarchy, Ereshefsky mentions the drive to introduce more ranks, leading to a hierarchy in flux (2001: 215); and to disagreements about the rank of certain taxa (p. 226). The signs are that, although the higher categories may have some pragmatic value, they are doing no explanatory work: they are not natural kinds.

In sum, at the first level I have found nothing against the view that the higher taxa, even some that are not monophyletic, are natural kinds. The second level is different. The signs are that the higher categories are not natural kinds and so the Linnaean hierarchy must be abandoned. Finally, we should note that abandoning the Linnaean hierarchy is not abandoning a hierarchy altogether, it is not abandoning a tree of life. It is abandoning the labeling of categorical ranks in that tree (Ereshefsky 1999: 299; Mishler 1999: 311). But, in light of (A) and (B), we may have to accept that there is more than one correct uncategorized tree of life, each reflecting legitimate explanatory concerns.

7. Conclusions My aim in this paper has been to examine certain issues in biology to show that they are really issues over which kinds are causal‐explanatory and hence “carve nature at its joints”. The issues are over which kinds are “natural” not over the mind‐independent existence of anything. So, describing the issues as about “realism”, as about what “exists”, is “real”, or is “objective” has led to unclarity and confusion. Throughout my discussion I have emphasized the importance of distinguishing between first‐level issues about taxa and second‐level issues about categories. Species pluralism is the view that there are several equally good accounts of what it is to be a species. Ereshefsky presents his argument from species pluralism to antirealism as an argument against the existence of the species category. I have argued that it is better seen as an argument against the explanatory significance of that category, hence an argument against that category being a natural kind. Not surprisingly, Ereshefsky sees his “pluralistic antirealism” as opposed to Kitcher's “pluralistic realism”. Yet, on close inspection, the two positions are similar: they agree that the species category itself is not explanatory but that (p.212) various types of that category are explanatory. However, they differ in that Kitcher's pluralism is more radical: Kitcher thinks that structural explanations in biology justify some species categories whereas Ereshefsky thinks that only historical evolutionary explanations can do so. I presented an argument that Ereshefsky is right: even if structural explanations motivate taxonomies they do not seem to show that a species category plays an explanatory role. Finally, I considered antirealism about the higher categories. At the first level, despite the urging of cladists, there seems to be no good reason to suppose that only monophyletic higher taxa are natural kinds. However, at the second level, the signs are that the higher categories are not natural kinds: they seem to do no explanatory work. If they don't, then the Linnaean hierarchy must be Page 13 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Natural Kinds and Biological Realisms * abandoned. Furthermore, the case for various taxonomies suggests that we may have to accept more than one uncategorized tree of life. Notes:

(*) First published in Joseph Campbell, Michael O'Rourke, and Matthew Slater (eds.), Carving Nature at its Joints: Topics in Contemporary Philosophy, vol. 8 (Devitt 2009e). Reprinted with kind permission from MIT Press. (1) The chapter draws on Devitt 2009a. (2) [2009 addition] See Chs. 2 and 9 in the present volume particularly. (3) Elliott Sober (1980: 203) struggles with what Ernst Mayr means by real. David Hull draws attention to the unclarity of these “realism” issues (1999: 25– 6). (4) Although, interestingly enough, an issue that Darwin himself was skeptical about: he talks of “the vain search for the undiscovered and undiscoverable essence of the term species” (1859: 381). (5) Dupré 1981; Sterelny and Griffiths 1999; Wilson 1999b; Okasha 2002. (6) Mayr's distinction is established but it is often overlooked; e.g. in Dupré 1981; Griffiths 1999; Sterelny 1999. In his brief discussion of Stanford and Ereshefsky on realism, Kevin De Queiroz (1999: 74–5) does distinguish the taxa issue from the category issue but he does so against a background of a definition of realism that applies only to taxa. The definition also takes realism to be a commitment only to mind‐independent existence, ignoring the far greater importance of explanatory significance in this debate. (7) As Hull (1997) emphasizes, many biologists require that a species concept be not only theoretically explanatory but also easily applicable: it must be “operational”. Hull rightly points out that “the philosophical arguments against operationism are decisive” but then goes on to be surprisingly tolerant of an “operational criteria for theoretical concepts” (p. 371). (8) And even more from commitment on the equally vexed issue of the nature of properties (supposing there are any). (9) We should acknowledge the popular view that species are individuals rather than kinds (Ghiselin 1974; Hull 1978). On that view the species of tigers is an individual constituted by the fusion of all tigers and any particular tiger is a part of that individual rather than a member of the tiger kind. Adjusting my discussion of biological realisms to take account of this view would have no significant effect. So I shall continue to write as if species are kinds.

Page 14 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Natural Kinds and Biological Realisms * (10) Richard Boyd claims that the types of species are unified by “an especially close homeostatic relation between the classificatory practices” (1999: 171); see also Wilson 1999b: 203–4. (11) In another paper, Ereshefsky gives a related reason for doubting the existence of the species category: the failure of attempts to distinguish species from the higher taxa, in particular the failure to do so in terms of the processes of speciation and interbreeding (1999: 269). (12) Thus, David Armstrong (1978b) is a selective realist, holding that empty predicates, negative predicates, and, most pertinently, disjunctive predicates, have no corresponding universal. He thinks that some predicates apply to the world in virtue of many universals. Most importantly, he looks to science to tell us which properties there are. (13) It is easy even for staunch realists to slip into loose ways of talking that suggest worldmaking. Thus Kornblith says that when we “group objects together under a single heading on the basis of a number of easily observable characteristics . . . we thereby create a nominal kind” (1993: 41). But we don't! We create a concept that picks out a kind that may or may not be “real” in Locke's terms but which has its members independently of our creation. And Boyd, talking of kinds with nominal essences, says that their “boundaries” are “purely matters of convention” (1999: 142). But they aren't! Our naming a kind picked out by a certain set of descriptions is conventional but the boundary of the kind thus picked out is not. [2009 addition] For more on worldmaking see Ch. 6 in the present volume. For more on nominal and real essences see Ch. 11 n. 4. (14) Despite this, I do have my doubts that any species category is really explanatorily significant. (15) Stanford (1995) seems to disagree; see Ereshefsky 1998 and Devitt 2009a for criticisms. (16) It is perhaps worth mentioning that being a member of a certain sub‐ species or variety is also explanatorily significant: Fido's being a pitbull explains a lot about his morphology, physiology, and behavior; cf. Joel Cracraft's description of concern about “the ontological status of subspecies” (1983: 100). (17) Boyd is also skeptical of monophyly (1999: 182). (18) It is also surely explanatorily significant that something is a predator or a parasite. Consider this e.g.: “The Lotke‐Volterra equations . . . describe the interactions of predator and prey populations” (Sober 1980: 202). But these are not the sort of classifications that concern the cladists. (Thanks to Marc Ereshefsky.) Page 15 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Natural Kinds and Biological Realisms * (19) Mishler concludes his summary of the argument for monophyly with this remarkably inadequate claim: “Because the most effective and natural classification systems are those that “capture” the entities resulting from processes that generate the things being classified, the general biological classification system should be used to reflect the tree of life” (1999: 309–10). It is probably the case that the classification systems in all sciences “capture” entities resulting from processes—entities don't come from nothing!—but it doesn't follow that they should be classified to reflect those processes.

Access brought to you by:

Page 16 of 16

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism *

Putting Metaphysics First: Essays on Metaphysics and Epistemology Michael Devitt

Print publication date: 2009 Print ISBN-13: 9780199280803 Published to Oxford Scholarship Online: February 2010 DOI: 10.1093/acprof:oso/9780199280803.001.0001

Resurrecting Biological Essentialism * Michael Devitt (Contributor Webpage)

DOI:10.1093/acprof:oso/9780199280803.003.0012

Abstract and Keywords This chapter defends the doctrine that Linnaean taxa, including species, have essences that are, at least partly, underlying intrinsic, mostly genetic, properties. The consensus among philosophers of biology is that such essentialism is deeply wrong, indeed incompatible with Darwinism. The chapter argues that biological generalizations about the morphology, physiology, and behaviour of species require structural explanations that must advert to these essential properties. The objection that, according to current ‘species concepts’, species are relational is rejected. These concepts are primarily concerned with what it is for a kind to be a species and throw little light on the essentialist issue of what it is for an organism to be a member of a particular kind. Finally, the chapter argues that this essentialism can accommodate features of Darwinism associated with variation and change. Keywords:   biological essentialism, Linnaean taxa, species, intrinsic, relational, Darwinism, structural explanation, species concepts, variation, change

Essentialism about species is today a dead issue (Sober 1980: 249) Folk essentialism is both false and fundamentally inconsistent with the Darwinian view of species (Griffiths 2002: 72)

Page 1 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * 1. Introduction The idea that biological natural kinds, particularly a species like dogs, have intrinsic underlying natures is intuitively appealing. It has been shown to be widespread even among children (Keil 1989). It was endorsed by a great philosopher, Aristotle. Under the influence of the logical positivists, Popper (1950), Quine (1960), and others, it fell from philosophical favor in the twentieth century until revived by Hilary Putnam (1975), Saul Kripke (1980), and David Wiggins (1980). Many philosophers probably now take the view for granted. If so, they are right out of touch with biologists and, especially, philosophers of biology. For, the consensus among philosophers of biology, and a widespread view among biologists, is that this sort of “Aristotelian essentialism” is deeply wrong, reflecting “typological” thinking instead of the recommended “population” thinking (Sober 1980: 247–8). This essentialism is thought to arise from a naive and uninformed view of biology, indeed to be (p.214) incompatible with Darwinism.1 This view is nicely presented and argued for in a paper by Samir Okasha (2002). I shall take that as my main text. I shall defend intrinsic biological essentialism. I think that the children are right and the philosophers of biology wrong.2 I start by saying something about essentialism in general and about the essentialism I shall defend in particular. A property P is an essential property of being an F iff anything is an F partly in virtue of having P. A property P is the essence of being an F iff anything is an F in virtue of having P. The essence of being F is the sum of its essential properties. Essences can be fully intrinsic; for example, the essence of being gold is having atomic number 79. Essences can be partly intrinsic and partly extrinsic and relational3; for example, the essence of being a pencil is partly being an instrument for writing, which an object has in virtue of its relation to human intentions, and partly having the sort of physical constitution that distinguishes it from a pen, which an object has intrinsically. Finally, essences can be fully relational and extrinsic; being Australian is probably an example because it seems that anything—Rupert Murdoch, Phar Lap (a horse), the Sydney Opera House, a bottle of Penfolds' Grange, the expression “no worries mate”, and so on —can have the property provided it stands in the right relation to Australia.4 The doctrine I want to defend, which I shall call “Intrinsic Biological Essentialism”, abbreviated sometimes to “Essentialism”, is that Linnaean taxa (p.215) have essences that are, at least partly, intrinsic underlying properties. This calls for some clarification and comment. (i) By “Linnaean taxa” I mean kinds that are thought to fall under the biological categories in the Linnaean hierarchy: kingdoms, phyla, classes, orders, families, genera, species, and even subspecies (varieties).5 I do not mean kinds like those of predators or parasites. And I do not mean the categories themselves. Page 2 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * Essentialism is a thesis about what it is for an organism to be, say, a dog not a cat, not about what it is for, say, dogs to be a species not a genus. (This distinction will loom large in sections 5 to 9.) The focus of my discussion will be on species but, I emphasize, Essentialism covers kinds that fall under all the categories. (ii) I include the qualification “at least partly” because I shall not take issue with the consensus that a species is partly an historical entity.6 (iii) In sexual organisms the intrinsic underlying properties in question are to be found among the properties of zygotes; in asexual ones, among those of propagules and the like.7 For most organisms the essential intrinsic properties are probably largely, although not entirely, genetic. Sometimes those properties may not be genetic at all but in “the architecture of chromosomes”, “developmental programs”, or whatever (Kitcher 1984: 123).8 For convenience, I shall often write as if the essential intrinsic properties were simply genetic but I emphasize that my Essentialism is not committed to this. (iv) Intrinsic Biological Essentialism would certainly be opposed by the consensus because of its commitment to intrinsic essences. But the consensus should not be opposed to biological essentialism in general because, as I am understanding essentialism, the consensus is that species have essences but these are extrinsic or relational. And Kim Sterelny and Paul Griffiths, in their excellent introduction to the philosophy of biology, Sex and Death, are explicitly not opposed to this (p.216) sort of essentialism: “the essential properties that make a particular organism a platypus . . . are historical or relational” (1999: 186). Of course, the very term ‘essentialism’ has become so distasteful to biologists because of its association with Aristotelian metaphysics that a biologist would doubtless be reluctant to admit to any sort of essentialism. But the essentialism I have defined need not come with those Aristotelian trappings. Many philosophers would be similarly reluctant because the term ‘essentialism’ strikes them as quaintly old‐fashioned, scholastic, even unscientific. But such reluctance would be a merely verbal matter. The issue of essentialism would remain even if the term ‘essentialism’ were dropped. It is the issue of in virtue of what an organism is a member of a certain Linnaean taxon; the issue of what makes an organism a member of that taxon; the issue of the very nature of the taxon. I stick with ‘essentialism’ because it is the term that philosophers of biology use for the doctrine that they want to reject and I want to promote. Those who are offended by the term should replace it with one of the other ways of characterizing the issue. (v) There is some controversy over whether species are natural kinds or individuals. Michael Ghiselin (1974) and David Hull (1978) seem to see individualism as an antidote to essentialism.9 But, as Okasha points out, “the issues about essentialism . . . do not depend on which view of the ontological Page 3 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * status of species we favour” (2002: 193–4; see also Kitcher 1989: 137–40).10 Thus, if a species is an individual rather than a kind, our essentialism issue for species becomes that of saying in virtue of what organisms are parts of a certain species, for example, the species Canis familiaris. And the consensus answer should be that it is entirely in virtue of the organisms' historical or relational properties, entirely “because they are part of the genealogical nexus” (Hull 1978: 309), whereas my Essentialism's answer would be that it is, at least partly, in virtue of the organisms' intrinsic underlying properties.11 Indeed the essentialism issue can be posed “nominalistically” in a way that is noncommittal on the ontological status of species: In virtue of what is an organism, say, a Canis familiaris? And that is how I do pose the issue in section 5. I mean to be neutral on the ontological issue but for convenience will mostly talk of species as if they were kinds. (p.217) (vi) Essentialism is primarily concerned with the natures of the actual groups identified by the folk and biologists for explanatory purposes. This “descriptive” issue needs to be distinguished from the “normative” issue of the natures of the groups that we should identify for explanatory purposes. Clearly, we might not be doing what we should be doing. However, I shall write as if we are. If we are not, then my Essentialism should be taken to cover the groups that we should be identifying for explanatory purposes as well. In section 2, I give evidence that the consensus really is opposed to Intrinsic Biological Essentialism. In section 3, I argue for the doctrine: explanations in biology demand that there be essential intrinsic underlying properties. I turn then to objections. In section 4, I describe the standard relational views of species which, according to the consensus, make Essentialism untenable. In section 5, I emphasize a distinction which is crucial to showing that the consensus is wrong about this. The distinction is between two questions. (1) What is it to be a member of any group that happens to be a species? (2) What is it for a group to be a species? In sections 6 to 9, I argue that the relational views of species are, primarily at least, answers to question (2). Essentialism, in contrast, is an answer to (1). Indeed, these relational views can, mostly, be happily wedded to Essentialism. Even the influential phylogenetic‐cladistic view can be wedded if it loses some implausible features.12 How has the consensus got it so wrong? My tentative diagnosis is that the error has arisen from conflating questions (1) and (2), a conflation encouraged by some mistaken thoughts about conspecificity. Finally, in section 10, I accommodate some general features of Darwinianism, associated with variation and change, features that are thought to undermine Essentialism. The accommodation requires acceptance of some indeterminacy in what constitutes a biological kind. But we must all accept that, whatever our views of Essentialism.

Page 4 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * If the arguments in sections 4 to 10 are right, the stated objections to an essentialist doctrine like Intrinsic Biological Essentialism fail. Perhaps there are some unstated objections that would succeed. And perhaps these could provide the basis for showing that the arguments in section 3 in favor of Essentialism are inadequate. Given the strength and longevity of the consensus in biology against intrinsic essentialism, it seems reasonable to predict this. Still, it remains to be seen whether it is so. At the very least I hope to show that the case for the consensus needs to be made a lot better than it has been.

(p.218) 2. Evidence of the Consensus I have claimed that the consensus among philosophers of biology is that doctrines like Intrinsic Biological Essentialism are wrong. Among those philosophers, the claim hardly needs support because the consensus is so established. Still, among philosophers in general, the claim does need support because, influenced by Kripke and Putnam, many find the claim incredible and so think I must be struggling with a straw man. The epigraphs to this paper, drawn from the works of Elliott Sober and Paul Griffiths, two leading philosophers of biology, are some evidence that I am not. Here is some more. The consensus starts by denying that members of a species share a distinctive set of genetic properties. Thus, according to Okasha, virtually all philosophers of biology agree that . . . it simply is not true that the groups of organisms that working biologists treat as con‐specific share a set of common morphological, physiological or genetic traits which set them off from other species. (2002: 196) Clearly, if the members of a species do not share a distinctive set of genetic properties then those properties could not be essential properties of that species. Indeed, Okasha claims that “biologists and philosophers of biology typically regard essentialism about species as incompatible with modern Darwinian theory” (2002: 191). And John Dupré claims that “it is widely recognized that Darwin's theory of evolution rendered untenable the classical essentialist conception of species” (1999: 3). Alex Rosenberg says: “The proponents of contemporary species definitions are all agreed that species have no essence” (1985: 203). Mohan Matthen claims that “species . . . are associated with no nonrelational real essence” (1998: 115). Sober expresses this consensus as follows: “biologists do not think that species are defined in terms of phenotypic or genetic similarities”; tigers are “not defined by a set of traits” (1993: 148). Sterelny and Griffiths put the point bluntly: “no intrinsic genotypic or phenotypic property is essential to being a member of a species” (1999: 186). Ghiselin puts it even more bluntly: “That John Doe has a particular set of genes is about as relevant to his being a specimen of Homo sapiens L. as it is to his working for the manufacturers of Brand X” (1974: 283).

Finally, if the essence of a species is not in the least intrinsic then it must be entirely relational. I have already quoted Sterelny and Griffiths's claim to this effect about the platypus. And they think that nearly everyone agrees with them: there is “close to a consensus in thinking that species are identified by their Page 5 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * histories” (p. 8). Their view is endorsed by Okasha (2002: 202). (p.219) Sober declares that tigers are “historical entities” (1993: 148). “Two organisms are conspecific in virtue of their historical connection to each other, not in virtue of their similarity” (1993: 150). Similarly, Marc Ereshefsky, speaking for “Darwin, the founders of the Modern Synthesis, and most cladists” (2001: 209). Finally, Hull claims: “If species are interpreted as historical entities, then particular organisms belong in a particular species because they are part of that genealogical nexus, not because they possess any essential traits. No species has an essence in this sense” (1978: 313). Ruth Millikan says much the same (2000: 19). The consensus is broad but some are not part of it. Thus David B. Kitts and David J. Kitts (1979) urge an intrinsic essentialism like mine. According to Richard Boyd (1999) and Robert Wilson (1999b), species are “homeostatic cluster kinds” and I take it that they think that they have at least partly intrinsic essences.13 And Philip Kitcher has this to say: “I want to remain agnostic on the issue of whether any species taxon has a nontrivial essence” (1984: 132 n. 16). I take the opposition to Intrinsic Biological Essentialism to be established. It is now time to argue for the doctrine.

3. An Argument for Intrinsic Biological Essentialism I shall offer two reasons for believing Essentialism. The first is superficial but still, it seems to me, indicative of where the truth lies. Such essential properties seem to be part of what “genome projects” are discovering. The projects seem to be throwing light on the very nature of certain species. Thus the New York Times recently reported that researchers hope “to discover, from a three‐way comparison of chimp, human and Neanderthal DNA, which genes have made humans human” (Wade 2006: A14).14 Philosophers of biology disparage this common view (Sterelny and Griffiths 1999: 7; Okasha 2002: 197) but the view is certainly appealing. The second reason is deep and shows why the view is appealing. We group organisms together under what seem, at least, to be the names of species or (p. 220) other taxa and make generalizations about the morphology, physiology, and behavior of the members of these groups: about what they look like, about what they eat, about where they live, about what they prey on and are prey to, about their signals, about their mating habits, and so on. These generalizations are the stuff of popular nature programs and are to be found throughout the writings of biologists and philosophers of biology. For example, we are told that ivy plants grow toward the sunlight (Sober 1993: 6); that polar bears have white fur; that Indian rhinoceri have one horn and Africa rhinoceri have two (p. 21); that Hawaiian Drosophila “routinely form interspecific hybrids in the wild” (p. 156); that the Australasian bittern is superbly camouflaged (Sterelny and Griffiths 1999: 32); that “Major Mitchell cockatoos occasionally hybridize with Page 6 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * galahs” (p. 189); that “Australian trees . . . are not just drought‐proof; they are fireproof as well” (p. 203); that “magnetotactic bacteria . . . come equipped with little compasses called magnetosomes, which they use to navigate away from oxygen‐rich surface water because oxygen is toxic to them” (p. 209). Generalizations of this kind demand an explanation. Why are they so? Why, for example, is there this difference between the Indian and African rhinos? Such questions could, of course, be seeking an explanation of the evolutionary history that led to the generalization being true. Set that aside for a moment. The questions could also be seeking an explanation of what makes the generalization true. Regardless of the history of its coming to be true, in virtue of what is it now true? What are the mechanisms? The truth of these generalizations cannot be brute facts about the world and so must be explained. Explanations will make some appeal to the environment15 but they cannot appeal only to that. There has to be something about the very nature of the group—a group that appears to be a species or taxon of some other sort—that, given its environment, determines the truth of the generalization. That something is an intrinsic underlying, probably largely genetic, property that is part of the essence of the group. Indeed, what else could it be?16 Some intrinsic underlying property of each Indian rhino causes it, in its environment, to grow just one horn. A different such property of each African rhino causes it, in its environment, to grow two horns. The intrinsic difference explains the physiological difference. If we put together each intrinsic underlying property (p.221) that similarly explains a similar generalization about a species, then we have the intrinsic part of its essence.17 The generalizations we have been discussing reflect the fact that it is informative to know that an organism is a member of a certain species or other taxon: these classifications are “information stores” (Sterelny and Griffiths 1999: 195). But being a member of a certain taxon is more than informative, it is explanatory. Matthen points out that “many biologists seem committed to the idea that something is striped because it is a tiger” (1998: 115). And so they should be: the fact that an individual organism is a tiger, an Indian rhino, an ivy plant, or whatever, explains a whole lot about its morphology, physiology, and behavior. At first sight, the explanation of the animal's stripes may seem rather superficial, but it is not really. For, when biologists group organisms together under some name on the basis of observed similarities, they do so partly on the assumption that those similarities are to be explained by some intrinsic underlying nature of the group. It seems to me clear that this is their practice, whatever they say about essentialism.18 So the apparently superficial explanation points to the deep fact that there is something intrinsic, probably unknown, partly in virtue of which the animal is a tiger and which causes it to be striped. That something is an essential intrinsic property. The sum of those properties, together perhaps with some historical ones, constitute the essence of a tiger. Sober rightly insists that the essence of a species must explain why its members are the way they are. It must be “a causal mechanism that acts on Page 7 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * each member of the species, making it the kind of thing that it is” (1980: 250). That is exactly what this (partly) intrinsic essence is. I distinguished two sorts of explanation that might be sought in asking why members of a species have a certain property. In so doing I am following in the footsteps of Ernst Mayr (1961). He regards an explanation of the mechanisms within members of a species that make a generalization true (regardless of the history) as concerned with “proximate” causation and part of “functional biology”. In contrast an explanation of the evolutionary history that led to the mechanism being present in the members of a species is concerned with “ultimate” causation and is part of “evolutionary biology”. The use of “ultimate” to describe the latter explanation seems like a gratuitous put‐down of the former. Kitcher's even‐handed description in response to Mayr's (p.222) is prima facie more appropriate: “there are indeed two kinds of biological investigation that can be carried out relatively independently of one another, neither of which has priority over the other” (1984: 121). Adopting Kitcher's terminology, I shall call the former sort of explanation “structural” and the latter “historical”. Structural explanations, as he says, seek to “explain the properties of organisms by means of underlying structures and mechanisms”. He gives a nice example: A biologist may be concerned to understand how, in a particular group of bivalve mollusks, the hinge always comes to a particular form. The explanation that is sought will describe the developmental process of hinge formation, tracing the final morphology to a sequence of tissue or cellular interactions, perhaps even identifying the stages in ontogeny at which different genes are expressed. (Ibid.) He goes on to claim that “explanations of this type abound in biology: think of the mechanical accounts of normal (and abnormal) meiosis, of respiration and digestion, of details of physiological functioning in all kinds of plants and animals” (ibid.). And my point is that, at bottom, structural explanations will advert to essential intrinsic, probably largely genetic, properties.19 It is because the bivalve mollusks have a certain intrinsic underlying nature that the hinge takes that form. That is the deep explanation.20

This discussion generates a number of questions. Here are two: (I) “Surely any of the generalizations we have been discussing could have exceptions: a small mutation may lead to an organism that seems to be a member of a species and yet lacks the property attributed to the species by a generalization. So the generalizations do not seem to be law‐like. How does Intrinsic Biological Essentialism deal with that?”21 (II) “It is of course the case that the truth of any such generalization must be explained by an intrinsic, probably largely genetic, property, but why does that property have to be an essential property of the kind in question?”22 I shall consider these questions in section 10, along with others

Page 8 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * arising from Darwinian views of variation and change. I shall argue that Essentialism has an adequate answer to all these questions. Sober claims that “evolutionary theory has removed the need for providing species with constituent definitions” and hence with intrinsic essences (p.223) (1980: 255). I suspect that this sort of focus on evolution, hence on historical rather that structural explanations, has misled biologists and philosophers of biology about essentialism.23 This having been said, I suspect that even historical explanations demand a partly intrinsic essence; that, for example, the explanation of how polar bears came to be white will ultimately depend on essential intrinsic properties of polar bears and of their grizzly ancestors. But I shall not attempt to argue this. I have presented a positive argument for Intrinsic Biological Essentialism. We might sum it up: structural explanations in biology demand that kinds have essential intrinsic properties. That is my first main point in defense of Essentialism. In the rest of the chapter, I will develop the case for Essentialism in the course of responding to objections.

4. Relational Species Concepts I start with what is alleged to be the central objection to Intrinsic Biological Essentialism: according to nearly all current “species concepts”—theories about the nature of species—species are relational. Okasha expresses this consensus as follows: “On all modern species concepts (except the phenetic), the property in virtue of which a particular organism belongs to one species rather than another is a relational rather than an intrinsic property of the organism” (2002: 201). Despite the consensus that these species concepts make Essentialism untenable, the nature of biological species is, and always has been, an extremely controversial issue: “The species problem is one of the oldest controversies in natural history” (O'Hara 1993: 231); it is “one of the thorniest issues in theoretical biology” (Kitcher 2003: p. xii).24 There are around two dozen species concepts and “at least seven well‐accepted ones” (Ereshefsky 1998: 103). I shall follow Okasha in placing them in “four broad categories”. In this section I shall briefly describe these concepts. In the next, I shall draw a distinction which is crucial to showing, in sections 6, to 9, that the consensus is wrong. 1. Phenetic concepts. On this sort of view, organisms are grouped into species on the basis of overall similarity of phenotypic traits. This is thought by its proponents to have the advantage of being fully “operational”. (p.224) Okasha says that phenetic concepts are “the least popular” (2002: 199) and this is hardly surprising because they arise from the “philosophical attitude . . . of empiricism” (Sokal and Crovello 1970: 29). “Phenetic taxonomists have often wanted to segregate taxonomy from theory” (Sterelny and Griffiths 1999: 196).25 This category of Page 9 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * species concept is the only one of the four that is not in the least historical and relational. I shall set it aside. 2. Biological Species concepts (BSC). The most famous example of BSC is due to Mayr. He defined species as “groups of interbreeding natural populations that are reproductively isolated from other such groups” (Mayr 1969: 26). Sterelny and Griffiths remark that “If the received view has a received species concept” it is BSC (1999: 188).26 3. Ecological Niche concepts (ENC). According to ENC, a species occupies a certain ecological niche. “A species is a lineage . . . which occupies an adaptive zone minimally different from that of any other lineage in its range and evolves separately from all lineages outside its range” (van Valen 1976: 70). Okasha puts the view succinctly: species “exploit the same set of environmental resources and habitats” (2002: 200).27 4. Phylogenetic‐Cladistic concepts (P‐CC). On this view we “identify species in terms of evolutionary history . . . [with] particular chunks of the geneological nexus. . . . Species come into existence when an existing lineage splits into two . . . and go extinct when the lineage divides, or when all members of the species die” (Okasha 2002: 200). Sterelny and Griffiths claim that “something like a consensus has emerged in favor of a cladistic conception of systematics” (1999: 194). Nonetheless, it has some surprising features, as we shall see (sec. 9). But perhaps the most important feature of the P‐CC concept for the purposes of this paper is that it is, as everyone agrees, incomplete. It needs to be supplemented by a theory of speciation, a theory that explains when a lineage has split in two. For this, as Okasha says, P‐CC “will have to rely on a concept of one of the other types” (2002: 201).

(p.225) 5. A Crucial Distinction It is alleged that, according to each of these species concepts, except the phenetic which we are setting aside, species are relational. These concepts are thought, therefore, to show that the nature of a species could not be partly intrinsic and hence that doctrines like Intrinsic Biological Essentialism are false. In assessing this thought it is very important to distinguish the question that Essentialism is supposed to answer from another which it isn't. Let Fs be some group that has been classified for biological purposes under one of the taxa; for example, a group of poodles, dogs, or Canis. The question that Essentialism answers has many forms, as already indicated (sec. 1 (iv)): (1) In virtue of what is an organism an F? What makes an organism an F? What is the nature of being F? What is the essence of being F?

Page 10 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * This is a question about the properties of organisms. When it concerns Fs that form a species, Mayr calls it the species “taxon” problem (1982: 253–4). So let us generalize this as “the taxon problem”. It needs to be distinguished from the very different, “higher level”, problem about the properties of those properties:

(2) In virtue of what are Fs a subspecies, a species, a genus or etc.? What makes a group of Fs a subspecies, a species, a genus or etc.? What is the nature of being a subspecies, a species, a genus or etc.? What is the essence of being a subspecies, species, genus or etc.? When it concerns species, Mayr calls it the species “category” problem (ibid.). So let us generalize this as “the category problem”.

“What is it to be a poodle not a bulldog?” is an instance of the taxon problem (1), “What is it for poodles to be a subspecies not a species?” is an instance of the category problem (2). The distinction between the two problems may seem obvious and yet it is easily conflated by certain forms of words. In particular, consider the question, “What is a species?” or “What is the nature/definition of a species?” These questions are ambiguous. They could be asking what sort of a nature any group has that happens to be a species, an instance of the taxon problem (1). But they are more likely to be asking what it is for any group to be a species, an instance of the category problem (2). The distinction between the two problems is absolutely crucial to this paper. Which problem are the species concepts supposed to answer? According to (p. 226) Okasha, speaking for the consensus, they are at least supposed to answer problem (1) for species: they are supposed to show, as we noted, that “the property in virtue of which a particular organism belongs to one species rather than another is a relational rather than an intrinsic property of the organism” (2002: 201). Yet Ereshefksy, in his introduction to a collection that includes many classic papers on species concepts (1992a), says that “our concern is with a definition of the species category” rather than of the species taxon (1992b: p. xiv; see also Kitcher 1984: 120). And according to Sterelny and Griffiths (1999: 211) and Wilson (1999b: 191–2), the species concepts provide answers to both taxon and category problems. I shall argue that, on the contrary, the species concepts are primarily concerned with (2) and throw little light on (1).28 Yet Intrinsic Biological Essentialism is concerned with (1). So, the central objection to Essentialism, based on the species concepts, fails. This is my second main point in defense of Essentialism. In section 6, I shall show how much the species concepts bear on the category problem (2). In sections 7 and 8, I shall show how little they bear on the taxon problem (1).

6. Species Concepts and the Category Problem (2) The species concepts straightforwardly answer problem (2) for species: they tell us about “the species category” (Sterelny and Griffiths 1999: 184).29 Thus the popular biological species concept (BSC) tells us that a group is a species in Page 11 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * virtue of being an interbreeding natural population that is reproductively isolated from other such groups. And the ecological niche concept (ENC) tells us that a group is a species in virtue of being a lineage which occupies an adaptive zone minimally different from that of any other lineage in its range and evolving separately from all lineages outside its range. And both these answers do indeed entail that being a species is relational: a group is a species in virtue of its breeding or niche relations to other groups. The story for the influential phylogenetic‐cladistic concept (P‐CC) is a bit more complicated. As we have noted, the P‐CC account of species adverts to the splitting of a lineage and so needs to be supplemented with a theory (p.227) of splitting, a theory of speciation. Thus, according to P‐CC, a group of organisms constitute one species at time t1 and their descendents constitute two daughter species at t2. But what makes it the case that the descendents are members of the daughter species rather than the original species? An account of speciation will tell us. When faced with the need to supplement P‐CC it is customary to wave a hand toward other species concepts to provide the needed theory of speciation. “The biological species concept, perhaps supplemented by the ecological species concept or by something else, reemerges as an account of speciation” (Sterelny and Griffiths 1999: 192). Thus, if P‐CC is supplemented by BSC it will take a lineage to split when it yields two groups each of which is interbreeding but reproductively isolated from the other. And if it is supplemented by ENC, it will take a lineage to split when it yields two groups exploiting different sets of environmental resources and habitats. With some such supplement in mind, P‐ CC, just like the other species concepts, straightforwardly yields an answer to problem (2) for species. Thus, return to Okasha's statement of P‐CC: “species come into existence when an existing lineage splits into two . . . and go extinct when the lineage divides, or when all members of the species die” (Okasha 2002: 200). Supplemented by a theory of speciation, this tells us what it is for a group of organisms to be a species rather than, say, a subspecies or a genus. And it tells us that this is a relational matter. Not only do the species concepts straightforwardly yield answers to problem (2) for species, that seems to be what they are intended to do. Ever since Darwin, the species concepts have been tied closely to views of speciation and to distinguishing when two groups constitute subspecies of the one species and when they constitute two distinct species of a genus. They are concerned, for example, with whether the British red grouse (Lagopus lagopus scoticus) and its continental relative (L. l. lagopus) are separate species; and with whether the divergent forms of the bluegill sunfish (Lepomis macrochirus) constitute a single species. They are concerned with what distinguishes species from other taxa. And, we should note, they do not seem to be concerned with the taxon problem

Page 12 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * (1); with, for example, the nature of the British red grouse or the bluegill sunfish.30 (p.228) Consider also the earlier‐mentioned controversy over species (sec. 4). The arguments here are all over problem (2). Thus, alluding to the controversy, Ereshefsky says: “Biologists differ widely on how to define the species category . . . [on how] to provide the essential property of the species category— a property found in all and only species taxa” (1992b: pp. xiv–xv). The controversy has led some (Mishler and Donoghue 1982; Kitcher 1984; Sterelny and Griffiths 1999: 194–201) to the view that “there is no unique factor common to all species” (Ereshefsky 1999b: p. xv). This pluralism, as Robert Wilson notes, is “about the species category” (1999b: 192).31 Biologists are concerned with whether groups that we have picked out for biological purposes should be counted as a subspecies, a species, a genus, or whatever. And with whether various considerations do, and should, play a role in settling such issues. In response to these issues, and inspired by cartographic generalization, Robert O'Hara urges that we take up the perspective of “systematic generalization” and then “we will be better able, not to solve the species problem, but rather to get over it” (1993: 232) It is quite clear that what he thinks we will be able to get over is fussing about when to judge a group to be a species. And, once again, the issues are not over problem (1), not over what is it for an organism to be a member of a group that we have picked out for biological purposes, irrespective of whether that group is a subspecies, species, genus, or whatever. So the issues are largely orthogonal to Essentialism. Essentialism is concerned with the nature of a group whatever the category it falls under. I turn now to consider the bearing of species concepts on the taxon problem (1). In section 7, I will argue that BSC and ENC do not bear on this problem and so the consensus is wrong. In section 8, I will try to diagnose the error. In section 9, I will consider the bearing of P‐CC.

7. BSC, ENC, and the Taxon Problem (1) One reason for thinking that species concepts are not intended to answer the taxon problem (1) for species is that they quite obviously have nothing to say in answer to this problem for taxa other than species.32 Yet if they were answering (1) for species we would expect them to be like Essentialism in saying (p.229) something, at least, in answer to (1) for the other taxa, in saying something about what it is to be a member of a particular genus, for example. Despite the consensus, neither BSC nor ENC provides an answer to (1). Let Fs be a group of organisms that is a species according to BSC or ENC; for example, the group of tigers. What do BSC or ENC, as they stand, tell us about how to complete ‘Some organism is an F in virtue of . . . ’? Since Fs are a species, BSC or ENC obviously tell us that whatever the completion it must specify some property or other of an organism that is at least compatible with the organism Page 13 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * being a member of some group or other that, briefly, interbreeds or occupies a niche. But, beyond that, they tell us nothing at all! They don't tell us what property makes an organism a member of the group of Fs in particular. Indeed, they don't even tell us what sort of property that must be. As we have just noted, BSC and ENC tell us what it is for Fs to be a species rather than, say, a subspecies or genus, but they are silent on what it is for an organism to be an F, say, a tiger rather than a lion. Analogously, an account of what it is for a group of objects to be tools rather than, say, pets or toys would not tell us what it is for an object to be a hammer rather than, say, a saw. In brief, BSC and ENC, as they stand, say nothing about what identifies a particular species, hence nothing about what constitutes its essence. The consensus view expressed by Okasha is simply false: it is not the case that, according to BSC and ENC, “the property in virtue of which a particular organism belongs to one species rather than another is a relational rather than an intrinsic property of the organism” (2002: 201). These concepts, as they stand, say nothing about this matter and so need not be at odds with any doctrine that does. Indeed, they are not at odds with Intrinsic Biological Essentialism. Essentialism's answer to taxon problem (1) is that the group F is identified (partly at least) by certain intrinsic underlying properties: it is in virtue of having such properties that an organism is a tiger rather than a lion. This is compatible with the BSC view that what makes Fs a species rather than a subspecies or a genus is a matter of interbreeding, and with the ENC view that it is a matter of occupying a niche. Indeed, it is better than compatible: Essentialism partly explains why the members of a species have the characteristics which, according to BSC or ENC, make them a species: it is partly because those members have a certain essential intrinsic properties that, in the given environment, they interbreed and occupy a niche. Far from being undermined by these species concepts, Essentialism is complementary to them. BSC and ENC do not entail a relational answer to (1) despite the consensus that they do. But perhaps we can see them as associated with such an answer. We can indeed find signs of two such answers in the literature. But neither could be a serious rival to Essentialism's answer. (p.230) The first answer is the simple idea that what makes something an F is that its parents are Fs. This seems to be suggested by the following: the reference of an individual to a species is determined by its parentage, not by any morphological attribute (G. C. D. Griffiths 1974: 102) If we suppose that humans first appeared about ½ million years ago, Homo sapiens is the name for the group that descended from the original organisms. (Ruse 1987: 344)

Page 14 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * Hull quotes the Griffiths passage approvingly (1978: 305). He goes on to consider whether a human‐like organism made by a scientist would be a human and this leads him to modify the simple idea: what makes something human is “being born of human beings and/or mating with human beings” (p. 306; emphasis added). A difficulty with the simple idea is that it rules out speciation: all organisms will be conspecific with their ancestors, however distant. Set that aside until later (sec. 10). The idea is open to an obvious objection: it is not really an answer to (1). It tells us that an organism is an F if its parents are Fs. But what is it for them to be Fs? The idea does not solve our problem, it simply moves it back a generation. And Hull's modification of the simple idea is open to a similar objection. It tells us that an organism is an F if it mates with Fs. But what makes the organisms it mates with Fs?

The second relational answer might be considered an elaboration of the first. It gets its inspiration from the typical naming practices of biologists: “Biologists coin new species terms by providing a sample, called a ‘type specimen’ ” (LaPorte 2004: 5). Could we then identify a species by referring to its type specimen? So, what makes this interbreeding or niche‐occupying group Fs is that it contains a certain type specimen. This idea seems to be suggested by some other remarks of Hull: The taxonomist . . . selects a specimen, any specimen, and names it . . . A taxon has the name it has in virtue of the naming ceremony, not in virtue of any trait or traits it might have. (1978: 308) Any organism related to [the type specimen] in the appropriate ways belongs to its species, regardless of how aberrant the type specimen might turn out to be or how dissimilar other organisms may be. (pp. 311–12) And consider this (entertaining) proposal for using BSC for “taxonomic definitions”:

Specify some individual, say Brigham Young, as your reference point, and then members of the same taxon are potential and actual interbreeders . . . (Ruse 1987: 344)33 (p.231) Combining this idea with BSC or ENC seems to suggest that what makes an organism a lion is that it is part of an interbreeding or niche‐occupying group that contains a certain historically identified type specimen, say Leo; and what makes this other organism a tiger is that it is part of another interbreeding or niche‐occupying group that contains a certain other historically identified type specimen, say Benji.

This answer is transparently inadequate. Relating an organism to a type specimen may sometimes be a convenient way to tell what species the organism belongs to—for example, if the type specimen is held in some museum—but it should not be taken seriously as an account of what constitutes being a member of a species. The answer may be epistemically useful but it is metaphysically hopeless.34 Why? Briefly, because being an F—for example, being a lion or being a tiger—is an explanatory property, as we noted in discussing the generalizations in section 3.

Page 15 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * The hopeless answer itself immediately generates the demand for an explanation that it cannot possibly fulfill. Why can lions interbreed with Leo and not Benji? It is clearly no help to be told that that is what it is to be a lion. And this failure is just the tip of the iceberg. Consider the following questions (construed structurally not historically): Why do tigers have stripes? It is no help to be told that it is because they can interbreed with Benji. That does not tell us why any tiger, including Benji, has stripes. Why do polar bears have poor eyesight? Once again, their relation to some Ur‐bear gives no explanation just as the relation of pieces of gold to the stuff in Fort Knox gives no explanation of why they are malleable. And so on through indefinitely many structural questions about the morphology, physiology, and behavior of species. All these questions concern facts about species that could not be brute: the facts have to be explained. The suggested answers to (1) cannot provide adequate explanations. The moral of this discussion is that any adequate explanation cannot appeal only to relational properties of members of the species in question because those relations cannot bear the explanatory burden. An adequate explanation must appeal to intrinsic properties of the organisms. It is something about the intrinsic natures of lions, tigers, polar bears, and so on that provides the explanation (along with some environmental factors). It is worth noting that the main point of the argument is not restricted to biology. Suppose we ask: why do paperweights make good weapons? The answer is not that they are the same tool as a certain specimen paperweight (p.232) kept in some museum. The answer is that the nature of paperweights makes them suitably heavy and easily grasped. And if they had a different nature, say that of buttons, then they would not make good weapons.35 So, contrary to consensus opinion, BSC and ENC do not give relational accounts of species identity. Indeed they do not give any account of species identity. Furthermore, they cannot be happily wedded to a relational account because such accounts are explanatorily hopeless. They can, however, be happily wedded to Intrinsic Biological Essentialism because intrinsic underlying properties can bear the explanatory burden. We are left with a puzzle. BSC and ENC do not give a relational answer to the taxon problem (1) for species and yet the consensus is that they do. What has gone wrong? How can we diagnose the error?

8. The Conspecificity Route to Error about the Taxon Problem (1) The obvious answer to the diagnostic question is that the error has come from somehow conflating the problems (1) and (2) that Mayr distinguished.36 BSC and ENC do indeed offer relational accounts of what it is to be a species. But that is a very different matter from offering a relational account of what it is to be a member of a group that is a species.

Page 16 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * Sterelny and Griffiths are interesting on this score. They nicely distinguish the taxon problem (1) from the category problem (2) in the “Further Reading” that concludes a chapter discussing the species concepts (1999: 211). Yet, surprisingly, the preceding discussion itself does not distinguish the problems (pp. 184–94). Indeed, they themselves draw attention to this conflation! Their defense is that “an answer to the taxon problem should solve the category problem, and vice versa” (p. 211). This raises our puzzle in an acute form. Why would anyone think that an answer to the one problem would answer the other? The discussions in Sterelny and Griffiths, and in Okasha (2002), suggest that the route from a category answer to a taxon answer may be via an answer to the conspecificity problem, the problem of saying in virtue of what organisms are in the same species. This route is most explicit in Wilson's discussion (1999b). Wilson takes what are, in (p. 233) effect, the BSC and P‐CC answers to the species category problem to imply answers to the taxon problem: they “imply that the properties determining species membership for a given organism are not intrinsic properties of the organism at all, but depend on the relations the organism bears to other organisms” (p. 192). How so? Because they imply that “conspecificity is not determined by shared intrinsic properties, but by organisms' standing in certain relations to one another” (p. 193). So the idea is that (a) BSC and P‐CC imply a relational nonintrinsic answer to the conspecificity problem and (b) this implies a relational nonintrinsic answer to the taxon problem. And the problem with the idea is that, although (a) is clearly tempting, it is false. (b), however, is true. I shall start with it. The first thing to note in considering (b) is that an answer to the conspecificity problem does not alone provide an answer to the taxon problem. A conspecificity answer tells us what it is for two organisms to be members of the same species and hence what makes Leo not conspecific with Benji. A taxon answer tells us what it is to be a member of a particular species and hence what makes Leo a lion and Benji a tiger. These are two very different matters. An answer to the conspecificity problem does not answer the taxon problem because it does not determine which species conspecific organisms are members of; it does not determine species identity; it does not tell us that these conspecifics are lions, those, tigers. So, even if a species concept did answer the conspecificity problem, as (a) claims it does, more would still need to be done to answer the taxon problem. Still, the two problems are related in a way that sustains (b). Suppose that the answer to the taxon problem is that an organism is a member of species F in virtue of being Q; and the answer to the conspecificity problem is that two organisms are conspecific in virtue of being R‐related. Now, necessarily, if two organisms are both Q then they are both Fs and hence conspecific. So the fact that they are both Q must determine, in a very strong way, that they are R‐related and hence conspecific: R must be the relation of sharing a property of Page 17 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * the Q sort, whatever that sort may be. Now suppose, as Intrinsic Biological Essentialism does, that Q is a partly intrinsic property of Fs, then R must be partly the relation of sharing that sort of intrinsic property. So if a species concept did imply a nonintrinsic answer to the conspecificity problem, as (a) claims it does, that would indeed count against Essentialism, as (b) claims. Turn now to (a) and consider BSC. I have allowed that (a) is tempting. Yet BSC, as it stands, says nothing at all about conspecificity. So why is (a) tempting? Because, as Wilson notes, citing Mayr, BSC implies something about conspecificity: it implies that “a given individual organism is conspecific with (p. 234) organisms with which it can interbreed” (1999b: 192–3). And it is easy to think that this amounts to (a). But it doesn't, because BSC does not imply that organisms are conspecific in virtue of interbreeding. The crucial error is to suppose that it does imply this, to suppose, quoting Wilson again, that BSC implies that “conspecificity is . . . determined by . . . organisms' standing in certain relations to one another” (p. 193; emphasis added).37 The point is a bit subtle and so I shall provide more details. BSC, as it stands, straightforwardly provides an answer to the category problem. Now that answer implies that conspecific organisms are members of a group that is, as a matter of fact, an interbreeding (and reproductively isolated) group. For, to be conspecific is, by definition, to be members of a group that is a species and, according to BSC's category answer, what makes a group a species is being an interbreeding group. The category answer tells us that conspecific organisms are members of an interbreeding group but it does not tell us in virtue of what they are members of that group. Indeed, BSC's category answer is compatible with Intrinsic Biological Essentialism's answer to the conspecificity problem: it is compatible with the view that organisms are conspecific in virtue of sharing a certain intrinsic underlying property and, perhaps, a history. The compatibility is easy to see. Essentialism is motivated by the need to explain the observable properties of a group of organisms (sec. 3). These properties include, of course, the property of interbreeding. So, according to Essentialism, it is because the members of a species share the intrinsic underlying properties necessary to make them conspecific that, in the given environment, those members interbreed and hence have the property than makes them a species according to BSC.38 Now we could, of course, supplement BSC as it stands with a relational answer to the conspecificity problem: organisms are conspecific in virtue of being able to interbreed. But, first, this supplement is not entailed by the BSC answer to the category problem and gets no support from the considerations that motivate (p.235) that answer. Without further argument, the supplement is gratuitous. And, second, the supplement is a very bad answer to the conspecificity problem. It is a very bad answer because, as (b) shows, it is at odds with Intrinsic Biological Essentialism. (b) shows that a relational conspecificity answer entails Page 18 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * a relational taxon answer and our earlier argument shows that relational taxon answers are explanatorily hopeless. The supplement is not something that BSC should be saddled with. Although (a) is certainly tempting it remains puzzling that people would give in to the temptation: it remains puzzling that they would not have doubts about this conspecificity route to anti‐Essentialism. For, although the answer to the conspecificity problem that BSC is wrongly thought to give does entail that the answer to the taxon problem must be relational and nonintrinsic it does not give such an answer. And as soon as one tries to give one, it should becomes apparent how explanatorily inadequate a relational answer must be. It least, this should be apparent if one keeps in mind the needs of structural explanations. So, it remains puzzling that a person who starts with the idea that BSC implies a relational conspecificity answer would not be led to contemplate one of two responses: abandoning BSC simply because of that implication; or, abandoning the idea that BSC has the implication.39 Our discussion of the puzzling conflation of the category problem with the taxon problem has focused on BSC. Yet the discussion applies just as much to ENC. This concludes our discussion of how little bearing BSC and ENC have on the taxon problem (1). We turn now to consider the bearing of P‐CC on that problem.

9. P‐CC and the Taxon Problem (1) At first sight P‐CC, unlike BSC and ENC, may seem to provide an answer to the taxon problem (1) for species. Let Fs be a group of organisms that is a species according to P‐CC. Whereas BSC and ENC told us little about how to complete ‘Something is an F in virtue of . . . ’, P‐CC may seem to tell us a lot. The idea would be that since Fs are a species, an organism is an F in virtue of being a member of a group with a particular “evolutionary history”; the species is “this particular chunk of the genealogical nexus”. What makes (p.236) this organism a lion not a tiger is that it is a member of a species having this particular place in “the branching tree‐of‐life” (Okasha 2002: 200). But this idea amounts to identifying a species simply by its relation to other species, to the species that preceded and succeeded it in its lineage: “You and I are members of Homo Sapiens, therefore, because we both belong to the segment of the genealogical nexus which originated in Africa some 300,000 [years] ago (on current estimates),

Figure 1.

Page 19 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * and which has not budded off any daughter species since that point” (pp. 200–1). But this answer to (1) is inadequate in just the same sort of way as our earlier second answer in discussing BSC and ENC: its relational identification of a species is explanatorily hopeless. To see this, it helps if we drop the actual names of species (like “Homo Sapiens”), which might provide some illicit information, and replace them with schematic names using which we can capture the relational information that is all we are entitled to on this P‐CC view. Thus suppose that species A splits into species B and C, then B splits into D and E and C splits into F and G. This is represented in the “tree‐ of‐life” as depicted in Figure 1.

What does the P‐CC view tell us about the nature of B? That B is descended from A and that B is distinct from C. And that's what P‐CC tells us that about the nature of C too. So that clearly does not distinguish B from C. No more does it distinguish D and E to know that each is descended from B, and F and G to know that each is descended from C. Furthermore, since P‐CC does not distinguish B from C it does nothing to distinguish D and E, descended from B, from F and G, descended from C. Suppose D were lions and G, tigers. Relational facts of the sort captured in this representation, which are the only facts that P‐CC allows to constitute the natures of species, would do nothing to distinguish lions from tigers, hence nothing to explain the morphological, physiological, and behavioral differences between them. Suppose that we want to explain why C has poor eyesight, all we could appeal to on this view is its (p.237) relation to A, F, and G. This is no more helpful in explaining the poor eyesight of C than was the relation of polar bears to some Ur‐bear in explaining theirs. The structural explanations we need must appeal to the intrinsic properties of a species. In sum, if P‐CC is taken to give an answer to the taxon problem (1) for species, its answer is a very poor one. But it would be more charitable to suppose that it does not really intend to give an answer. Indeed, why would anyone think that it does? We have already noted a possible explanation in section 8: the route from a category answer to a taxon answer may be via an answer to the conspecificity problem, the problem of saying in virtue of what organisms are conspecific. We have agreed with Wilson that a relational answer to the conspecificity problem implies a relational answer to the taxon problem; that is (b) in section 8. Wilson also thinks that P‐CC, like BSC, implies such an answer to the conspecificity problem: P‐CC implies that “conspecificity is determined by a shared pattern of ancestry and descent” or by something similar (1999b: 193); that is (a) in section 8. Now P‐CC does indeed seem to give such a relational answer to the conspecificity problem, just as it seems to give a relational answer to the taxon problem. But that conspecificity answer would be bad because that taxon answer would be explanatorily hopeless, as we have seen. So, if P‐CC really did involve such an answer, it should be abandoned. But it is more charitable to suppose that, despite appearances, P‐CC does not really propose an answer to the conspecificity problem.

Page 20 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * In section 8, I argued that the account given by Intrinsic Biological Essentialism of what it is for an organism to be a member of a particular species—an answer to (1)—can be wedded happily to the accounts that BSC and ENC give of what it is for a group to be a species—answers to (2). Can it also be wedded happily to the influential P‐CC? No. Adjustments would have to be made. There is no problem wedding Essentialism to P‐CC's view that species are historical entities because Essentialism is not committed to a fully intrinsic essence. The wedding would yield the view that a species is constituted partly by intrinsic, probably genetic, properties and partly by a particular history: an organism is a member of a certain species F in virtue of having a certain intrinsic properties and being part of a particular chunk of the geneological nexus. Those intrinsic properties are the ones that, together with environmental properties, explain the morphological, physiological, and behavioral properties of members of F. The chunk of the genealogical nexus in question should be the one that plays a role in explaining the evolution of the species. But this wedding of Intrinsic Biological Essentialism to P‐CC could not include two rather surprising features of P‐CC. (i) P‐CC does not allow anagenesis, (p. 238) the forming of a new species without any split in the old. No matter how dramatically a lineage changes it will not form a new species unless it splits (Hennig 1966; cf. Simpson 1945). So if Homo sapiens had evolved from protists without any splits, all the organisms in this lineage would be in the same species. Kitcher aptly notes that “this strikes many people as counterintuitive (even insane)” (1989: 151). Essentialism could not go along with this rejection of anagenesis (assuming, as we obviously do, that some groups covered by Essentialism are species).40 For, as Sterelny and Griffiths point out, the rejection of anagenesis has the consequence that “there is no upper limit to the amount of evolutionary change that can take place within one species” (1999: 7). So there would be no limit to genetic variation in a species and hence its essence could not consist of genetic properties.41 But the rejection of anagenesis is deeply implausible. Why? Because a doctrine that, in some possible world, places Homo sapiens and protists in the same species seems to have lost track of the explanatory significance of an organism being a Homo sapiens or a protist. (ii) Suppose that a species A splits off a side branch that forms a daughter species B but the population otherwise remains unchanged; that is to say, apart from the members of B, the descendents of the members of A do not differ phenotypically or genetically from their ancestors. So, had there not been the split forming B, those descendents would have all been members of A, on anyone's view of species. Yet P‐CC has the consequence that because B did come into existence those descendents form a new species C. Essentialism cannot accept this. According to Essentialism, if a population remains unchanged then its members must be conspecifics.42 But the P‐CC view is another implausible one, and for a closely related (p.239) reason. If a population remains Page 21 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * unchanged then its members should all be grouped together in explaining their common morphological, physiological, and behavioral properties. That's what the explanatory role of species seems to require.43 In sum, Essentialism could be wedded to P‐CC at the cost of dropping P‐CC's two surprising, and implausible, features. This concludes the discussion of the bearing of species concepts on the taxon problem (1) for species. Despite the consensus, BSC and ENC do not give a relational answer to that problem and, if P‐CC is taken to do so, its answer is a very poor one. My tentative diagnosis of the error in the consensus is that it arises from a conflation of the taxon problem with the category problem, a conflation encouraged by some mistaken thoughts about conspecificity. Perhaps also a focus on evolution has taken attention away from the needs of structural rather than historical explanations.

10. Variation and Change The consensus among philosophers of biology is that doctrines like Intrinsic Biological Essentialism are at odds with Darwinian evolutionary theory (Sober 1980; Griffiths 2002). We have been discussing what is thought to be the most important problem for such doctrines: contemporary historical views of species. We must now consider a number of other alleged problems centering on variation and change. Variation

(A) Sober claims that “no genotypic characteristic can be postulated as a species essence; the genetic variability found in sexual populations is prodigious” (1980: 272). Others write in the same vein. Thus Wilson rejects genetic essentialism because: The inherent biological variability or heterogeneity of species with respect to both morphology and genetic composition is, after all, a cornerstone of the idea of evolution by natural selection. (1999b: 190) (p.240) And Okasha claims:

Intra‐specific genetic variation is extremely wide—meiosis, genetic recombination and random mutation together ensure an almost unlimited variety in the range of possible genotypes that the members of a sexually reproducing species can exemplify. It simply is not true that there is some common genetic property which all members of a given species share and which all members of other species lack. (2002: 196) Surely, one thinks, this must be an exaggeration. Surely there are genetic properties that humans share and that say chimpanzees, let alone worms, lack. Indeed, even Mayr, no friend of essentialism, talks of “the historically evolved genetic program shared by all members of the species” (1963: 17).44 And Okasha goes on: “This is not to deny, of course, that there are important genetic similarities between members of a Page 22 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * single species . . . species taxa are distinguished by clusters of covarying [chromosomal and genetic] traits, not by shared essences” (2002: 197). Great! So the clusters are the essences! On the strength of these remarks, it seems as if the consensus should be simply that the crude idea that there is, say, “a tiger gene” is wrong. But to reject that crudity is not to reject the idea that a certain cluster or pattern of underlying, largely genetic, properties is common and peculiar to tigers.45 So my third main point in defense of Intrinsic Biological Essentialism is: an intrinsic essence does not have to be “neat and tidy”. And, because the intrinsic essence is identified by its causal work, we need not be concerned that the identification will be ad hoc: the essence of the Indian rhino is the underlying property that does, as a matter of fact, explain its single horn and other phenotypical features.

(B) Okasha emphasizes the importance of variation to natural selection: “Darwinianism leads us to expect variation with respect to organismic traits, morphological, physiological, behavioural and genetic. For genetically‐based phenotypic variation is essential to the operation of natural selection” (2002: 197). Sober thinks that this variation clashes with essentialism's commitment to the Aristotelian “Natural State Model”: essentialism takes the variation to be the result of “interfering forces” taking an organism away from its “natural state” (1980: 257–9); to be “the result of imperfect manifestations of the idea implicit in each species” (Mayr 1963: 16); to be “deviation” from an “ideal” (Griffiths 2002: 78–9). This contrasts with the Darwinian view: “Individual (p.241) differences are not the effects of interfering forces confounding the expression of a prototype; rather they are the causes of events that are absolutely central to the history of evolution” (Sober 1980: 264). Furthermore, “the Natural State Model presupposes that there is some phenotype which is the natural one which is independent of a choice of environment” (Sober 1980: 268). Essentialism need not go along with the teleological thinking of the Aristotelian Model and Intrinsic Biological Essentialism does not.46 That doctrine can and should accept the Darwinian view of variation: variation within a species is indeed to be expected; species are indeed, as Griffiths says, “pools of variation” (2002: 78). Essentialism is committed simply to the view that in the pool of variation among the members of a species there are shared intrinsic, probably largely genetic, properties. And Essentialism rejects the idea that it is “not natural” for a corn plant of a particular genotype to wither and die, owing to the absence of trace elements in the soil (cf. Sober 1980: 268). Gradual change

Hull puts the problem for essentialism thus: “according to evolutionary theory, species develop gradually, changing one into another. If species evolved so gradually, they cannot be delimited by means of a single property or set of properties” (1965: 203; see also Ruse 1987: 347). According to Griffiths essentialism “is precisely the ‘typological’ perspective on species that Darwin had to displace in order to establish the gradual transformation of one species into another” (2002: 77; see also Caplan 1980: 73).47 But Darwin didn't have to. Page 23 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * Suppose that S1 and S2 are distinct species, on everyone's view of species, and that S2 evolved from S1 by natural selection. Essentialism requires that there be an intrinsic essence G1 for S1 and G2 for S2. G1 and G2 will be different but will have a lot in common. This picture is quite compatible with the Darwinian view that the evolution of S2 is a gradual process of natural selection operating on genetic variation among the members of S1. Indeed, gradual change is obviously compatible with having essential intrinsic properties: rivers, mountains, continents, planets, and so on, are all the result of gradual change and yet all have partly intrinsic natures. Still, there may seem to be a worry, nicely expressed by Sober: evolution is a gradual process. If species A gradually evolves into species B, where in this lineage should one draw the line that marks where A ends and B begins? Any line (p.242) will be arbitrary. Essentialism, it is alleged, requires precise and nonarbitrary boundaries between natural kinds . . .  (1993: 147) This raises three issues: indeterminacy (or vagueness), arbitrariness, and “worldmaking”. Indeterminacy

Ereshefsky, paraphrasing Hull (1965), starkly puts the problem that indeterminacy is alleged to pose: “The boundaries of species are vague . . . there is no genetic or phenotypic trait that marks the boundary from one species to the next. Therefore no trait is essential for membership within a species” (1992c: 188–9). But this is a mistake: Essentialism does not require sharp boundaries between species. On the Essentialist picture, the evolution of S2 from S1 will involve a gradual process of moving from organisms that determinately have G1 to organisms that determinately have G2 via a whole lot of organisms that do not determinately have either. There is no fact of the matter about where precisely the line should be drawn between what constitutes G1 and what constitutes G2, hence no fact of the matter about where precisely to draw the line between being a member of S1 and being a member of S2. Essences are a bit indeterminate. There are two reasons not to be worried by this. First, indeterminacy is everywhere. It is indeterminate whether a certain x is a mountain, or a certain y, a planet,48 but this does not show that that there is no essence to being a mountain or a planet. Mount Everest has the somewhat indeterminate essence of being a mountain and yet is determinately a mountain; Mars has the somewhat indeterminate essence of being a planet and yet is determinately a planet. Second, there is just the same level of indeterminacy about species whatever one's (Darwinian) view of them and of essentialism, as indeed the passage from Sober indicates. For, everyone agrees that there comes a point where two organisms that have some common ancestor are nonetheless of Page 24 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * different species. Yet there is no determinate matter of fact about precisely where that point is. And it is very easy to spot the root of the problem. We are tempted to say that an offspring and its parent are conspecific whatever the mutation (as Okasha indicates; 2002: 197). But if we do say this, it is obvious that all organisms will (probably) come out conspecific. Biology faces a classic sorites problem. The indeterminacy that biology must learn to live with is no special problem (p.243) for Intrinsic Biological Essentialism.49 That is my fourth main point in the defense of the doctrine. We have been talking about the issue of indeterminacy in the world. This is likely to raise a worry about arbitrariness. There may indeed be some arbitrariness in the way we talk about the world. But that is not to say that there is arbitrariness in the world. Arbitrariness

Biologists choose, for various explanatory purposes, to introduce names for certain groups of organisms thought to be species. There could be some arbitrariness about what groups to choose. Let us start with the worst case. Suppose that we allow for anagenesis, as my Essentialism must (sec. 9):50 a new species can be formed without any split in a lineage. Suppose next that evolution were not only gradual but also steady: the morphological, physiological, behavioral and genetic properties of organisms in the lineage change at a steady rate. Essentialism alone does not rule this out. How then would we choose where to draw our (indeterminate) lines in naming the species of this lineage? Clearly, there would be a deal of arbitrariness about this choice.51 But we should not exaggerate how much. Our explanatory purposes in introducing a name for a species demand that we draw the lines around a group that is small enough to share a whole lot of important properties and large enough to yield broad generalizations. That is what is required for structural explanations. And, as G. G. Simpson points out, “such arbitrary subdivision does not necessarily produce taxa that are either ‘unreal’ or ‘unnatural’ ” (1961: 60–1). Furthermore, wherever we draw the lines in naming a group “F”, it is still the case that the intrinsic essence of being F, together with the environment, explains the morphological, physiological, and behavioral properties typical of Fs. Turn next to the best case. This is the situation if the hypothesis of punctuated equilibrium (Eldredge and Gould 1972) is right. On this hypothesis, evolution is far from steady. Species do not change much over most of their existence and then, in a relatively short period of time, either go extinct or evolve into daughter species. So, on my Essentialist picture, the need to explain the morphological, physiological, and behavioral properties of organisms would (p. 244) dictate that organisms in the period of stasis formed a species that should be named. We would draw our (indeterminate) lines in the period of rapid change. Our choice would hardly be arbitrary at all.

Page 25 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * The important point for our purposes is that, wherever the truth lies between the worst and the best cases, arbitrariness poses no threat to Essentialism. The groups we name will still have partly intrinsic essences. Indeed, arbitrariness is really a problem for the species category rather than the taxa. “Worldmaking”

We do have a choice about what groups of organisms to name “F”. It is sadly common to confuse this with a choice we certainly do not have: the choice about which things are F. This is, in effect, the distinction between making theories and making worlds, a distinction the importance of which can hardly be exaggerated.52 We name a group of organisms “F” for explanatory purposes and hence, even at worst, the choice of what group to name is mostly not arbitrary. But, however arbitrary it is, indeed even if it was totally arbitrary, we would not thereby make those organisms F. When biologists chose to apply the name ‘Drosophila melanogaster’ to a vast number of insects, they did not thereby make those insects Drosophila melanogaster. They always were Drosophila melanogaster and would have been even if there had been no biologists around to call them anything.53 It is common to talk as if, in doing science, we impose our concepts to “divide up reality”. But this is not literally so: we choose our concepts in an attempt to discover the causally significant features of a nature that is already “divided up”.54 Monsters

This discussion provides the wherewithal to deal with “monsters”, offspring that differ greatly from their parents. Monsters are thought to refute Essentialism because they lack what might plausibly be proposed as the intrinsic essences (p.245) of their parents' species. For example, Okasha claims that “if a member of the species produced an offspring which lacked one of the [essential] characteristics, say because of a mutation, it would be very likely to be classed as con‐specific with its parents” (2002: 197). Sterelny and Griffiths put the point more firmly: “No intrinsic genotypic or phenotypic property is essential to being a member of a species. . . . People born with the wrong number of chromosomes, eyes, or arms are still human beings. So the essential properties that make a particular organism a platypus, for example, are historical or relational” (1999: 186). Now Okasha is surely right that we would very likely classify any offspring as conspecific with its parents. But the sorites problem shows that we cannot always be right to do so, whatever we think of Essentialism: as Hull says, “Obviously . . . there must have been instances in which non‐horses (or borderline horses) gave rise to horses” (1978: 306). So what should the Essentialist say about monsters? One of two things. (a) If the mutations are gross enough, we should indeed say that the offspring is not of the same species as its parents. And that surely is what we would say, as monster movies sometimes illustrate. I doubt that we would even hesitate to say it of embryos that are so monstrous that they would not grow into viable organisms and are spontaneously aborted. (b) In other circumstances we should say that the status Page 26 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * of the offspring is indeterminate. Return to our schematic example of the evolution of S2 from S1. At the beginning of that process, there were organisms that determinately had G1 and so were determinately members of S1, and at the end, there were organisms that determinately had G2 and so were determinately members of S2. But in between there were organisms that did not determinately have G1 or determinately have G2 and so were not determinately members of S1 or determinately members of S2. All we can say is that the further an organism gets from determinately having G1 to determinately having G2, the further it gets from being determinately a member of S1. This is vague of course, but that's the way a lot of the world is, not just living things. Monsters are no special problem for Essentialism.55 (p.246) Laws

We are now in the position to respond to the questions raised in section 3 about my treatment of biological generalizations. Question (I) was: “Surely any such generalization could have exceptions: a small mutation may lead to an organism that seems to be a member of a species and yet lacks the property attributed to the species by a generalization. So the generalizations do not seem to be law‐ like. How does Intrinsic Biological Essentialism deal with that?” Essentialism surely does demand that these generalizations be law‐like rather than accidental. In a group of animals, it does not just happen to be the case that the members of a certain subgroup have one horn and the members of another, two. It is because the first subgroup are Indian rhinos and the latter, African rhinos; it is part of their very natures to have (in their environments) one horn and two horns respectively. But how can the generalizations be law‐like if there could be exceptions?56 There are several things we might say in answer. First it is common, perhaps even the rule, for laws in the special sciences to have exceptions: they hold only ceteris paribus. So why should this be a problem for biology in particular? Indeed, if Nancy Cartright (1983) is right the situation is not much different in physics. Second, statistical generalizations can be law‐like. Thus the claim that, say, 90 per cent of Fs are P can be law‐like: it can sustain the subjunctive conditional that if something were an F it would very likely be P.57 Finally, we can say that universal biological generalizations are indeed law‐like but that there is some indeterminacy about precisely which organisms they would cover. ‘All Fs are P’ may be a law in that anything that would be determinately F would be P but there might be some organisms that would not be determinately F or determinately not F and so there would be no determinate matter of fact about whether the law covered them. Note that this is not primarily an epistemological problem of telling what organisms the generalizations cover: it is primarily a metaphysical problem. Of course, even where there is a determinate matter of fact that generalizations cover certain organisms there can still be a problem discovering this; thus, many black birds (p.247) in Australia were

Page 27 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * determinately swans at a time when biologists believed that all swans were white.58 Essentialism?

Question (II) was: “It is of course the case that the truth of any such generalization must be explained by an intrinsic, probably largely genetic, property, but why does that property have to be an essential property of the kind in question?” Suppose that the generalization is ‘All Fs are P’ and that the explanatory intrinsic property is G. So it is agreed that, ultimately, it is because Fs have G that they have P. The question asks why we must take G to be an essential property of Fs. My answer rests on the just‐argued claim that the generalization is law‐like. So, anything that would be determinately an F would be P (in the appropriate environment). But now, in virtue of what is that the case? The answer is that anything that would be an F would have G. Indeed what other answer could we seriously entertain given that having G explains why all actual Fs are P? We have now answered question (II). For, if anything that would be an F would have G then having G is essential to being an F: that is what it is to be an essential property.

11. Conclusion I have proposed the doctrine, Intrinsic Biological Essentialism: Linnaean taxa have essences that are, at least partly, underlying, probably largely genetic, intrinsic properties. The consensus in biology and philosophy of biology is that any such essentialism is deeply mistaken. In section 2, I set out evidence that this is indeed the consensus. In section 3, I presented my central argument for Essentialism: the ubiquitous generalizations of biology need structural explanations that rest on essential intrinsic underlying properties of kinds. That was my first main point in defense of Essentialism. In section 4, I described current species concepts. The consensus view is that these make doctrines like Essentialism untenable because, according to these concepts, species are relational. In section 5, I emphasized a distinction (p.248) that is crucial to my defense of Essentialism from this consensus view. It is the distinction between two problems, a taxon problem (1) and a category problem (2). (1) What is the essence of being F (where Fs are a group under one of the biological taxa)? (2) What is the essence of being a subspecies, species, genus, or etc.? This distinction yields two ways to understand the question “What is a species?” The question could be asking about the nature of any group that happens to be a species or it could be asking about what it is to be a species. My second main point in defense of Essentialism, argued for in the next two sections, was that the (partly) relational species concepts are primarily concerned with (2) whereas Essentialism is concerned with (1).

Page 28 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * In section 6, I argued that, not only do the species concepts straightforwardly yield answers to question (2) for species, that seems to be what they are intended to do. In section 7, I argued that, contrary to the consensus, the biological species concept and the ecological niche concept do not answer (1) nor are they even associated with relational answers that are close to being explanatorily adequate. Indeed they can both be happily wedded to Essentialism's (partly) nonrelational answer. Where has the consensus gone wrong? My tentative diagnosis in section 8 was that the error arises from a conflation of the taxon problem with the category problem, a conflation encouraged by some mistaken thoughts about conspecificity. In section 9, I considered the influential phylogenetic‐cladistic concept (P‐CC). P‐ CC might be taken to give a relational answer to (1) as well as (2) but, if it were, its answer would be explanatorily inadequate. Can Essentialism's answer to (1) be wedded to P‐CC's answer to (2)? Essentialism can easily accommodate P‐CC's view that species are historical entities. However it cannot accommodate two of P‐CC's features: its rejection of anagenesis and its view that a species must go extinct when it has a daughter.59 But those features seem unwelcome anyway. Finally, in section 10, I argued that some general features of Darwinism do not undermine Essentialism. Variation within a species can be seen to be compatible with Essentialism once one realizes that an intrinsic essence does not have to be “neat and tidy”—my third main point in defense of Essentialism—and that Essentialism is not wedded to the Aristotelian “Natural State Model”. Essentialism can accept the gradual change of one species into another. Still, there are some concerns raised by the lack of sharp boundaries between species. First, Essentialism must accept a certain indeterminacy about species. But this is no worry because this indeterminacy has to be accepted whatever one's (p.249) (Darwinian) view of species and of essentialism; biology faces a sorites problem. That was my fourth main point in defense of Essentialism. Next, Essentialism is compatible with there being a certain amount of arbitrariness in choosing which groups of organisms to name as species. But this choice, however arbitrary, must not be confused with a choice we do not have: the choice to make things member of a group we have named. The fact of indeterminacy enables Essentialism to deal with the problem of monsters— organisms lacking what might plausibly be proposed as the intrinsic essences of their parent's species—and to maintain the law‐like status of biological generalizations despite apparent exceptions. I have dealt with the objections to Intrinsic Biological Essentialism that I have found in the literature. Perhaps there are other objections that would be more effective. Perhaps it can be shown that my argument in favor of Essentialism— the argument from explanation—is inadequate. Given the strength and longevity of the consensus against such an essentialist doctrine, it seems reasonable to predict this. Still, it remains to be seen whether it is so. At the very least I hope Page 29 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * to have shown that the case for the consensus needs to be made a great deal better than it has been. If the arguments of this chapter are good, the consensus relational view about particular species is quite mistaken. And many claims that biologists make day in and day out about the living world require species to have natures that they do not have according to this consensus.60 (p.250) Notes:

(*) First published in the Philosophy of Science, 75 (Devitt 2008). Reprinted with kind permission: © 2008 by the Philosophy of Science Association. (1) Michael Ruse places Kripke, Putnam, and Wiggins “somewhere to the right of Aristotle” and talks of them showing “an almost proud ignorance of the organic world” (1987: 358 n.). John Dupré argues that the views of Putnam and Kripke are fatally divergent from “some actual biological facts and theories” (1981: 66). [2009 addition] The standard story is that biology was in the grip of classical essentialism until saved by Darwin. Polly Winsor (2006) argues persuasively that this story is a fabrication of Ernst Mayr's. (2) This chapter was prompted by writing another one defending the thesis that the notion of rigidity we need for kind terms is one of rigid application not one of rigid designation (Devitt 2005d). The view that natural kind terms are rigid appliers has the metaphysical consequence that a member of a natural kind is essentially a member. This sort of “individual essentialism” needs to be distinguished from the “kind essentialism” that is the concern of the present paper. (3) Biological essentialism is usually taken to be concerned only with what is intrinsic (e.g. Mayr 1963: 16; Sober 1993: 146; Wilson 1999b: 188). This reflects the influence of Aristotle. I think it more helpful to define essentialism in a more general way so that issues come down to the sort of essence that a kind has. (4) Locke called an underlying intrinsic essence that is causally responsible for the observable properties of its kind a “real essence”. This is contrasted with a “nominal essence” which is picked out by reference‐determining descriptions associated with a kind term. So, having atomic number 79 is the real essence of gold and the essence of being Australian, whatever it may be, is merely nominal. Kripke and Putnam showed that natural kind terms like ‘gold’ are not associated with reference‐determining descriptions and so do not pick out nominal essences; they pick out real essences without describing them. This is not to say that a term could not pick out a nominal essence that is also real; indeed, ‘having atomic number 79’ is such a term (cf. Boyd 1999: 146).

Page 30 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * (5) I say “thought to fall” because I sympathize with the doubts of some about this hierarchy; see Ereshefsky 1999, 2001; Mishler 1999. [2009 addition] See Ch. 10, sec. 6, of the present volume. (6) However, I say that the essences are “at least, partly” intrinsic rather than simply “partly” because I do wonder whether all species are, or should be, partly historical. Citing the possibility of regularly produced hybrids like the lizard Cnemidophorus tesselatus, Philip Kitcher claims that “it is not necessary, and it may not even be true, that all species are historically connected” (1984: 117). (7) What I would like is a term for asexual organisms that is like ‘zygote’ for sexual ones in referring to the beginning of an organism. John Wilkins informs me that there is no one term for this. Others he mentions include ‘bud’ and ‘gemmule’. He has also drawn my attention to other uses of ‘propagule’. Thus, consider the following definition: In animals, the minimum number of individuals of a species capable of colonizing a new area. This may be fertilized eggs, a mated female, a single male and a single female, or a whole group of organisms depending upon the biological and behavioral requirements of the species. In plants, a propagule is whatever structure functions to reproduce the species: a seed, spore, stem or root cutting, etc. (http://www.radford.edu/ ∼swoodwar%20/CLASSES/GEOG235/glossary.html) (8) Webster and Goodwin (1996) promote the idea of “morphogenetic fields”. (9) “Individualism about species is an idea with close links to antiessentialism, both conceptually and historically” (P. Griffiths 1999: 211). (10) Richard Boyd goes so far as to say that the distinction between species being individuals or kinds “is almost just one of syntax” (1999: 164). (11) Also, we should note, if Canis familiaris is an individual, we can ask about its individual essence just as we can about that of any individual (n. 2). And the consensus answer should be that its essence is its being constituted by organisms that share historical or relational properties whereas my Essentialism's answer would be that those organisms must also share certain intrinsic underlying properties. (12) [2009 addition] Not so: the wedding is possible even with the implausible features; see nn. 41 and 42. (13) Hilary Kornblith favors the view that species are homeostatic cluster kinds, notes that the members of the cluster need not be intrinsic, but does not take a stand on whether any of them are (1993: 111, n. 10). Griffiths writes approvingly

Page 31 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * of the homeostatic cluster view but argues that species have purely historical essences (1999: 217–22). (14) Consider also this recent news report in the Scientific American online: “ ‘DNA barcodes are giving us a direct signal of where species boundaries lie,’ says Paul Herbert, an evolutionary biologist at the University of Guelph in Ontario and a progenitor of the genetic bar code effort” (Biello 2007). (15) The role of the environment is very obvious with plants. Thus the height of corn in a field depends on the temperature, the soil, and so on. (16) The point is not, of course, that the explanation of any generalization, even any biological one, demands an intrinsic property, just that the explanation of a generalization of the kind illustrated demands one. (17) So this intrinsic part is a real essence, in Lockean terms; see n. 4. (18) The following comment of Sterelny is interesting in this respect: “Some, perhaps most, evolutionary biologists take speciation to occur only when there have been intrinsic changes”. He finds this “puzzling for the view that species are historically defined entities is close to the consensus view in evolutionary biology”. He is inclined to blame the influence of the folk who, as we noted (sec. 1), tend to be intrinsic essentialists (1999: 130). I think that the biologists and the folk are, deep down, tuned into the demands of explanation. (19) Griffiths thinks not but his argument conflates structural explanations with historical explanations (1999: 210–11, 219–21). (20) Hull unfavorably contrasts “classificationists” seeking “the unit of identification” with “phylogeneticists” seeking “the unit of evolution” (1965: 204). I think that the classificationists should be seen as seeking units of structural explanation, a very worthwhile pursuit. (21) “Philosophers of biology have often noted that there seem to be no laws which apply to all and only members of a species taxon (Hull 1978; Rosenberg 1985)” (Okasha 2002: 209). (22) I am indebted to Peter Godfrey‐Smith for raising this question. (23) See e.g. Matthen 1998: 117–21; Griffiths 1999: 219–22; Millikan 2000: 18– 20. (24) Although, interestingly enough, an issue that Darwin himself was skeptical about: he talks of “the vain search for the undiscovered and undiscoverable essence of the term species” (1859: 381).

Page 32 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * (25) Sterelny and Griffith include under phenetic concepts those that define species in terms of genetic similarity (1999: 184). I clearly do not include these. I take the phenotype of an organism to be observable properties of it distinct from, but caused by, its genotype (along with the environment). (26) Popular as it is, BSC has been the subject of extensive criticism; see e.g. Sokal and Crovello 1970; van Valen 1976; Cracraft 1983; Sober 1993: 155–6; Kitcher 1984: 118–20; 1989: 141–5; Mallet 1995; Dupré 1999; Sterelny and Griffiths 1999: 186–90. Mallet claims provocatively that the BSC concept “owes nothing either to genetics or to Darwinism” (p. 295). (27) Ghiselin (1987: 374–8) has some severe criticisms of ENC. (28) Note that this is not the general claim that answers to (2) will throw little light on (1); indeed, see n. 30. It is a claim that the relational species concepts throw little light on (1). (29) We are setting aside the phenetic concept but it is interesting to note that it is not obvious that this concept does answer (2). After all, subspecies like poodles and genera like canis could equally be identified by an overall similarity of phenotypic traits. What sort of similarity marks out species in particular? (30) I take these examples from Mallet 1995. This chapter starts with a nice discussion of the history of species concepts that makes their concern with (2) very apparent. Mallet himself urges a genotypic cluster definition of species as an answer to (2): “we see two species rather than one if there are two identifiable genotypic clusters. These clusters are recognized by a deficit of intermediates” (p. 296). This view puts him right outside what the philosophers of biology consider the consensus. Yet, he claims, “many, perhaps most, systematicists are currently using the genotypic (or morphological) cluster definition” (p. 298). Whether or not Mallet's answer to (2) is correct, it implies an answer to (1) that clearly is as congenial as could be to Essentialism. (31) [2009 addition] Pluralism has led to a “realism” issue about the species category; see Ch. 10 of the present volume. (32) But the phenetic concept does answer the species taxon problem and could answer the taxon problem for other taxa; cf. n. 29. (33) Similarly, Matthen takes an organism to be a member of a particular species in virtue of belonging to the same extended reproductive community “as the originally ostended individual” (1998: 120). (34) And for that reason it is charitable to construe any remark that seems to suggest this answer as making only an epistemic point.

Page 33 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * (35) Despite this, I am not making any claim about explanation in general, just one about what is required for explanations of these phenomena in biology and similar ones elsewhere. (36) Mayr's distinction is established but it is often overlooked: Dupré 1981; Stanford 1995 (on which see Devitt 2009a); Griffiths 1999; Sterelny 1999. (37) Matthen provides another clear example of the error (1998: 117–21). (38) [2009 addition] Perhaps it will help if we generalize the point to all categories and taxa. Inspired by the term ‘conspecific’, let us call organisms that are members of the same taxa in a category C, “con‐Cic” Now consider these questions: (1.) In virtue of what is an organism a member of a taxon T? (2.) In virtue of what is a taxon T in a category C? (3.) In virtue of what are organisms con‐Cic? The answers to 1 and 2 entail an answer to 3. For, an answer to 1 tells us when organisms are members of the same taxon T, and an answer to 2 tells us when T is in C. Essentialism answers 1. Answers to 2 differ according to category. Furthermore, there tend to be disagreements over the answers in each category. BSC offers one answer to 2 for the category species. It offers no answer to 1. So it alone offers no answer to 3 in the case where con‐Cic is conspecific.

(39) Could the lack of doubt arise from conflating the false view that organisms are conspecific in virtue of being able to interbreed with the true view (assuming BSC) that they are conspecific in virtue of that in virtue of which they are able to interbreed? (Thanks to Michael Dickson.) (40) It is an interesting empirical question whether there are many plausible actual cases of anagenesis. (41) [2009 addition] The claim that the rejection of anagenesis is incompatible with Essentialism is mistaken. Call the group of organisms constituting the lineage from protists to the final descendents of Homo sapiens, “PH”. According to P‐CC, PH would be a species if there had been no splitting in the lineage. However implausible this is it is no problem for Essentialism because Essentialism has nothing to say about when a group is a species. Essentialism is simply committed to the view that if PH were a species (or any other Linnaean category) then it would have a partly intrinsic essence. And PH would have such an essence: there would be something intrinsically in common to all those organisms, although probably not very much. The truth underlying the mistaken claim, and the problem for P‐CC, is that PH is not a group worth identifying for the purposes of structural explanations, hence not a group worth identifying as a species in any circumstances. Its essence is not explanatorily interesting enough.

Page 34 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * (42) [2009 addition] This claim is also mistaken. Given the story, it follows from Essentialism that A and C share the intrinsic part of their essences. Of course, the two groups differ in their histories: A is the ancestor of B but C is not. According to P‐CC this makes A and C different species. But this is quite compatible with Essentialism because, once again, Essentialism has nothing to say about when a group is a species. The truth underlying the mistaken claim, and the problem for P‐CC, is that A and C are not groups worth distinguishing for the purposes of structural explanations. (43) Sterelny and Griffiths claim that, according to P‐CC, the levels of the traditional Linnaean hierarchy above the base level of species “make little sense” (1999: 201). If this were taken as a view of the taxa then Essentialism should not go along with it. But P‐CC does not support such a view of the taxa. The claim should be taken rather as a view of the categories, in which case it is quite compatible with Essentialism (Devitt 2009a, e, which is Ch. 10 in the present volume). (44) Kitcher (1984: 132–3 n. 27) refers to other similar suggestions in the literature. And note this claim by Hebert, as reported in the earlier‐cited item from the Scientific American online (n. 13): “We have very low levels of variation within a species and this deep divergence between species” (Biello 2007). (45) The evidence seems to point to genes that switch other genes on and off— e.g. Hox genes—being particularly important to the nature of a biological kind (Carroll 2005). (46) Nor need it go along with Hull's “three essentialist tenets of typology” (1965: 201). (47) Similarly Ereshefsky, writing about the essentialism of Lyell and Lamarck, claims that “their conception of species as evolving entities conflicts with this essentialist requirement” (1992b: p. xv). (48) The recent debate by the International Astronomical Union shows that Pluto is a good example of this indeterminacy. (49) “Essentialism is in principle consistent with vague essences” (Sober 1980: 253). Sober also draws attention to the fact that Aristotle was aware of “line‐ drawing problems” (pp. 252–3). (50) [2009 addition] Not so: see n. 41. (51) “The idea then is that if phenotypic change does not proceed by large jumps (saltations), then species are not objectively identifiable over time” (Sterelny and Griffiths 1999: 180; they do not endorse this idea).

Page 35 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * (52) I have argued this at length elsewhere (1991b, particularly ch. 13; 2001a, which is Ch. 5 in the present volume). Overlooking the distinction seems to rest on something like a use/mention confusion. (53) Kyle Stanford (1995) has a different view; see Ereshefsky 1998 and Devitt 2009a for criticisms. (54) It is easy even for staunch realists to slip into loose ways of talking that suggest worldmaking. Thus Kornblith says that when we “group objects together under a single heading on the basis of a number of easily observable characteristics . . . we thereby create a nominal kind” (1993: 41). But we don't! We create a concept that picks out a kind that may or may not be “real” in Locke's terms (n. 4 above) but which has its members independently of our creation. And Boyd, talking of kinds with nominal essences, says that their “boundaries” are “purely matters of convention” (1999: 142). But they aren't! Our naming a kind picked out by a certain set of descriptions is conventional but the boundary of the kind thus picked out is not. (55) Philip Kitcher, in commenting on an early version of this chapter, claimed that “knockout” mutants produced by modifying “normal” zygotes show that my treatment of monsters is too quick. It seems to me that my discussion accommodates these knockouts well enough: a minor mutant of a Drosophila melanogaster may still count as a Drosophila melanogaster because it has the essential intrinsic property that explains the characteristics it shares with “normal” Drosophila melanogaster; a gross mutant would not count as a Drosophila melanogaster because it does not have that property; the status of other mutants, doubtless most of the mutants, is simply indeterminate. We can learn about Drosophila melanogaster from these mutants, as we did, even if they themselves are not determinately Drosophila melanogaster. There is plenty of room for subtlety here. And if I am right in my arguments, something along the lines of my proposal has no viable alternative. (56) Note that exceptions that arise from varying the environment are not a problem. Indeed, typical generalizations about an organism are implicitly restricted to its “normal” environment. (57) Griffiths points out that “the generalizations of the special sciences often fail to live up to the ideal of a universal exceptionless law of nature. . . . Nevertheless . . . they have “counterfactual force” (Griffiths 1999: 216). Referring to history, social sciences, geology and meteorology, Boyd notes that “causally sustained regularities . . . need not be eternal, exceptionless, or spatiotemporally universal” (1999: 152). [2009 addition] Marc Lange (1995) has argued persuasively that statements of the form ‘The S is T’ about a species S can express natural laws even though there could be Ss that are not T.

Page 36 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Resurrecting Biological Essentialism * (58) [2009 addition]. We need to distinguish two sorts of potential “exceptions” to a law‐like ‘All Fs are P’: those arising from a mutation and those not. The objection seems to concern the former sort. My third, and final, response addresses this sort, denying that there can be such exceptions. Organisms that appear to be exceptions are not determinately F. But there can be exceptions that do not arise from mutations. My first and second responses address this sort. (59) [2009 addition] Not so: Essentialism can accommodate these implausible features of P‐CC; see nn. 41 and 42. (60) This chapter started with “Some Heretical Thoughts on Biological Essentialism”, an 8‐page piece I wrote in 2003, on the basis of little reading, and sent to a number of experts for comment. This had two surprising consequences. First, the volume of response was astounding: initial responses together with follow‐up discussions amounted to 100 pages. Second, given the consensus, I expected the experts to identify deep flaws in these “heretical thoughts”. Yet this did not happen. I was corrected, informed, and guided on many matters and yet my basic argument for biological essentialism seemed to me to survive fairly intact. The experts I am indebted to for their heroic attempts to set me straight at that point are: Peter Godfrey‐Smith, Paul Griffiths, Stephen Schwartz, Stephen Stich, and, particularly, Joseph LaPorte, Karen Neander, and Samir Okasha. The first version of the paper of which this chapter is a version was delivered at the University of Queensland in Nov. 2005 and later versions have been delivered at many other universities. The chapter has benefited greatly from those events and also from the written comments of Matt Barker, Alberto Cordero, Michael Dickson, Marc Ereshefsky, Philip Kitcher, Joseph LaPorte, Mike Levin, Georges Rey, Iakovos Vasiliou, John Wilkins, and Rob Wilson. Finally my thanks to Macquarie University for the position of Visiting Associate in Oct. and Nov. 2005, during which the first version of this chapter was mostly written.

Access brought to you by:

Page 37 of 37

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Naturalism and the A Priori *

Putting Metaphysics First: Essays on Metaphysics and Epistemology Michael Devitt

Print publication date: 2009 Print ISBN-13: 9780199280803 Published to Oxford Scholarship Online: February 2010 DOI: 10.1093/acprof:oso/9780199280803.001.0001

Naturalism and the A Priori * Michael Devitt (Contributor Webpage)

DOI:10.1093/acprof:oso/9780199280803.003.0013

Abstract and Keywords This chapter starts from the rejection of a priori knowledge on two grounds: first, confirmation holism removes any strong motivation for thinking that mathematics and logic are immune from empirical revision; second, the idea of a priori knowledge is deeply obscure, as the history of failed attempts to explain it show. The chapter defends this position from Rey's argument for a reliablist a priori and Field's for an a priori logic. It argues that Rey has not explained a way of knowing at all, hence not an a priori one: he has not shown how the beliefs reliably generated by his ‘logical sub-system’ are epistemically nonaccidental. The dominant idea of Field's argument is that logic must be seen as a priori because we need logic to get evidence for anything. The chapter gives a reason for thinking that this idea is ‘fishy’: an evidential system can undermine itself. Keywords:   naturalism, a priori, mathematics, logic, confirmation holism, empirical, Rey, reliablism, Field

In Coming to Our Senses (1996), I argue for and apply a naturalistic semantic methodology to defend an anti‐holistic truth‐referential view of meaning. The naturalism in question is an epistemological doctrine that I take from Quine: there is only one way of knowing, the empirical way that is the basis of science (whatever that way may be). So I reject “a priori knowledge”. I do not give a detailed argument for my rejection but I do give two reasons (2.2). Briefly, first, with the recognition of the holistic nature of confirmation, we lack a strong motivation for thinking that mathematics and logic are immune from empirical

Page 1 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Naturalism and the A Priori * revision; and, second, the idea of a priori knowledge is deeply obscure, as the history of failed attempts to explain it show.1 In this chapter, I will defend this view of the a priori from the criticisms of two other naturalistically inclined philosophers with whom I am usually in a great deal of agreement, Georges Rey (1998; also 1993) and Hartry Field (1998; also 1996). Rey disagrees with me explicitly and sharply. Field does so implicitly and less sharply.2

I. Rey's Reliablist A Priori Rey rightly insists “that whether or not there is a priori knowledge is an empirical issue” (1998: 25); on my view, every issue is an empirical one. But he thinks that this issue is open and may well be settled in favor of a priori (p.254) knowledge.3 In response to my first reason against the a priori—the lack of motivation—Rey is rather scornful. In response to my second reason—the obscurity—he appeals to reliablist theories of knowledge and claims to give at least a sketch of how we might have a priori knowledge of logic, mathematics, and analytic truths. He finds my comments on this sketch in Coming “a little bewildering” (p. 41 n. 9). I shall discuss his responses in turn. But I start with some preliminary points. (1) I am claiming that knowledge can be justified only by experience, that the evidence for it must be experiential. I doubt that there is any innate knowledge and so am inclined to think that experience must also be part of the source of a person's knowledge. But that is another matter and not my concern in Coming. Rey sometimes writes as if he thinks some knowledge may be innate. Suppose that some is. My second reason against the a priori still applies. If what is innate is indeed knowledge, it must be justified. We have some idea how we might establish the justification for innate empirical knowledge: experiences of the worldly facts that are the subject of the beliefs might play a role via adaptation in the production of the innate beliefs.4 But we have no idea how we might establish the justification for innate a priori knowledge. (2) Although Rey is not a naturalist in my sense, he claims to be one in some sense. I think that he is a bit confused about the sense in which he can be one, given his position on the a priori. Two sharply different doctrines are often called “naturalism”, one metaphysical and the other epistemological. Metaphysical naturalism is physicalism: the view, roughly, that all entities are physical entities and that the laws they obey are in some way dependent on physical laws. This is a reductive doctrine. It has nothing to say about ways of knowing except that they must be, like everything else, physicalistically acceptable: so it alone entails nothing one way or the other about a priori knowledge. Rey can endorse this doctrine and it is clear that he does. Quine and I do too, but we call it “physicalism” not “naturalism”. The other doctrine called “naturalism” is an epistemological one that is not reductive and is opposed to a Page 2 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Naturalism and the A Priori * priori knowledge. Quine is expressing this doctrine in rejecting (p.255) first philosophy and insisting that reality must be examined scientifically.5 It is what he aims to capture in his vivid metaphor of the seamless web and I aim to capture in my claim that the only way of knowing is the empirical way. Epistemological naturalism applies to all knowledge. So it applies to knowledge of ways of knowing themselves, yielding what Quine calls “naturalized epistemology”. In rejecting my naturalism, Rey is rejecting this epistemological naturalism, exemplified in naturalized epistemology and taken from Quine. So he is contradicting himself when he goes on to claim that the naturalism that he assumes “is Quine's own” (1998: 32). Furthermore, this claim is at odds with his own argument which attempts to show that epistemological naturalism may well be false: he is arguing that there may well be a priori knowledge and first philosophy! He does go on to quote approvingly Quine's description of naturalized epistemology (pp. 32–3), but it is clear from his discussion that all he is thereby endorsing is the use of the empirical method to argue for the a priori. Now if one must argue this—and one mustn't and shouldn't—it is certainly better to do so empirically rather than a priori: better not to start in sin even if one ends up there. But using the empirical method from time to time does not make you a naturalist, else everyone would be one. What makes you a naturalist, in Quine's epistemological sense, is a commitment to there being no other method, a commitment that Rey manifestly does not have. In sum, Rey is a metaphysical naturalist or physicalist. He is neither an epistemological naturalist nor a naturalized epistemologist, but he does think that the a priori ought to be argued for empirically. (3) Rey's discussion of Quine's attack on the analytic/synthetic and the a priori/ empirical distinctions is subtle and highly illuminating. Yet in one respect it strikes me as somewhat obtuse: Rey trivializes Quine's revisability thesis by characterizing it as follows: “any belief can be reasonably revised in the light of experience”.6 He finds this indistinguishable from “banal fallibalism” (1993: 72): “people could be wrong about anything; they can make errors in reasoning, rely on experts that mislead them, or just reason themselves into strange corners” (1998: 26–7). For example, Rey, in balancing his checkbook (p.256) concludes, “16 + 17 + 18 + 23 + 100 = 174” but abandons this arithmetic truth on being told by the bank that he is wrong. He reasons that surely the bank is better at addition that he is (1993: 70–1). He points out that this fallibalism “has nothing to do with the a priori”, for surely no rationalist ever denied that you could make errors in your a priori reasoning (1998: 27). True enough,7 but this fallibalism also has nothing to do with Quine's thesis, for surely Quine and the empiricists were aware that the rationalists accepted this fallibalism. Banal fallibalism can't be the right way to understand Quine and the dispute over the a priori.

Page 3 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Naturalism and the A Priori * An empiricist ought to accept a distinction between two ways that further experiential evidence can and should lead a person to change her mind about a statement p. (i) On the one hand, the evidence might bear for or against p itself. (ii) On the other hand, the evidence might throw light on the goodness of her thinking about p. She is a fallible calculator. New evidence may throw no direct light on p but may suggest that she has made a mistake in her thinking about the relation of the evidence to p. Quine's revisability thesis is surely concerned only with (i): “no statement is immune to revision” (Quine 1953: 43) in that experiential evidence might directly bear against it. The thesis is simply concerned with the relation between evidence and statement not with the relation between evidence and the view that a particular person (or even a particular community) has thought well about the statement. In fact, the thesis is the epistemological naturalism that I have just described, captured by the metaphor of the seamless web. Understood in this way the thesis strikes at the heart of apriorism and is far from trivial. Rey has responded to criticisms that may seem similar to this but his responses do not address adequately this particular point. First (in response to my 1993: 53 n., among others) he discusses the following view: the Quinean “point is that someone could hold on or revise any statement and still be rational” (1993: 71). But this is not the present point. Rey is right that the apriorist should accept that it may be rational for someone to stop holding any belief in the face of evidence of type (ii); for example, Rey at the bank. The Quinean point is that the content of the belief can be objectively disconfirmed by recalcitrant experience. Second (in response to Field 1996: 4) he describes something like the distinction between (i) and (ii) and then construes it as a “distinction between matters of empirical evidence and matters of reason alone”, a distinction he rightly thinks the Quinean is not entitled to (1998: 27). But the distinction between (i) and (ii) should not be construed in this way. It is a distinction between two ways in which experiential evidence can lead to (p.257) belief revision. (ii) is not a matter of revision by “reason alone” but a matter of revision because empirical evidence—for example, the disagreement of the bank with Rey's addition— suggests a failure in reasoning. This having been said, Rey's discussion raises an interesting possibility. It is common to confront the apriorist with historical examples of allegedly a priori knowledge abandoned in the face of experience; for example, the Euclidean view of space. In the light of the above distinction, it is always open to the apriorist to respond that this experience was relevant in way (ii). The evidence does not count for or against the statement in question, it simply shows that our process of nonempirical justification was defective in this case. Of course, it remains to be argued that this is a plausible response in a particular case. In my view, it is rather clearly not in the case of Euclidean geometry.

Page 4 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Naturalism and the A Priori * In any case, the argument against apriorism and for the seamless web that we should take from Quine does not rest primarily on these historical examples. It rests primarily on the two reasons I gave. The first of these involves confirmation holism, but not quite in the way Rey seems to think (1998: part I; 1993: 78–81). The argument starts by pointing out how scientific laws that are uncontroversially empirical are holistically confirmed. Evidence for this is not to be found only in the discussions of Duhem and Quine: most of the evidence comes from the movement in the philosophy of science inspired by Thomas Kuhn and Paul Feyerabend. It is then plausible to extend this holism to all beliefs, even those of logic and mathematics; there is no motivation for a seam in the web. I turn now to Rey's response to this reason. Rey has a lot of rhetorical fun mocking the remarks that Quine and I make about the empirical way of knowing and about the application of this way to logic and mathematics. He thinks that I am under “the illusion” that these remarks amount to “a serious theory” (1997: 146). I am not. I agree with Rey that “no one yet has an adequate theory of our knowledge of much of anything” (1998: 29); as I say, “we do not have the rich details of the empirical way of knowing that we should like to have” (1996: 50). In any case, Rey's mockery is largely beside the point. Since we do not have a serious theory that covers even the easiest examples of empirical knowledge, the fact that we do not have one that covers the really difficult examples from logic and mathematics hardly reflects on the claim that these are empirical knowledge too. We all agree that there is an empirical way of knowing. Beyond that, this part of the argument against the a priori needs only the claim that the empirical way is holistic. We have no reason to believe that a serious theory would show that, whereas empirical scientific laws are confirmed in the holistic empirical way, the laws of logic and mathematics are not; that it would show (p.258) there is a principled basis for drawing a line between what can be known this way and what cannot. I would be the first to concede that this part of the argument alone is far from conclusive, a long way from proving that the holism extends to logic and mathematics. That is why I put a lot of weight on the rest of the argument: my second reason, which is about the obscurity of a priori knowledge. In this part, I attempt to show that the alternative a priori explanation of our knowledge of logic and mathematics, indeed of anything, is very unpromising. If this is right, we have a nice abduction: the best explanation of that knowledge is that it is empirical. The a priori way of knowing is typically characterized by what it is not: it is not empirical. But what we need if we are to take the a priori way seriously is some idea of what it is. We need a positive account, not just a negative one. Why? This question may seem particularly pressing since I have just agreed that we do not have a serious theory of the empirical way. However, there are two crucial differences in the epistemic status of the two ways. First, the existence of the Page 5 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Naturalism and the A Priori * empirical way is not in question: everyone believes in it and Rey is even urging us to use it to show that there is an a priori way. In contrast, the existence of the a priori way is very much in question. Second, even though we do not have a serious theory of the empirical way, we do have an intuitively clear and appealing general idea of this way, of “learning from experience”. In contrast, we do not have the beginnings of an idea of what the a priori way might be; we lack not just a serious theory but any account at all. Rey claims to provide just such an account, appealing to the idea that knowledge is true belief arrived at by a reliable process. He is pained by my unenthusiastic response. This response is certainly very brief (1996: 51 n.; it draws on my 1993). I shall expand it here. The objection to a priori knowledge is that we don't know what it would be for something to be known a priori. So a successful resurrection of a priori knowledge must describe a nonempirical way of knowing, a process for justifying a belief that does not give experience the role indicated above. The difficulty in meeting this demand is well‐demonstrated by the failure of traditional attempts: on the one hand, these assumed that we have Cartesian access to meanings; on the other hand, they took knowledge of logic for granted. The trouble with Rey's ingenious proposal is that it does not meet the demand either. He has seriously underestimated what is required for the resurrection. Rey proposes that we know logical truths a priori because they are produced by a sub‐system of the brain that enjoys a reliability “completely independently of whatever input (i.e. experience) an agent may receive” (1993: 91); for example, (p.259) Ellen realizes in her brain “a non‐axiomatic system of natural deduction, relying entirely on the operation of standard rules like modus ponens, universal generalization, conditionalization, etc.” An example of its output is: (R) Nothing bites all and only those things that don't bite themselves. This “would be knowledge because each of the rules she used were surely justified if anything is. As soundness proofs of first‐order logic show, they are, indeed, absolutely reliable in this sense: it is impossible for them to produce a falsehood as a theorem” (1998: 34). The system's insensitivity to sensory input is intended to show that this is not an account of the empirical way of knowing. My objection is that it is not an account of a way of knowing at all.

It has been clear, at least since Plato, that for a belief to count as knowledge, it must not only be true, it must be justified: there has to be something about the way it was produced or sustained that makes it epistemically nonaccidental. Thus, if I believe on no good basis that it will rain tomorrow, or that a certain number is a prime, and I should turn out “by accident” to be right, my belief does not count as knowledge. The problem with Rey's account is that it does not

Page 6 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Naturalism and the A Priori * show how Ellen's belief in (R) is, in the appropriate way, epistemically nonaccidental. Consider another belief of Ellen's: the clearly empirical belief (E) that Tom is mortal. Suppose that Ellen knows that Tom is a man. Suppose further that (E) is the product of a sub‐system that took Ellen's knowledge that Tom is a man as input and that always yields the output that x is mortal given the input that x is a man. Inspired by Rey, we then argue: “(E) is knowledge because the rule she used was surely justified if anything is. As a vast amount of empirical research has shown, this rule is reliable in this sense: given a truth of the form ‘x is a man’ as input it will always produce a truth as output. And the rule took a known truth as input.” Clearly something is wrong with this argument. More needs to be said to show that the production of truths like (E) by “the mortality sub‐system” is epistemically nonaccidental. To help see this, suppose that the sub‐system in Ellen was itself produced by a “random” process that also produced other sub‐ systems which frequently yield false outputs; for example, given the same known input, yield the output that Tom is wise, round, brave, etc.; or, given other inputs such as that x is a bachelor, yield the output that x is rich. Among these many sub‐systems, one just happens to reliably produce truths, the mortality sub‐ system. We need to say more to rule out that the sub‐system is thus accidental. We might say something about how the sub‐system was produced: that experiences of a mortal‐man world played an essential role in that production. But it is not necessary to say (p.260) this.8 It is necessary to show that the sub‐ system, however it was produced, is causally sensitive in an appropriate way to the fact that we live in a mortal‐man world. The outputs of Ellen's sub‐systems in the imagined situation are not knowledge because those sub‐systems are impervious to the way the world is; impervious to whether it is a mortal‐man, wise‐man, round‐man, brave‐man, rich‐bachelor world. The problem with Rey's proposal can be put briefly: he does not say more. What he says about his logical sub‐system is analogous to my Rey‐inspired argument about the mortality sub‐system. We have just seen that that argument fails. Rey's argument fails for the analogous reason. He needs to show that the production of truths like (R) is epistemically nonaccidental. Once again, we can bring out the problem by supposing that the logical sub‐system in Ellen was produced by a “random” process that also produced other similar sub‐systems which frequently yield false outputs. Perhaps these sub‐systems include one that realizes the gambler's fallacy; one that fails to take proper account of the base in probabilistic reasoning; one that commits the fallacy of asserting the consequent. Rey needs to say something to rule out that it is a mere accident that one of Ellen's sub‐systems yields truths. And, of course, he cannot say that it yields truths because it is causally sensitive in an appropriate way to the fact that we live in a logical world. For then the knowledge of (R) would be empirical.

Page 7 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Naturalism and the A Priori * It does no good to insist that the logical sub‐system is reliable in that it regularly produces truths, for that is true also of the mortality sub‐system which, as we have just seen, may not be producing knowledge at all. Being reliable in this respect that Rey emphasizes is simply not enough. It does no good to appeal to proof theory to demonstrate how sure we theorists are of the reliability, in that respect, of the logical sub‐system. For the reliability of the sub‐system in that respect is not in question; and, in any case, we theorists are also sure of the reliability of the mortality sub‐system in that respect. We need to know something more about Ellen to establish that she arrived at (R) by a procedure that is, in a broader respect, epistemically reliable. It does no good to claim, as Rey does, that “the logical truths are those sentences that are true by virtue of the pattern of operators alone, independently of the assignments to the referential devices”9 and to insist that Ellen arrives at her belief in these truths because her sub‐system is appropriately sensitive to (p.261) these patterns (1998: 35). For, it may be a mere accident that her sub‐system is sensitive to these patterns rather than others that would lead her into falsity. Suppose, for example, that the sub‐system was produced by a random process that produces other such sub‐systems sensitive to a vast variety of different patterns. Ellen just happens to be one of the few people lucky enough to have scored a valid sub‐system. Manifestly, we would need a story about how Ellen's beliefs are nonetheless justified. In saying all this I am not insisting that Ellen can only know (R) if she knows that she knows—the KK principle—or if she has available to herself some justification of the rules of the logical sub‐system (cf. Rey 1998: 36; 1993: 92), nor am I simply refusing to accept a reliablist approach to knowledge.10 I think that the reliablist idea for empirical knowledge, briefly indicated above, may well be along the right lines: Ellen knows (E) if the mortality sub‐system that produced it is causally sensitive in an appropriate way to the fact that we live in a mortal‐ man world. This can be the case without Ellen knowing that it is, let alone knowing that it must be the case for her to know (E). I am insisting that there be a justification for Ellen's believing (E), not that she knows the justification. Putting this reliablist idea together with my naturalism, I think that it may well be the case that Ellen knows (R) because it was produced by a logical sub‐ system that is causally sensitive in an appropriate way to the fact that we live in a logical world. (We do not know the details, of course, but this does not count against (R) being empirical because we are hardly better off with (E).) There is a difference between the logical and the mortality sub‐systems which seems to play a role in Rey's thinking: the former unlike the latter yields necessary truths.

Page 8 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Naturalism and the A Priori * If empirical knowledge can be the result of a process that reliably . . . issues in true beliefs in relevant circumstances, why couldn't a priori knowledge be the result of a process that reliably issues in true beliefs in all possible circumstances? (1998: 34, see also p. 28; 1993: 92, 95) The answer is: because that metaphysical difference in the output is not epistemically relevant. In the case of the mortality sub‐system, we have seen that we have to say more than that it is reliable in Rey's respect if the contingent (E) that is its output is to count as knowledge. It would be strange indeed if we were relieved of the responsibility of saying more in the case of the logical sub‐system by the fact that its output (R) is necessary; so the requirements on (p.262) knowing a necessary truth would be less demanding! Indeed, a person might have sub‐systems that yield necessary truths which are obviously empirical and so where the need to say more should be uncontroversial; for example, the truths that water is H2O and that Hesperus is Phosphorus. The fact that a statement is necessarily true can no more show that the process of arriving at it is epistemically nonaccidental than can the fact that it is true. That fact does not undermine my argument.

Aside. How might the necessity of the output of the logical sub‐system be accommodated by a reliablist account of empirical knowledge of the output? What could the appropriate causal sensitivity to a logical world amount to? We are tempted to say that Ellen's mortality sub‐system is sensitive to a mortal‐man world in that if the world were different she would not have had that system. If we then apply this approach to Ellen's logical sub‐system, we seem to have a problem: the world cannot be different from the logical world; it is necessarily logical. So, it is common to think, the logic of subjunctive conditionals makes it trivially the case that if the world were nonlogical she would not have had that logical sub‐system. I doubt this view of subjunctive conditionals.11 In any case, perhaps we were wrong to be tempted by this approach. It builds a certain sort of infallibility into knowledge: Ellen knows (E) only if she could not have the mortality sub‐system unless she lived in a mortal‐man world. This seems to overlook the message of Descartes's First Meditation and the underdetermination of theories by the evidence. Perhaps what we need to say is that Ellen would not have had the mortality sub‐system unless her world appeared to be a mortal‐man world, unless she had experiences appropriate to such a world. She might have had other experiences, perhaps caused by an Evil Demon, even though she does live in such a world. Similarly, perhaps we should say that she would not have had the logical sub‐system unless she had experiences appropriate to living in a logical world. She might have had other experiences, perhaps caused by an Evil Demon, even though the world she lives in is necessarily logical. The point that we need to say more than Rey does for (R) to be knowledge may be more obvious if we change the examples a little: replace talk of sub‐systems with talk of general beliefs. Suppose that (E) was not produced by the mortality sub‐system but is inferred from the general belief that all men are mortal. Page 9 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Naturalism and the A Priori * Clearly, (E) will count as knowledge only if the general belief does. So we need to say more to show that general belief does. Now suppose that (p.263) (R) was not produced by a non‐axiomatic system of natural deduction but is inferred from some general logical beliefs. Once again, the epistemic status of (R) depends on the status of the general beliefs. So we have to say more to show that they are knowledge. It is hard to see how the change from general beliefs to sub‐systems of rules could remove the need to say more. I wonder if the talk of reliability is confusing the issue. So, consider what Rey tells us about (R) that might constitute its justification. First, of course, it is true. Second, it is the product of a sub‐system that regularly produces such truths. Obviously, we have no justification so far; think of the mortality sub‐system. Third, the logical sub‐system is insensitive to sensory input. But this insensitivity clearly cannot justify the system's output. Note that it would do Rey no good to suppose that his logical sub‐system is innate. What nonempirical story could possibly be told of its innate presence that would support the view that its output is knowledge? The evidence suggests that some “good” and some “bad” logical sub‐systems may be innate. What could justify the “good” ones apart from some empirical story? In sum, the comparison with the mortality sub‐system shows that the respect in which, according to Rey, his logical sub‐system is reliable is insufficient to establish that its output is knowledge. To characterize a way of knowing he needs to say more, showing that the system is, in a broader respect, epistemically reliable. There is no reason to suppose that if he were to say more he would characterize a way of knowing different from the empirical way.12

II. Field's A Priori Logic The attraction of the empirical view of logic is nicely brought out by Hartry Field: there is a 100% correlation between what is indeed true in logic and what the evidential systems that idealize our epistemic practice say is true in logic. How is this correlation to be explained? (1996: 13) A satisfactory explanation of this correlation between the logical facts and our logical beliefs should “make sense of the idea that if the logical facts had been different then our logical beliefs would have been different too” (pp. 13–14). And making sense of that idea is precisely what the empirical view of logic claims to do.

(p.264) Still, Field thinks that this claim is wrong: “we can simply make no sense whatever of the question of what we would believe were the logical facts different” (p.17) because of logic's “inextricable role in ascertaining the dictates of evidential systems” (pp. 18–19). Field argues that logic, by which he has in mind classical logic, is a priori. He thus implicitly disagrees with my Quinean naturalism. However, his requirements on a priori knowledge are unusually Page 10 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Naturalism and the A Priori * moderate and his conclusion is very qualified, so the disagreement may not be great. I shall not discuss all the nuances of his argument but simply respond to what I take to be his argument against my Quinean view. According to Field the view that logic is a priori amounts to the following two claims: (i) that it is reasonable to infer according to the rules of that logic without any empirical evidence for the legitimacy of those rules; (ii) that those rules are empirically indefeasible, in the sense that no possible combination of observations should count as evidence against their legitimacy. (1998: 1) There are also two related claims, (iw) and (iiw) concerning “logical principles” (Quine's “logical truths”) which are statements inferable from the rules of logic alone. I shall discuss (i) and (iw) first and then (ii) and (iiw).

Field thinks that “logic is almost incontrovertibly a priori in the weak sense of (i) and (iw)”. His reason is clear: in order for anything to count as evidence for anything, we have to use logic; logic licences the inferences from evidence to conclusion and so must come first; “if one is debarred from believing logical principles prior to empirical evidence for them, one is effectively debarred from ever believing them (p. 2; also 1996: 5–7). This exemplifies the dominant idea of Field's discussion: logic is special and must be seen as a priori because we need logic to get evidence for or against anything. Clearly Field has a point. A person must begin the pursuit of knowledge using some innate evidential system—a set of deductive and inductive rules—before she has any empirical evidence for its legitimacy. So, in that pragmatic respect, it is indeed “reasonable” for her to use this system—call it “S”—and logic comes out weakly a priori according to Field's definition (i). However, in another respect, most of us would hold that it remains to be seen whether it is reasonable for her to do so. In this other respect, we are concerned with the justification of S, with whether or not S is a good system, whatever the particular person may have reason to believe about this. This other respect demands a revision of Field's (i): for logic to be weakly a priori is for its rules to be legitimate and for that legitimacy not to depend on any empirical evidence; legitimacy must be established by some nonempirical way of knowing. On this (p.265) revised definition, the aprioricity of logic is certainly not incontrovertible. I, for one, do controvert it. Of course, Field is an epistemological nonfactualist and so denies that there is any such other respect, any issue of justification beyond what is pragmatically reasonable:

Page 11 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Naturalism and the A Priori * It makes no sense to ask whether logic really is justifiable a priori; the only issue is whether it is advantageous to employ an evidential system that licenses adhering to a logic whatever the empirical evidence may be. (1996: 19) So Field feels entitled both to claim that logic is weakly a priori even though he has not described a nonempirical way of knowing it, and to see this claim as naturalistic (p. 19).

I think that his nonfactualism is too extreme. I do not dispute most of the considerations that lead him to it; for example, trade‐offs to do with reliability and power; and the problem of saying just how successful an evidential system must be to be “good enough” (1998: 6–9). So I accept that there is a certain amount of indeterminacy in judgments about whether one system is better than another, and about whether a system is justified (1991b: 77–8). But the indeterminacy does not seem to me to stretch as far as Field thinks. He allows that the choice of an evidential system is only “partly” nonfactual. I don't think he has shown that the nonfactual parts include the a priori issue. He claims: “If, as I have been arguing, questions of justification or evidence are not fully factual, then presumably questions of a priori justification aren't either” (1998: 11). I resist the presumption. Consider S. I accept that there may be many possible evidential systems that are not determinately better or worse than S. Still, so far as I can see, nothing Field says shows that S might not be determinately a good system, a justified system,13 even if not determinately the best system. Suppose that it is good. Suppose, as Field does, that it contains classical logic: deductions within S always follow the rules of classical logic. Now suppose that every possible rival evidential system that is not determinately worse than S also contains classical logic. If this supposition is correct then we have good reason to think that classical logic is determinately justified. So far as I can see, Field does not show that this supposition is not correct. The reasons Field gives for thinking that justification in general is partly nonfactual do not show that the justification of logic is nonfactual. (p.266) If logic is indeed determinately justified, then we need an explanation of how it is. Quineans will seek an empirical explanation. Since S is innate, the explanation must be found in evolutionary history: S has been a success in the past and as a result of that has been inherited by the person in question. Spelling out the details here is at least as difficult as it is anywhere else in epistemology. If Field is to persist in his view that logic is a priori, he must reject any such empirical explanation. But then he would, after all, owe us an account of how logic was otherwise justified, an account of a nonempirical way of knowing. I shall return to this point.

Page 12 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Naturalism and the A Priori * Turn now to Field's (ii) and (iiw) and the issue of the empirical indefeasibility of logic. First, some preliminary remarks about the empirical revision of logic. S is a system of inference rules including the rules of classical logic, for example, modus ponens. So, a person embodying S is disposed to infer according to the pattern: If p then q, p, So, q. And, as a result, she is disposed to believe instances of the schema, ‘If p then q and p, then q’. But the person who embodies S need not have any theory of these matters. So she need not believe that modus ponens is a valid inference nor that ‘If p then q and p, then q’ is a logical truth. Indeed, in my view, it is almost certain that she does not have such beliefs innately, and unlikely that she will ever come to have them, given our ordinary indifference to logical theory. Call the epistemological theory recommending S, and hence the rules of classical logic, “T”. The Quinean empiricist view of logic is more explicitly concerned with T rather than S, with beliefs about logic rather than with logical practices.14 The claim is that these beliefs, although central in the web of belief, are nonetheless revisable in the light of experience. Still, this claim has clear consequences for S, the system that “holds the web together”. Suppose that experience leads a person to abandon T in favor of T’, a theory that recommends an evidential system S' built around a nonclassical logic. Then clearly the person should abandon S in favor of S'. In this way the Quinean thinks that our logical practices are themselves open to rational revision in the light of experience.

This is not to say that it is psychologically possible for a person to make any of these revisions in the light of experience. Perhaps there are all sorts of psychological difficulties in rationally replacing T by T’, or, having done so, (p. 267) in replacing S by S'. However, we should not exaggerate these difficulties in light of the improvements in reasoning we observe in ourselves and others brought about by the study of logic and probability theory. In any case, the psychological possibility of making these replacements is not the point. The point is that experience could supply rational grounds for these replacements. (Nonrational replacements are obviously possible; for example, by violence or surgery.) Still, there seems to be a problem. A person's rational revision of her evidential system S is supposed to come from her abandoning T in the face of empirical evidence. But she must use S, in order to bring the evidence to bear on T. How then could this use show that T is false and hence that she ought not to use S? Field is very impressed by this problem: “It is hard to see how the logic employed could itself be subject to empirical confirmation” (1998: 12). This is the dominant idea of his discussion. As a result, he thinks that we have no choice but to treat our logic as a priori.

Page 13 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Naturalism and the A Priori * The Quinean solution to this problem is captured by the famous image from Neurath of rebuilding a boat whilst staying afloat on it. We can rebuild any part of the boat but in so doing we must take a stand on the rest for the time being. So we cannot rebuild it all at once. Analogously, we can contemplate revising any part of our evidential system S but in so doing we must hold fast to the rest of the system for the time being. So we cannot contemplate revising it all at once. This applies to the logic that is built into S. Field rightly insists that we need some such logic to draw any conclusions from the empirical evidence. Still, we could contemplate revising T and the system S that it recommends. Suppose that system S', recommended by T’, holds fast to most of classical logic but weakens the distributive law. We could employ S' to test our total science against the empirical evidence. If total science comes out better on this test than on our present test employing S, then we have empirical grounds, of the usual Duhem– Quine holistic sort, for preferring S'. To say that we could do this is not, of course, to say that we should do it. As I understand the situation, we do not have any promising S' to be empirically tested against S, despite the quandary of quantum theory. As Field says, “without some idea as to what a better system of evidence would be, it makes little sense to criticize the one we have”. And coming up with a better system is “no easy matter” (1996: 11). So the possibility of revision is a theoretical one not a practical one at this time. But that is all that Quinean epistemology requires. Field is, of course, aware of this Quinean solution to the problem of using logic to test logic, but he strangely ignores it in his 1996 discussion. Even his 1998 treatment of it is rather brusque. He complains that the solution is (p.268) “vague” and doubts that we can “spell out” the Quinean inductive method (p. 13). This criticism is unconvincing. The vagueness cannot be denied: consider, for example, my talk of total science “coming out better” using S' than S. Quine is up‐front about the vagueness, as Rey notes (1998: 29). But it is hard to see why this vagueness constitutes an effective criticism of the Quinean view of logic. Perhaps it would if realistic alternative evidential systems that treated logic as a priori had been spelled out. Field often writes as if he thinks that there have been, mentioning hypothetico‐deductive and Bayesian formalizations (e.g. 1998: 12). Doubtless these formalizations throw a good deal of light on our inductive methods, but I take it as generally agreed that they fall very far short of a realistic, albeit idealized, picture of those methods. Rey is surely right in claiming that “no one yet has an adequate theory of our knowledge of much of anything” (1998: 29). At this point we do not have a good detailed theory of the evidential system we should use. This is sad, but no special problem for Quine.

Page 14 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Naturalism and the A Priori * Field has a further point to make against the Quinean theory (a point he put to me in conversation). The evidential system recommended by the Quinean theory is not really S, embodying classical logic, but rather the system described a page back with the help of the image of Neurath's boat. This latter “higher level” system uses a “lower level” system like S whilst keeping open the possibility that S might be replaced on empirical grounds in the way described. Suppose that we accept this view of how S is empirically revisable. Still, how could the higher level Quinean system itself be empirically revisable? Employing that system according to which all justification is based on experience, how could experience show that the system is wrong and that some justification is not so based? Applying the dominant idea of his discussion once again, Field thinks that the Quinean system could not be thus revisable and so it must be regarded as a priori. For the Quinean, it is a priori that there is no a priori!15 This is a neat and perplexing point. Still, I think there is a reason for thinking that his dominant idea, however appealing, is “fishy”.16 Start by considering epistemological theories rather than evidential systems. Whether or not these theories are empirically revisable, they are indubitably revisable somehow. Thus, an epistemologist may start out committed to a priori knowledge because she can think of no other way to explain knowledge of logic and mathematics. Then she reads Quine, becomes impressed with confirmation holism, struck by the obscurity of a priori knowledge, and adopts (p.269) the Quinean theory. This seems to be a rational procedure in the descriptive sense, even if not in the normative sense: it seems to be governed by her underlying evidential system, whatever that may be. Later she may have second thoughts. She may, like Rey, be influenced by reliablism and change back to a belief in the a priori. Again this seems like a descriptively rational procedure. Now consider the implications of this for evidential systems. Our epistemologist's initial evidential system either takes justification to be based only on experience or it does not; either it is Quinean or it is apriorist. If it is Quinean then she should change it once she comes to believe that logic is a priori; the same rational procedure that leads her to change her epistemological theory should lead her to change her evidential system. Of course, if her initial evidential system is apriorist, then she should not change it at this point. But she should change it when she later converts to Quinean epistemology. And, either way, she should change it again when she has second thoughts and returns to apriorism. The moral of this is that an underlying evidential system, whatever it may be, can govern a procedure which, by its own lights, supplies a rational basis for its own revision. So there is something wrong with Field's dominant idea: an evidential system can undermine itself. Field uses the idea all the time to argue against the empirical revision of an evidential system. Yet, if the idea is good, it counts against any rational revision of the system at all.

Page 15 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Naturalism and the A Priori * One might respond to this by claiming that a person cannot rationally revise her evidential system. So the changes that our epistemologist went through are not governed by her evidential system but are the result of malfunctions in the system, of “noise”: the changes are no more rational than those brought about by surgery or violence. But this line seems very implausible. And it is not a line that is likely to appeal to Field. Although he opposes the empirical revision of logic he is happy to contemplate its “conceptual” revision (1998: 3, 4, 12). So, Field's neat point against Quinean epistemology seems to be in trouble: if evidential systems are to be rationally revisable at all, they must be able to undermine themselves. Of course, this alone does not show that they are empirically revisable. But, first, it leaves untouched the Quinean argument that they are empirically revisable. And, second, it prompts an argument against their being otherwise revisable. For, how are they otherwise revisable? If there is this nonempirical way of knowing, we need an explanation of it. Yet there is no explanation in sight. That is my dominant idea. Earlier, I argued that Field had not shown that the justification of logic is a nonfactual issue and so he owes an explanatation of how logic is nonempirically justified. His talk of revising logic on “conceptual” grounds reinforces the debt. For the Quinean, conceptual grounds are parts of the empirical grounds, the parts distant from experience and central to the web. Field must reject this (p. 270) because his conceptual grounds are contrasted with empirical ones. So he needs to tell us about these nonempirical grounds. I have argued that Field's dominant idea is “fishy” but I have not located the source of the smell. And this is something that should be done, because the dominant idea is appealing. Alas, I do not know how to do it. I comfort myself with the thought that we know so little about our evidential system. Notes:

(*) First published in Philosophical Studies (Devitt 1998). Reprinted with kind permission from Springer. (1) [2009 addition] For more detailed arguments see my 2005a, b, and 2009c, which is Ch. 13 in the present volume. (2) I criticized Rey 1993 in my 1993, a criticism modified slightly in my 1996 (p. 51 n.). Ancestors of Rey's present response to my views and my present response to his response were delivered at a conference on “Naturalistic Semantics and its Methodology” at the University of Maribor in June 1996 and at the D. C. Williams Conference on “Naturalism, Analyticity and the A Priori” at the University of Maryland in Apr. 1997; see Rey 1997 and Devitt 1997c: sec. 4. An ancestor of Field 1998 was also delivered at the D. C. Williams conference.

Page 16 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Naturalism and the A Priori * (3) And he agrees (1998: 41 n. 8) with Joseph Levine view (1993, 1998) that my “molecular localism”, according to which a few of the inferential properties of a word may constitute its meaning, commits me to a priori knowledge. The crux of my response to Levine (1996: sec. 1.11; 1997c: secs. 2 and 8) is that localism is a purely semantic doctrine with no commitment to anything epistemological. In particular, localism need not, and should not, subscribe to the view that competence with a word alone yields knowledge about its meaning, a Cartesian view that has played such a prominent role in traditional doctrines of analyticity and in semantics generally. [2009 addition] For more against the Cartesian view see Ch. 13, sec. 3.2, in the present volume. (4) [2009 addition] See Ch. 13, part I, preliminary point (2), in the present volume for a discussion of this issue. (5) See e.g. Quine 1981: 21, 67, 72. The distinction between his naturalism and his physicalism is implicit in a passage on p. 85. (In response to p. 72, Rey draws a red herring “about whether empirical science as a whole can be justified by some means external to it” (1998: 41 n. 10). Quine is pointing out that if epistemological naturalism is true then there is no such justification. Rey responds by rejecting the reverse conditional: that if epistemological naturalism is not true—if there is a priori knowledge—then there is such a justification. This is indeed “a quite different issue” but it is not one Quine has raised.) (6) My thoughts on this matter were sharpened in extensive correspondence with Rey prior to his 1993. (7) [2009 addition] Although I have since come to doubt this; see Ch. 13 n. 14, in the present volume. (8) In my 1993, I wrongly insisted that it was necessary (p. 55). (9) Rey's view here, and in the accompanying discussion, seems to be that the logico‐syntactic form of a logical truth is wholly responsible for its truth. This is surprising: does he not think that the world has something to do with its truth? I certainly do (1996: sec. 1.6; 1998: sec. 1). (10) Indeed, for all I know, my view may be consistent with the particular reliablist theories that Rey cites as the basis for his view (1993: 91 n.). (11) As does Field (1996: 17). (12) [2009 addition] See also Ch. 13 n. 3 for the suggestion of a further consideration against Rey's attempt “to save the a priori”. (13) That there is no determinate matter of fact whether some systems are good enough to be justified does not show that there is no determinate matter of

Page 17 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Naturalism and the A Priori * whether any system is good enough to be justified; cf. baldness and other cases of vagueness. (14) See e.g. 1952: pp. xiii–xiv. (15) Much to the delight of Rey (1997). (16) I return to this issue in Ch. 13, sec. 2.4, of the present volume.

Access brought to you by:

Page 18 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

No Place for the A Priori *

Putting Metaphysics First: Essays on Metaphysics and Epistemology Michael Devitt

Print publication date: 2009 Print ISBN-13: 9780199280803 Published to Oxford Scholarship Online: February 2010 DOI: 10.1093/acprof:oso/9780199280803.001.0001

No Place for the A Priori * Michael Devitt (Contributor Webpage)

DOI:10.1093/acprof:oso/9780199280803.003.0014

Abstract and Keywords This chapter argues that all knowledge is empirical. It attempts to undermine the motivation for the a priori that comes from examples in mathematics, philosophy, and logic. In particular, it argues that logic can be seen as empirical if rule-circular arguments are allowed. And a priorists cannot disallow such arguments because any justification of a priori reasoning would have to be rulecircular. The chapter argues that the whole idea of the a priori is unexplained and deeply obscure. Traditional attempts at explanation that appeal to analyticity fail in two ways. They rest on an unexplained acceptance of logical truths and on the mistaken view that competence with a concept is sufficient for knowledge about it. A consideration of the contemporary views of Peacocke, Bealer, and BonJour helps to bring out the obscurity. Keywords:   a priori, empirical, mathematics, philosophy, logic, rule-circularity, concepts, Peacocke, Bealer, BonJour

1. Introduction Why believe in the a priori? The answer is clear: there are many examples, drawn from mathematics, logic, and philosophy, of knowledge that does not seem to be empirical. It does not seem possible that this knowledge could be justified or revised “by experience”. It must be justified in some other way, justified a priori. So we have a motivation for the a priori. But there is severe problem: the a priori seems deeply obscure. What is it for a belief to be justified a priori? What is the

Page 1 of 21

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

No Place for the A Priori * nature of this nonempirical method of justification? Without satisfactory answers the a priori is left mysterious. In other works I have defended the naturalistic view that there is no a priori by attempting to undermine the motivation for the a priori and by demonstrating its obscurity (1996a, 1998, which is Ch. 12 in the present volume, 2002a, 2005a, b).1 In this chapter, I shall summarize this attempt and then develop it further. (p.272) But I start with two preliminaries. First, what is the empirical method of justification? An answer starts from the metaphysical assumption that the worldly fact that p would make the belief that p true. The empirical justification of the belief is then to be found in its relationship to experiences that the worldly fact would cause. Justified beliefs are produced and/or sustained by experiences in a way that is appropriately sensitive to the way the world is. This is very brief and we shall return to the question later. Still it is hard to say much more. Second, our concern with the a priori is with the justification of beliefs not primarily with their source. Experience is clearly not the source of many mental states: they are innate. Perhaps some of these are beliefs. I rather doubt this but suppose, nonetheless, that some were. That would raise two interesting questions. Could these innate beliefs be innately justified—justified, but not by the experiences of the believer—in a naturalistically respectable way? If so, would that justification fit the empirical model of justification we have briefly sketched? There seem to be two naturalistically respectable possibilities for innate justification. The first starts with a belief of some of our distant ancestors, a belief formed as a result of experiences that justified it in the normal empirical way. Now suppose that the belief is extremely beneficial to the survival of those that hold it. Then there might be a process of natural selection leading in time to that belief being innate. That alone would not make the belief innately justified because its beneficial effects may have nothing to do with whether it is true; for example, one could imagine false religious beliefs being beneficial to survival. But suppose that the belief is as a matter of fact true and that it was because it was true that it was beneficial and hence selected for. Such an innate belief would have been produced by a reliable mechanism and I think we should count it as innately justified; this selection process would be a reliable way for us to inherit the justificatory work of those distant ancestors. And I think that this justification would fit the empirical model well enough. It is worth noting that such innately justified beliefs would be a bit analogous to justified beliefs formed on the basis of testimony; for example, to learning that it is raining from someone reliable who has just experienced the rain. Of course, hearing testimony is an experience whereas receiving beliefs through your genes is not. Still in each case the believer's justification would be in a sense Page 2 of 21

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

No Place for the A Priori * indirect, not coming from experiences directly produced by the worldly fact that makes the belief true. (p.273) The second possibility does not seem to fit the empirical model.2 Again we start with a true belief of some of our distant ancestors. This time, however, they held that belief “by accident” without any proper empirical justification at all. The story then continues as before: suppose that because the belief is true it is extremely beneficial; it is then selected for and so becomes innate. Now if this belief is to count as justified, the justification must come entirely from the process of natural selection itself. No experiences of the worldly fact that makes the belief true played a role in producing or sustaining the belief in those distant ancestors: they simply happened on this belief which was beneficial because true and which was then selected for. Once again the belief would seem to have been produced by a reliable mechanism and should, I think, be counted as innately justified. Whereas the earlier first possibility is of an empirical justification being inherited by natural selection and so fits the empirical model well enough, this second possibility is of a justification by natural selection itself and so does not seem to fit the model at all. I doubt that there are any innate beliefs and doubt even more that there are any that are innately justified in either of these possible ways, But if some were innately justified in the second possible way, a naturalistic philosopher would have to broaden his view of acceptable justification beyond empirical justification (as usually conceived). But this broadening would not give us anything like a priori justification (as usually conceived).3 I shall ignore this possibility in what follows.

(p.274) 2. Undermining the Motivation 2.1 The Naturalistic Alternative Summarized

The task in undermining the motivation for the a priori is to show how the troublesome examples of allegedly a priori knowledge might be accounted for naturalistically. I have attempted this, drawing on Quine (1961, 1966, 1969, 1975b) and before him Duhem (1906). In brief, the key is breaking free of a naive atomistic picture of justification. We must view justification in a more holistic way: beliefs, even whole theories, do not face the tribunal of experience alone, but in the company of auxiliary theories, background assumptions, and the like. Such holism is well‐supported by the revolution in the philosophy of science inspired by Thomas Kuhn (1970). In light of this holism, it is argued, we have no reason to believe that whereas scientific propositions, which are uncontroversially empirical, are confirmed in the holistic empirical way, the propositions of mathematics, logic, and philosophy are not; no reason to believe that there is a principled basis for drawing a line between what can be known this way and what cannot; no reason to believe that there is, in Quine's vivid metaphor, a seam in the web of belief.

Page 3 of 21

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

No Place for the A Priori * I shall develop this view by considering in turn, in more detail, the problems posed for it by mathematics, philosophy, and logic. 2.2 Naturalism and Mathematics

Obviously, these brief remarks scarcely begin to solve the epistemological problem of mathematics. There are two reasons why this is not a great concern to the project of undermining the motivation for the a priori. First, as Georges Rey (1998) is fond of pointing out, we are not close to solving the epistemological problem of anything.4 Since we do not have a serious theory that covers even the easiest examples of empirical knowledge—examples where experience plays its most direct role—the fact that we do not have one (p.275) that covers the really difficult examples from mathematics hardly reflects on the claim that these are empirical too. We all agree that there is an empirical way of knowing. Beyond that, the present project needs only the claim that the empirical way is holistic. We have no reason to believe that a serious theory would show that, whereas empirical scientific laws are confirmed in the holistic empirical way, the laws of mathematics are not. Second, there is a special reason for not expecting the epistemological problem of mathematics to be anywhere near solved: the metaphysical problem of mathematics—what mathematics is about—remains so intractable. How could we solve the epistemological problem when we remain in such darkness about the metaphysical one? The point is that we no longer have any reason to think that, if we solved the metaphysical problem, the epistemological problem would not be open to an empirical solution.5 I emphasize that this is not the claim that we now have anything close to an empirical justification of a mathematical proposition. It is the much weaker claim that we now have no good reason to think that such a justification could not be given. The weaker claim is all that is needed to undermine the motivation for the a priori. Furthermore, I am not denying the striking epistemological differences between mathematics and science. (i) There is an obvious difference between observing and inferring, and an obvious difference between inferring deductively and inferring nondeductively or “ampliatively”. Where mathematical justification largely involves deductive inferences from “self‐evident” assumptions in proofs, scientific justification largely involves ampliative inferences from observations in experiments. But the claim is that all these differences could be accommodated in the naturalistic picture; for example, that the justification of the self‐evident assumptions could be empirical. (ii) Despite the holistic story, a scientific claim—even, say, the general theory of relativity— seems to answer fairly directly to certain evidence in a way that a mathematical claim does not. Nonetheless, the naturalist urges, this difference is just a matter of degree.6

Page 4 of 21

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

No Place for the A Priori * 2.3 Naturalism and Philosophy

George Bealer rightly points out that “in philosophy, the use of intuitions as evidence is . . . ubiquitous . . . these intuitions . . . determine the structure of (p. 276) contemporary debates in epistemology, metaphysics, and philosophy of logic, language, and mind”. He goes on to say that “in our context when we speak of intuition, we mean ‘rational intuition’ or ‘a priori intuition’ ” (1999: 30). This view of intuitions is, of course, fairly standard in philosophy. Indeed, it is common to suppose that only if intuitions are a priori can they play their evidential role in the characteristic “armchair” method of philosophy. From the naturalistic perspective, this common thought is mistaken. We have no need to see philosophical intuitions as a priori. We can see them as being members of a general class of empirical intuitions. I shall now summarize a view of these empirical intuitions that I have argued for elsewhere (1994; 1996a: 48–86; and, particularly, 2006c, which is Ch. 14 in the present volume).7 The intuitions are judgments that are empirical theory‐laden “central‐processor” responses to phenomena, differing from many other such judgments only in being immediate and unreflective, not based on any conscious reasoning. We should trust a person's intuitions as evidence about some kind we are investigating only to the degree that we have confidence in the person's empirically based expertise about that kind. Sometimes the folk may be as expert as anyone: intuitions laden with “folk theory” are the best we have to go on. For most kinds, it clearly is not: we should trust intuitions laden with established scientific theories. Even where we are right to trust an intuition in the short run, nothing rests on it in the long run. We can look for more direct evidence in scientific tests. In such a scientific test we examine the reality the intuition is about. Intuitions often arise in “thought experiments”. Instead of real experiments that confront the expert with phenomena and ask her whether they are members of the kind F, we confront her with descriptions of phenomena and ask her whether she would say that they were members of F. Thought experiments call on the same empirically based beliefs about the world as real experiments, and their results have the same empirical status. In light of this, turn now to philosophical intuitions and the “armchair” method of philosophy. The traditional explanation of this method, illustrated by Bealer, is that philosophers are conducting thought experiments that probe their concepts to yield a priori rational intuitions; they are doing “conceptual analysis”. The naturalistic explanation accepts that philosophers are conducting thought experiments but construes these differently. The philosophers are not probing concepts but rather intuitions about kinds. This is just as well because knowledge of concepts, being a species of semantic knowledge, is very hard to come by. In contrast, philosophers have acquired considerable knowledge (p. 277) of many kinds over a lifetime of acquaintance with them. The philosophers' intuitions that draw on this knowledge are not a priori but empirical. The philosophers are conducting thought experiments of the sort just mentioned, Page 5 of 21

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

No Place for the A Priori * counting themselves as experts about the kind in question. Thus, consider “the analysis of knowledge”, a famous example of the method and one that Bealer discusses (1992: 100). The philosopher, as expert as anyone in identifying cases of knowledge, confronts descriptions of epistemic situations and considers whether the situations are cases of knowledge. On the basis of these empirical intuitions about cases the philosopher constructs an empirical theory about the nature of knowledge. The naturalist does not deny armchair intuitions a role in philosophy but does deny that their role has to be seen as a priori: the intuitions reflect an empirically based expertise in the identification of kinds. So I am urging that philosophical intuitions should be seen as empirical central‐ processor responses to phenomena. This view has the great advantage of being theoretically modest: it treats these intuitions like intuitions in general. It accommodates the evidential role that intuitions play in philosophy without resort to the a priori. 2.4 Naturalism and logic

The naturalistic alternative seems to face a difficulty with logic. For, on this alternative, experience justifies beliefs in the interior of the web via logical links with beliefs at the periphery, via logical links with beliefs “close to experience”. But these justifications depend on the logical links themselves being justified: clearly a belief is not justified by other beliefs unless those others give it genuine support. And, many will claim, the justification of these links could not come from experience; it has to be a priori. Laurence BonJour has put the problem vividly: if there is no a priori insight . . . no prediction will follow any more than any other . . . any . . . sort of connection between the parts of the system will become essentially arbitrary. (2001b: 679)  . . . the rejection of all a priori justification is tantamount to intellectual suicide. (2001a: 626) In brief, the problem is that logic must be seen as a priori because we need logic to get evidence for or against anything.

To assess this problem we need to start with an important distinction. On the one hand, there are the rules that govern a person's practice in forming beliefs, that constitute her “evidential system” (Field 1996, 1998). These must include rules for forming beliefs from perceptual experiences and the logical rules that concern us here, rules for inferring one belief from another. On (p.278) the other hand, there are assumptions or theories about such rules. Thus it is one thing for a rule of inference R—for example, modus ponens—to be among the rules that govern a person's thinking, it is another thing to theorize that R does so govern.8 And it is also another thing to theorize about R's justification. So we have two types of epistemological theorizing about R that contrast with the epistemic activity of inferring according to R. The first type of theorizing is a Page 6 of 21

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

No Place for the A Priori * piece of descriptive epistemology, the second, a piece of normative epistemology. The normative concern about the justification of R is with whether it gives epistemic warrant to its conclusion when it operates on true premises; the concern is whether it is a good rule. A belief we form about the epistemic status of a rule, like any other belief, raises an epistemic question: is the belief justified? To avoid confusion, I shall in future talk of the goodness of rules rather than of their justification, and continue to talk of the justification of beliefs. Against this background, we can now frame the issue. It is clearly the case that for beliefs arrived at by a process governed by R to be justified then the following must be true: TR: R is a good rule. And the issue is: what is the justification of TR? How do we know that R is a good rule?

We have seen that BonJour thinks that the justification must be by an a priori insight or rational intuition. And so too does Bealer (1992: 100–2; 1999: 30–4) who describes the intuition as an “intellectual seeming”. This view may seem appealing if R is a rule of deductive logic like modus ponens. For we do at least have a lot of insight into deductive rules. But many, perhaps most, of the rules that govern our epistemic practices are ampliative rules. And these are rules that we don't have much insight into, whether a priori or not. We can, of course, wave our hands and talk of enumerative induction, abduction, simplicity, and the like, but we are unable to characterize these in the sort of detail that would come close to capturing the rules that must constitute our actual evidential systems; for example, we are unable to specify when an explanation is good, let alone the best, or when we should take the belief that all observed Fs have been G to justify the belief that all Fs are G. Aside (p.279) from that, some of these vague rules are controversial; for example, scientific realists love abduction, Bas van Fraassen does not. In sum, when we move beyond deduction, we have few if any specific and uncontroversial rules to be insightful about. The nonskeptics among us will share the very general insight that, whatever the rules that govern our epistemic practices may be, those rules are for the most part good. So, if S is the sum of these largely unknown rules, if S is our evidential system, we believe TS: The rules in S are for the most part good. Now if the claim facing the naturalist is that TS is justified a priori the claim hardly seems tempting, given that we don't know what S is. It seems more plausible to view our general insight that TS is true as supported by the empirical success of S, whatever S may be. Similarly, someone afloat on a boat may not know the methods by which it was built but, noting its seaworthiness, infers that the methods, whatever they were, are good. In sum, when we focus on the largely unknown ampliative parts of S, our confidence in S seems as Page 7 of 21

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

No Place for the A Priori * empirical as anything. To that extent, TS does not even appear to be justified by a priori insight. Be that as it may, it will be thought that, at least, our insights into specific deductive rules like modus ponens are a priori. But why must we see the support for the deductive rules as different in principle from that for the ampliative rules? We need to have confidence in S as a whole if we are to avoid skepticism. We can see that confidence as coming from the overall empirical success of S. Then the justification of our belief in the deductive parts of S is no different in principle from that in the ampliative parts. Similarly, all parts of S are empirically revisable. Thus, suppose that experience leads us to abandon TS in favor of TS', a theory that recommends an evidential system S' built around a nonclassical logic. Then clearly we should use S' instead of S. In this way our logical practices are themselves open to rational revision in the light of experience.9 These practices are far from “arbitrary”: they are recommended by an experience‐based epistemology. This raises what may seem to be the most serious problem for naturalism. On the one hand, I talk of TS being justified by the empirical success of S. Yet that alleged justification must come via S itself. So, the attempt to support (p.280) TS seems circular. On the other hand, I talk of the possibility of experience leading us to abandon TS in favor of TS'. Yet experience must be brought to bear on TS by using S and so could not show that TS is false and hence that we ought not to use S. The attempt to refute TS seems self‐defeating. A naturalist might attempt to respond to this by appealing to Quine's famous metaphor of Neurath's boat.10 Quine likens our web of belief to a boat that we continually rebuild whilst staying afloat on it. We can rebuild any part of the boat—by replacement or addition—but in so doing we must take a stand on the rest of the boat for the moment. So we cannot rebuild it all at once. Similarly, we can justify or revise any part of our knowledge but in so doing we must accept the rest for the time being. So we cannot justify or revise it all at once. So the claim that one of S's rules, say R, is good or not could be supported by an argument that uses other rules of S but not R itself; thus, perhaps one could use inductive and deductive rules to justify the view that abduction should or should not be revised. There would be nothing circular or self‐defeating about that. So if we could do that for claims about each rule of S in turn, we could justify or revise TS without circularity or self‐defeat. And the justification or revision would be naturalistically kosher. This is an attractive idea but I doubt that such a justification or revision of an epistemic rule would be generally available. In thinking about this it is important to remember that S must include rules governing its own potential replacement, rules governing the choice between TS and its rival TS' that recommends a different system S'. It is hard to see how these rules, vaguely indicated by the Page 8 of 21

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

No Place for the A Priori * Neurath metaphor, could themselves be justified or revised in the Neurath way.11 In what follows I shall consider only the circularity problem, setting aside the self‐defeat problem.12 We must start by distinguishing “premise‐circularity” from “rule‐circularity”. An argument is premise‐circular if it aims to establish a conclusion that is assumed as a premise in that very argument. Premise‐ circularity is obviously reprehensible. But my argument for TS is not guilty of it because it does not use TS as a premise. An argument is rule‐circular if it aims to establish a conclusion that asserts that the rules used in that very argument are justified. My argument tries to establish TS which asserts the justification of S, the system used in that argument to establish TS. So the argument is (p.281) certainly rule‐circular. Guided by the Neurath image, the argument accepts the non‐epistemological part of our web for the moment and seeks to justify the epistemological part, TS. And that justification is governed by just the same rules that govern the justification of anything, the rules of S. Rule‐circularity is not obviously reprehensible and some think it is not reprehensible at all (Papineau 1993; Psillos 1999). But there are two reasons for concern about it brought out nicely by Paul Boghossian (2000, 2001). The first reason is that even though rule‐circularity is not blatantly question‐begging like premise‐circularity, it still has a question‐begging air. For, in general, an argument for some conclusion can be criticized “either by questioning one of its premises or by questioning the implicated rule of inference R” (Boghossian 2000: 246). If R is questioned, one would have to defend it by justifying the belief that it is a good rule: one would have to justify TR. So in the rule‐circular case where the conclusion of the argument is TR itself, the use of R may seem to beg the question. At least, as Boghossian points out (2000: 251–3) drawing on Michael Dummett (1991: 202), it seems to beg the question of the skeptic who genuinely doubts TR. Yet it is not clear that it begs the question of someone who does not doubt TR but is simply looking for a justification for his belief in it. The naturalist may feel that he need lose no sleep over failing to satisfy the skeptic. But then there is the second reason for concern about rule‐circularity, what Boghossian calls “bad company”: “unless constraints are placed on the acceptability of rule‐circular arguments, . . . we will be able to justify all manner of absurd rules of inference”, for example Prior's notorious rule of ‘tonk’‐ introduction (2000: 247). As Crispin Wright says, “a rule‐circular ‘justification’ would seem to be available for any rule whatever” (2001: 50). This is unacceptable: we clearly do need some constraint on rule‐circular arguments. So more work needs to be done on the legitimacy of rule‐circularity.13 But this should not be a cause for rejoicing among apriorists. First, it has not been shown that rule‐circularity is always illegitimate or that it is illegitimate in the naturalist's argument for TS. Second, and more important, any justification that Page 9 of 21

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

No Place for the A Priori * the apriorist might offer of TS would also involve a sort of rule‐circularity. Consider the rationalists, BonJour and Bealer, for example. According to them we form some beliefs on the basis (p.282) of a priori intuitions or intellectual seemings, a process that is analogous to the uncontroversial forming of beliefs on the basis of perception. One respect in which the processes are analogous is that they are both noninferential. We might crudely state the rule for the perceptual process: P: In normal conditions, if you have a perceptual experience that p—if it perceptually seems to you that p—then believe that p. The rationalist idea is that there is also a rule that we might crudely state:

I: In normal conditions, if you have an a priori intuition that p—if it intellectually seems to you that p—then believe that p. So, according to the rationalist, I is part of our evidential system S and hence is one of the rules that TS claims to be good. Now the rationalist is as obliged to justify TS as the naturalist. How could he do that? He would have to appeal ultimately to a priori intuitions. And that is indeed what BonJour and Bealer do appeal to, as already noted. So the rationalist justification of TS uses I, one the very rules that, according to the rationalist, TS claims to be good. So the justification is rule‐circular.

The moral of this is that any justification of an epistemological belief in our evidential system, including our logic, whether apriorist or naturalist, will be rule‐circular (assuming that the Neurath way fails). So it is no skin off the nose of my project in this part of the paper that the naturalist justification must be rule‐circular. For, that project is to undermine the motivation for thinking that that there must be a priori knowledge by showing that all beliefs could be justified empirically. Manifestly, the fact that any naturalist's justification of the epistemological belief must be rule‐circular provides no motivation for the a priori given that any apriorist's justification must also be rule‐circular. Of course, there would be a motivation if it could be established that the apriorist's rule‐ circularity is legitimate but the naturalist's is not. But the chances of finding a non‐question‐begging argument to this effect seem close to nil. So the project is intact. Everyone agrees that there is an empirical way of knowing. The Duhem–Quine thesis, supported by the history of science, is that this way of knowing is holistic. I have argued that we have no good reason to think that our troublesome knowledge of mathematics, philosophy, and logic could not be accommodated within this holistic empirical picture. We are far short of a detailed epistemology for this knowledge, of course, but we are far short of a detailed epistemology for any knowledge. Now, if I am right about all this, we have clearly removed the theoretical need to seek another, a priori, way of knowing. This is the first part of the case against the a priori, but it cannot stand alone. The rest of the case is that the whole idea of the a priori is deeply obscure.

Page 10 of 21

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

No Place for the A Priori * (p.283) Many will remain unconvinced of the possibility of an empirical justification of the troublesome knowledge and will continue to think that the justification of this knowledge must be a priori. This thought would be rational if there were any grounds for optimism about the a priori. But, I shall now argue, there are no such grounds, only grounds for pessimism. If this is right, it is not rational to believe in the a priori.

3. Demonstrating the Obscurity 3.1 What is a priori knowledge?

The aim in this part is to show that the whole idea of the a priori is too obscure for it to feature in a good explanation of our knowledge of anything. If this is right, we have a nice abduction: the best explanation of all knowledge is an empirical one. We are presented with a range of examples of alleged a priori knowledge. But what are we to make of the allegation? What is the nature of a priori knowledge? We have the characterization: it is knowledge “not derived from experience” and so not justified in the empirical way; “a warrant . . . is a priori if neither sense experiences not sense‐perceptual beliefs are referred to or relied upon to contribute to the justificational force particular to that warrant” (Burge 1998: 3). But what we need if we are to take the a priori way seriously is a positive characterization, not just a negative one. We need to describe a process for justifying a belief that is different from the empirical way and that we have some reason for thinking is actual. We need some idea of what a priori knowledge is not just what it isn't.14 (p.284) Why? After all, I have been emphasizing how little we know ultimately about empirical justification. So why pick on the a priori?15 The answer is that there are two crucial differences in the epistemic status of the two alleged methods of justification. First, the existence of the empirical method is not in question: everyone believes in it. In contrast, the existence of the a priori way is very much in question. Second, even though we do not have a serious theory of the empirical way, we do have an intuitively clear and appealing general idea of this way, of “learning from experience”. It starts, as noted (sec. 1), from the metaphysical assumption that the worldly fact that p would make the belief that p true. A belief is justified if it is formed and/or sustained by the experiences of a mind/brain in a way that is appropriately sensitive to the putative fact that p. Many instruments—thermometers, voltmeters, etc.—are similarly sensitive to the world. Of course, the mind/brain differs from these instruments: beliefs are much more complex than the “information states” of instruments and their sensitivity to the world is mediated, in a holistic way, by many others. Still, the mind/brain is similar enough to the instruments to make empirical justification quite unmysterious, despite the sad lack of details. In contrast, we do not have

Page 11 of 21

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

No Place for the A Priori * the beginnings of an idea of what the a priori way might be; we lack not just a serious theory but any idea at all. 3.2 The Traditional Analyticity Explanation

The difficulty in giving a positive characterization of a priori knowledge is well‐ demonstrated by the failure of traditional attempts based on analyticity. A typical example of alleged a priori knowledge is our belief that (B) All bachelors are unmarried. Now, according to the tradition,

(1) The content of the concept is the same as that of , thus making (B) analytic. This seemed promising for an account of a priori knowledge because it was thought that, simply in virtue of having a concept, a person was in possession of “tacit knowledge” about the concept; in virtue (p.285) of having , a person tacitly knew (1). So a person's conceptual competence gave her privileged “Cartesian” access to facts about concepts. The required nonempirical process of justification was thought to be one that exploited this access, a reflective process of inspecting the contents of concepts to yield knowledge of the relations between them which in turn yielded such knowledge as (B). This alleged process is that of “conceptual analysis”.

Even if we grant that we have this Cartesian access to conceptual facts like (1), the account fails. For how would a person get from (1) to (B)? By arguing along the following lines. From (1) she infers (2) The content of is the same as that of . From this and

(3) is true, we can then infer

(4) is true, and hence conclude (B). But where did the justification of (3) come from? It does no good to say, rightly, that is a logical truth, for what justifies logical truths? Logical truths were, of course, one of the main things that we were supposed to know a priori. Yet, no satisfactory nonempirical account has ever been given of how we could justify logical truths. And what about the inferences in this argument? (B) will be justified only if the view that these inferences are good is justified (2.4). Where does that justification come from? Without an answer to these questions about the justification of logic we have still not explained a nonempirical way of knowing.

In any case, we should not grant the Cartesian view that conceptual competence gives privileged access to contents, despite its great popularity. I urge a much more modest view of competence according to which it is an ability or skill that Page 12 of 21

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

No Place for the A Priori * need not involve any tacit theory, any semantic propositional knowledge; it is knowledge‐how not knowledge‐that (1996a, 2006d). Why then should we believe the immodest Cartesian view, particularly since it is almost entirely unargued? The content of a person's thought is constituted by relational properties of some sort: “internal” ones involving inferential relations among thoughts and “external” ones involving certain direct causal relations to the world. Take one of those relations. Why suppose that, simply in virtue of her thought having that (p.286) relation, reflection must lead her to believe that it does? Even if reflection does, why suppose that, simply in virtue of that relation partly constituting the content of her thought, reflection must lead her to believe that it does? Most important of all, even if reflection did lead to these beliefs, why suppose that, simply in virtue of her competence, this process of belief formation justifies the beliefs, or gives them any special epistemic authority, and thus turns them into knowledge? These suppositions seem gratuitous. We need a plausible explanation of this allegedly nonempirical process of belief formation and justification. Of course, if one were justified in believing (1), and took logic for granted, then one would have a route to justifying other than the usual empirical route arising from experiencing the nonsemantic world. But the point is that there is no reason to believe in a Cartesian route to justifying (1). The route to justifying (1) would be that of empirical semantics.16 This having been said, the naturalist may, from the perspective of a reliablist epistemology, have to allow a truth in the traditional explanation, albeit not one that is any help to the a priori.17 Suppose that (1), or something similar, really were the case. Then anyone who has the concept might be disposed to believe the necessary proposition . She might be disposed to believe this even though she did not have the Cartesian access to her concepts that would yield semantic knowledge of (1) (or, indeed, an empirical semantic theory that would yield this knowledge). She might be disposed to believe it simply in virtue of the fact that did partly constitute . A consequence of this is that acquiring would be a reliable way of coming to this true belief. So, a reliablist must then allow that her belief is justified (although, of course, she does not know its justification). That would be a truth in the traditional explanation. But this is no help to the a priori. It would show that the empirical process of acquiring a concept involved a process that justifies a necessary belief. But that justification does not differ in any epistemologically significant way from the empirical justification of a contingent belief, for example of : there is still no Cartesian route to justification. Just the same sort of empirically reliable mechanism must be in place in both cases for the beliefs to be justified. The difference between the cases is strictly semantic: if the mechanism appropriate for the justification of is Page 13 of 21

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

No Place for the A Priori * not in place, then the person will not have the concept and hence will not even entertain that proposition; there is no analog of this with .18 Let us turn now to the views of some contemporary apriorists as further evidence of the difficulty of explaining the a priori. I think that we can predict that any attempted explanation will involve Cartesianism and/or taking logic for granted. And if the a priori is left unexplained, it is left mysterious, even mystical. 3.3 Christopher Peacocke

Peacocke follows the tradition in thinking that the sort of understanding that comes with possessing a concept yields a priori knowledge: “it is intuitive that understanding makes available some a priori ways of coming to know propositions” (2005: 751). “The theory of possession conditions is the crucial resource on which truth‐conditional theories need to draw in explaining why certain ways of coming to know are a priori ways” (p. 753). He proposes what he calls a “metasemantic theory of the a priori” which is illustrated in the following claim about conjunction: “the possession condition for the concept of conjunction . . . will entail that thinkers must find the transition from A&B to A compelling” (p. 753). Peacocke thinks that this compulsion reflects tacit knowledge of the nature of the concept that can yield a priori knowledge of the inference from A&B to A. “The moderate rationalist holds that any case of a priori status can be explained as such by appeal to the nature of the concepts involved in the content known a priori” (p. 755). I do not question the compulsion that comes with understanding conjunction. But what is left unexplained is how this compulsion, or anything else about the possession conditions of the concept of conjunction, yields the justification that would turn a belief in the inference from A&B to A into knowledge. Peacocke's account presupposes a Cartesian access to the possession conditions of a concept, an access which I have just argued that we have no reason to believe that we have. But even if we went along with this Cartesianism, it is unclear where the justification would come from. Thus suppose that a person has Cartesian access to the fact that a certain inferential role constitutes the content of a concept. And suppose that this inferential role makes the concept a concept of conjunction. How would that justify the person's view that an inference from A&B to A is good? How do we get from knowledge of the content of a concept to knowledge of logic? It is hard to see how an answer to this will not take our knowledge of logic for granted. (p.288) In sum, I think that Peacocke's account suffers from the same defects as the traditional one.

Page 14 of 21

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

No Place for the A Priori * 3.4 George Bealer

For Bealer, as we have noted, a priori knowledge arises from “rational intuitions”, “intellectual seemings”. He emphasizes that these intuitions have a particular phenomenology: “After a moment's reflection you ‘just see’ ” the truth of De Morgan's Law (1992: 103, see also p. 107; 1999: 30). But what justifies these allegedly a priori intuitions? Bealer, like Peacocke, looks to our grasp of concepts for the answer: the tie “between intuitions and the truth . . . is simply a consequence of what, by definition, it is to possess—to understand—the concepts involved in our intuitions” (1999: 29–30). Bealer distinguishes possessing a concept merely “nominally” from possessing it “determinately”. Possessing it nominally is compatible with the sort of “misunderstanding” exemplied in Tyler Burge's case (1979) of someone who wrongly thinks that arthritis can be in the thigh; and with the sort of “incomplete” understanding exemplified in Burge's case of someone who does not know whether or not a contract must be written. But such misunderstanding or incomplete understanding is incompatible with possessing a concept determinately (1999). In other words, determinate possession requires having a true theory of what the concept is about. Bealer then sets out to give an account of determinate possession. His sensitivity to various problems leads, over many pages, to an account that is fiendishly complicated (pp. 38–47). However, I think that we can abstract from these complications. But, first, a word on the phenomenology. Bealer places a good deal of weight on the phenomenology of having what is thought to be an a priori intuition. This is common among apriorists; see, for example, Alvin Plantinga (1993: 106). What needs to be emphasized is that nothing in the experience of having an intuition supports the view that it is a priori or, indeed, supports any view of what justifies the intuition. In particular, it does not show that the insight is not justified in a holistic empirical way. This theoretical issue is way beyond anything in the phenomenology. Turn now to Bealer's account of the determinate possession of a concept. We can abstract from the complications of this account because everything hinges on one key aspect of it: the view that if having intuition I is partly constitutive of possessing concept C (in the sense defined/explained), and a person possesses C and has I, then I is true. And, of course, Bealer thinks that there are such concepts and intuitions (otherwise there would be no a priori knowledge). Suppose that this were so. Why would I be justified? (p.289) More importantly, what would make that justification a priori? The truth in the traditional analyticity explanation that we allowed earlier (3.2) shows how the justification could be empirical. Coming to possess C is an empirical process. If having I is partly constitutive of possessing C then we could find the justification of I in that empirical process. The very same empirical process leads to both the

Page 15 of 21

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

No Place for the A Priori * justification and the possession. Clearly if the justification of I is to be a priori it has to be found in some other process. Bealer owes us an account of this a priori process. It is helpful to compare what one supposes he must say with the traditional explanation of the a priori. On that explanation, first, Cartesian access was alleged to yield knowledge of the relation between, say, and ; see (1). To get from this to the knowledge that all bachelors are unmarried, (B), we needed, second, to apply knowledge of logic. I objected to both of these features. I take it that if Bealer has an explanation of a priori justification it must be a two‐step one also, but one resting solely on Cartesian access and hence able to do without the appeal to logic. First, simply in virtue of possessing C a person must tacitly know that having I is partly constitutive of this possession and, second, hence must tacitly know that I is true. Now even if one thinks that people have some Cartesian access, it would be bold indeed to think that they have this much. Thus, suppose that we grant the first step, how do we make the massive leap to the truth of I that comes with the second? Remember that Putnam (1975) and two‐factor theorists have proposed theories according to which having certain beliefs (e.g. that all lemons are yellow) is partly constitutive of possessing a concept even though the beliefs may be false. Yet, according to the explanation we are attributing to Bealer, simply in virtue of possessing C and hence knowing that the possession is partly constituted by having I, a person must thereby know that I is true, thus falsifying the semantic theories of Putnam and others. This is very hard to buy. Indeed, it would really amount to little more than the claim that I simply is justified a priori without any explanation of how it is. Perhaps Bealer has in mind some other explanation. 3.5 Laurence BonJour

Finally, let us consider BonJour. He rests a priori justification on “rational insight”: “a priori justification occurs when the mind directly or intuitively sees or grasps or apprehends . . . a necessary fact about the nature or structure of reality” (1998: 15–16). So, our problem of explaining the a priori becomes that of explaining rational insight. Where is the justification to be found in this quasi‐ perceptual process of apprehending a necessary fact? (p.290) BonJour offers the beginnings of an explanation but he does not claim much for it (pp. 180–6).19 Indeed, with admirable frankness, he acknowledges that “we do not presently have anything close” to an adequate explanation of rational insight (2001b: 674). That seems to leave the a priori deeply mysterious. Not according to BonJour: “the supposed mystery pertaining to rationalism . . . has been . . . greatly exaggerated” (1998: 31); allegations that rationalism is “objectionably mysterious, perhaps even somehow occult” “are very hard to take seriously” (pp. 107–8); “the capacity for rational insight,

Page 16 of 21

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

No Place for the A Priori * though fundamental and irreducible, is in no way puzzling or especially in need of further explanation” (p. 16).20 What is the source of this extraordinary confidence in an unexplained and apparently mysterious capacity? It comes partly from the earlier rejected view (2.5) that to deny the a priori is to commit “intellectual suicide”. But it comes also from “the intuitive or phenomenological appearances” of rational insight (p. 107): BonJour thinks that these appearances, when examining examples of alleged a priori knowledge, provide a prima facie case for rationalism that is “extremely obvious and compelling” (p. 99). I have just rejected the force of this sort of appeal to phenomenology (3.4). So, in my view, BonJour leaves the a priori unexplained and mysterious. I shall end with a few more remarks about the hopelessness of explaining it and the extent of the mystery. 3.6 The Mystery of the a Priori

Although we do not have a serious theory of empirical justification, we do have an intuitively clear and appealing idea, an idea that treats the mind/brain as an instrument sensitive, via experience, to the way the world is (3.1). We would certainly like to know a lot more about this but it is not in the least mysterious. The contrast with allegedly a priori justification is stark. What sort of link could there be between the mind/brain and the external world, other than via experience, that would make states of the mind/brain likely to be true about the world? What non‐experiential link to reality could support insights into its necessary character? There is a high correlation between the logical facts of the world and our beliefs about those facts which can only be explained by supposing that there are connections between those beliefs and facts. If those connections are not via experience, they do indeed seem occult. (p.291) At this point, it remains a mystery what it would be for something to be known a priori. Any attempt to remove this mystery must find a path between the Scylla of describing something that is not a priori knowledge because its justification is empirical and the Charybdis of describing something that is not knowledge at all because it has no justification.21 The evidence suggests that there is no such path. Hankering after a priori knowledge is hankering after the unattainable. Our knowledge of many things has not yet been given a satisfactory empirical explanation, just as the evolution of many organisms has not yet been given a satisfactory Darwinian explanation. But it is no more appropriate to respond to the former by claiming that the knowledge is a priori than to respond to the latter by claiming that the evolution is by “intelligent design”. These responses have no scientific substance: they simply label the present absence of a satisfactory explanation. Page 17 of 21

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

No Place for the A Priori * The nice abduction is established: our knowledge of mathematics, philosophy and logic cannot be explained a priori; an empirical explanation of it is the best.22 Notes:

(*) First published in Michael J. Shaffer and Michael L. Veber (eds.), What Place for the A Priori? (Devitt 2009c). Reprinted with kind permission from Open Court Publishing Co. The has been slightly shortened for this chapter by summarizing its discussion of intuitions in section 2.3. A fuller discussion of intuitions can be found in Devitt 2006c, which is Ch. 14 in the present volume. (1) On the naturalistic view, epistemology is part of science. It is important not to misunderstand this, as van Fraassen seems to (2000: 261–71). It goes without saying that epistemology implies the methods of science. But van Fraassen seems to take the naturalist view to be that basic science, or special sciences like biology, medicine, and psychology, imply the methods of science, a view that he rejects. That is not my view of naturalism (1991b: 75–9). I take epistemology to be itself a special science. As such it is no more simply implied by another science than is any other special science: it has the same sort of relative autonomy, and yet dependence on basic science, as other special sciences. So we should not go along with Quine's view that epistemology is a “chapter of psychology” (1969: 82). Naturalized epistemology, like any special science, applies the usual methods of science, whatever they may be, mostly taking established science for granted, to investigate its special realm. In the case of epistemology that realm is those very methods of science. The aim is to discover empirically how we humans learn, and should learn, about the world. We have no reason to suppose that the methods that have yielded knowledge elsewhere cannot yield knowledge in epistemology. (2) I am much indebted to Kim Sterelny and Stephen Stich for eloquently bringing this possibility to my attention. (3) Nor, of course, does a much less interesting broadening to cover beliefs about one's own mental states that are justified by introspection. This discussion of innateness shows that Louise Antony is wrong in sensing a disagreement with me (2005a, b) over whether “a naturalized approach to knowledge requires repudiation of the a priori” (2004: 1). I repudiate an a priori that is committed to a nonnaturalistic justification whereas she defends an a priori that is “innateness plus reliability” (p. 2). I am not against either innateness or a reliablist approach to knowledge. Antony goes on to claim that the a priori she defends “should be pretty satisfying” for those interested in the traditional notion (p. 2). This is surely not so. What has interested nearly all philosophers under the name ‘a priori’ has been a nonnaturalistic way of justifying beliefs, as the citations in this present paper make plain. Furthermore, I would argue that what Antony counts as a priori is not close to being even coextensive with what the tradition counts; Page 18 of 21

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

No Place for the A Priori * in particular, Antony would not count logical beliefs (in contrast to inferences), most philosophical beliefs, and a whole lot of ordinary analytic beliefs, as a priori. I would also argue similarly against Georges Rey's reliablist defense of the a priori (1998), thus adding another criticism to my earlier ones (1998, which is Ch. 12 in the present volume). In general, it seems to me a mistake for Antony, Rey, and other naturalistically inclined philosophers—e.g. Alvin Goldman (1999)—to attempt to “save the a priori”. Epistemological naturalism is indeed a radical doctrine. [2009 addition] Lisa Warenski provides another example of this mistake, just as the present volume goes to press. She argues, with great subtlety, that “a suitably‐modest fallibilist notion of a priori justification is, in principle, compatible with naturalism” (2009: 424). For this to be so, there would have to be a naturalistically respectable account of a nonempirical method of justification and this is something that she does not attempt. However, she does gesture approvingly (p. 421) toward the sort of reliablist account I have criticized in Ch. 12. And she seems to find comfort in the idea that “there may be multiple forms of a priori justification” (p. 408) and hence not “a single unified naturalistic account” (p. 420). But the main problem is that we don't have any good naturalistic account of a priori justification. (4) This is not to deny that we have made progress in epistemology. Indeed, a good deal of the impressive scientific progress in recent centuries has come from learning how better to learn about the world (2005a: 110; Ch. 4, sec. 4.2 in the present volume). Still these advances have not solved the basic epistemological problems. (5) I make a similar response (2005a: 107–8) to the common view that necessities can only be known a priori. There is no reason to believe that if we solved the metaphysical problem of necessity we would not be able to explain our knowledge of necessities empirically. (6) This paragraph was prompted by some very helpful comments from John Bigelow and Kim Sterelny. (7) See also 2006d, ch. 7; 2006e; Kornblith 1998. (8) In actual fact, we surely do not infer simply in accord with modus ponens, as Gilbert Harman has made plain (1999: 18–23): if we believe that p and that if p then q we might indeed infer that q but we might be so convinced of not q that we infer in accord with modus tollens and abandon our belief that p. The relation between psychological processes of inference, even of good inference, and logical implications is complex. Still, modus ponens is surely involved in some of our inferences and those inferences are good only because modus ponens is valid.

Page 19 of 21

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

No Place for the A Priori * (9) Hartry Field has urged on me that, even if the empirical justification of logic has some plausibility, the empirical revision of it has not: we have no reason to believe that any evidential system we would find acceptable would allow this sort of empirical revision of logic. Perhaps not. But then we have no reason to believe that an acceptable system would not allow the revision. Perhaps I would be wiser to follow Field's advice and remain neutral on the matter of empirical revision, resting the naturalist case on the empirical justification of logic. (10) [2009 addition] This was the approach to justifying rules of inference suggested in Ch. 4, sec. 3.1, of the present volume. (11) I discuss this problem elsewhere in responding to Field (Devitt 1998: 61–3; this is Ch. 12 in the present volume). (12) But see Devitt (2005a: 110–11). (13) Boghossian (2000: 248–51) proposes that a rule‐circular argument is in order provided that the rule in question is meaning‐constituting. This presupposes, as Boghossian acknowledges, a conceptual‐role account of the meaning of the logical constants. Wright (2001) is a detailed and interesting critical discussion of Boghossian's proposal. (14) I shall not be concerned with the issue of the fallibility of claims to a priori knowledge, interesting though the issue is. Georges Rey describes the view that such claims might be wrong as “banal fallibilism” (1998: 26) and mocks the idea that traditional rationalists rejected such fallibilism. I am inclined to think that he is wrong about this, given their extreme Cartesianism about the mind. In any case, contemporary rationalists like BonJour and Bealer do not reject fallibilism. This is, of course, wise given the sad history of mistaken claims to a priori knowledge (Kornblith 2000: 67–70). But then, however we understand the view that claims to a priori knowledge can be mistaken, it looks as if those claims could do little epistemic work. One way to understand the view is that, although the process of a priori justification is infallible, yielding outputs that are not open to empirical revision, a person might be mistaken in thinking she has gone through that process and so might make mistaken claims to a priori knowledge. But then for those claims to do any epistemic work we would need evidence, presumably empirical evidence, that they had been arrived at by the approved a priori process. Getting that evidence is surely going to be hard, particularly if I am right in arguing below that we don't know enough about this alleged process to know what to look for. Another way to understand the view is that the a priori process itself is fallible, yielding results that are open to empirical revision. But then it looks as if claims to a priori knowledge could do even less work. For, even if we knew that the claims had been arrived at by the a priori process, we would still need to assess them, in the usual holistic way, against the empirical

Page 20 of 21

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

No Place for the A Priori * evidence to see if they should be revised. (This note was prompted by a helpful correspondence with Hilary Kornblith about his 2000.) (15) Bealer (1999: 52 n. 23) makes a point along these lines. (16) Stipulating contents might provide an unusual example of an empirical semantic route to justifying the likes of (1). A person does not, as a matter of fact, stipulate (1), does not stipulate that the content of her concept is the same as that of , but suppose that she did. If her memory of such matters is reliable then any time later that she remembers the stipulation she will be justified in believing (1). But remembering is an empirical process. (17) I am indebted to Bob Kirk for making me to see this. (18) For a defense of this view, see Devitt 1996a: 30–6; 1997c: 356–8). (19) In my view (2005a: 113), and that of others (Boghossian 2001 and Rey 2001), BonJour's explanation is very unpromising. (BonJour 2005a, b, c and Devitt 2005a, b constitute a debate over the a priori.) (20) Bealer is also anxious to resist the charge that the a priori knowledge arises from “a supernatural power or a magical inner voice or anything of that sort” (1992: 101; see also 1999: 29–30). (21) I argue (1998, which is Ch. 12 in the present volume), in effect, that Rey's attempt (1998) to give a reliablist account of the a priori falls victim to Charybdis. (22) Earlier versions of this chapter were delivered at Macquarie University and at the Eastern Division Conference of the American Philosophical Association in New York, December 2005. I am grateful for comments at those meetings, particularly those of James Pryor.

Access brought to you by:

Page 21 of 21

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Intuitions *

Putting Metaphysics First: Essays on Metaphysics and Epistemology Michael Devitt

Print publication date: 2009 Print ISBN-13: 9780199280803 Published to Oxford Scholarship Online: February 2010 DOI: 10.1093/acprof:oso/9780199280803.001.0001

Intuitions * Michael Devitt (Contributor Webpage)

DOI:10.1093/acprof:oso/9780199280803.003.0015

Abstract and Keywords Intuition mongering is common in the theory of reference and in philosophy generally. Why is this appropriate? And why is it appropriate for linguists to take intuitions as the main evidence for a grammar. The Chomskian answer to the latter question is that the intuitions are derived by a rational process from a representation of linguistic principles in the mind. Stich has suggested (although not endorsed) an analogous answer to the question about referential intuitions. This chapter takes a different view, arguing for a naturalistic and non-Cartesian view of intuitions in general. They are empirical central-processor responses to phenomena differing from other such responses only in being immediate and fairly unreflective. The view yields a naturalistic view of the characteristic method of ‘armchair philosophy’. Keywords:   intuitions, reference, linguistics, Chomskian, Stich, naturalism, Cartesian, empirical, armchair philosophy

1. Introduction Intuition mongering seems to be essential to the characteristic method of “armchair” philosophy. Intuitions are consulted in “thought experiments” to test philosophical hypotheses. Is this way of proceeding appropriate? Many think it is because they take the intuitions to be known a priori. But this is not something that I, and other Quinean naturalists, can accept because we think that there is no a priori knowledge.1 So, is it in order for naturalistic philosophers to consult intuitions?

Page 1 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Intuitions * Consider the philosophy of language in particular, an area of philosophy with a heavy reliance on intuitions. It is common to think that the task of, for example, the theory of reference is simply to systematize our ordinary intuitions about reference. Stephen Stich sums this view up as follows: “the theory of reference is attempting to capture the details of a commonsense theory about the link between words and the world” where that theory involves, at least, a generalization of the intuitions. Stich thinks that this view is “favored, albeit tacitly, by most philosophers” (1996: 6; the view is not endorsed by Stich). I think he is right about that. Still, this common view is puzzling. It is puzzling because the obvious way to describe the task of the theory of reference is that it is to explain the nature of reference, to explain the nature (p. 293) of a certain word–world relation. If we start from this view, surely as good a starting place as one could have, why take the task to be to capture the folk theory of this relation? That would seem to be appropriate only if we assume that the folk must be right about reference. But why assume that? Why think that the folk have infallible insight into the nature of this particular word–world relation? We don't suppose that they are authorities on physics, biology, or economics, why suppose that they are on semantics? Stich has a neat suggestion (p. 40). Folk semantics might be viewed the way linguists standardly view folk linguistics. Their standard view is that speakers derive their grammatical intuitions about their language from a representation of the grammatical principles of the language; the intuitions are, as I put it, “the voice of competence”.2 We might then take a similar view of referential intuitions: speakers derive them from a representation of referential principles. So, just as, according to the linguists, the true grammar that they seek to discover is already stored in the mind of every speaker, so too, according to this view, is the true theory of reference. Linguistic intuitions, whether about syntax or reference, are not the result of the sort of empirical investigation that judgments of the world usually require. Rather, we might say, speakers have a “Cartesian access” to facts about their language simply in virtue of being competent in it and thus embodying representations of its principles. Someone who took the standard linguists' view of linguistic intuitions might well be tempted to this analogous view of referential intuitions. So, it is interesting to note that Noam Chomsky is not tempted. He expresses skepticism about “contemporary philosophy of language” and its practice of “exploring intuitions about the technical notions ‘denote’, ‘refer’, ‘true of’, etc.”. He claims that there can be no intuitions about these notions, just as there can be none about ‘angular velocity’ or ‘protein’. These are technical terms of philosophical discourse with a stipulated sense that has no counterpart in ordinary language. (1995: 24)

Page 2 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Intuitions * So Chomsky is skeptical about the use philosophers make of referential intuitions. But he is not, of course, similarly skeptical about the use linguists (p.294) make of linguistic intuitions. Why the difference? If skepticism about referential intuitions is appropriate, then surely just the same skepticism is appropriate about the linguistic ones, and for just the same reason. All the terms in linguistic theory are, in the relevant sense, technical and theory‐laden. A few like ‘grammatical’ and ‘sentence’ may have counterparts in ordinary language but this need not give intuitions deploying those counterparts the privileged status that would arise from their being the voice of competence. Those intuitions could simply be the result of years of empirical folk linguistics. And that is what I shall argue that they are.

In the next section, I shall present a view of intuitions in general. In the final section I shall apply this to intuitions in linguistics, philosophy of language, and philosophy generally.3

2. Intuitions in General On my view, the intuitions that concern us are judgments that are empirical theory‐laden central‐processor responses to phenomena, differing from many other such judgments only in being immediate and unreflective, not based on any conscious reasoning. These intuitions are surely partly innate in origin4 but are usually and largely the result of past reflection on a lifetime of worldly experience.5 A clarification. It may be that there are many unreflective empirical judgments that we would not ordinarily call intuitions: one thinks immediately of perceptual judgments like ‘That grass is brown’ made on observing some scorched grass, or ‘That person is angry’ made on observing someone exhibiting many signs of rage. Perhaps we count something as an intuitive judgment only if it is not really obvious. I shall not be concerned with this. My claim is that intuitions are empirical unreflective judgments, at least. Should more be required to be an intuition, so be it. (p.295) In considering intuitions and their role in science, it is helpful to distinguish the most basic intuitions from richer ones. Suppose that we are investigating the nature of a kind F—for example, the kind gene, pain, or echidna. The most basic intuitions are ones that identify Fs and non‐Fs; for example, “This is an echidna but that isn't”. The richer ones go on to tell us something about Fs already identified; for example, “Echidnas are monotremes that look like porcupines”. The richer ones may be much less dependable than the basic ones: a person may be good at recognizing F's without having much reliable to say about them; this is very likely the situation of the folk with pains. Identifying uncontroversial cases of Fs and non‐Fs is only the first stage of an investigation into the nature of Fs: the second stage is to examine those cases to see what is common and peculiar to Fs. Sometimes we have a well‐established theory to help with the first stage; thus we had Mendelian genetics to identify the genes that were examined by molecular genetics in the second stage. But Page 3 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Intuitions * sometimes we do not have such help: we start pretty much from scratch; we are at the stage of proto‐science. At that stage, the most basic intuitions are particularly important. In the absence of reliable theory, we must start by consulting the people who are most expert about F's to see what they identify as Fs and non‐Fs: we elicit their most basic intuitions about being an F in “identification experiments”. We are then in a position to begin our investigation. Until recently, at least, this was our position with pains. When we are starting from scratch, we need the basic intuitions, but we do not need the richer ones. This is not to say that we should not use them. They may well be a useful guide to what our investigation will discover about Fs; they are “a source of empirical hypotheses” (Gopnik and Schwitzgebel 1998: 78). We should trust a person's intuitions, whether basic ones or richer ones, to the degree that we have confidence in her empirically based expertise about the kinds under investigation. Sometimes the folk may be as expert as anyone: intuitions laden with “folk theory” are the best we have to go on. Perhaps this is the case for a range of psychological kinds. For most kinds, it clearly is not: we should trust intuitions laden with established scientific theories. Consider, for example, a paleontologist in the field searching for fossils. She sees what seems to be a bit of white stone sticking through grey rock, and thinks “a pig's jawbone”. This intuitive judgment is quick and unreflective. She may be quite sure but unable to explain just how she knows.6 We trust her judgment in a way that we would not trust folk judgments because we know that it is the (p.296) result of years of study and experience of old bones; she has become a reliable indicator of the properties of fossils. Similarly we trust the intuitions of the physicist over those of the folk about many aspects of the physical world where the folk have proved notoriously unreliable. And recent experiments have shown that we should have a similar attitude to many psychological intuitions. Thus, the cognitive psychologist, Edward Wisniewski, points out that researchers who study behavior and thought within an experimental framework develop better intuitions about these phenomena than those of intuition researchers or lay people who do not study these phenomena within such a framework. The intuitions are better in the sense that they are more likely to be correct when subjected to experimental testing. (1998: 45) Even where we are right to trust an intuition in the short run, nothing rests on it in the long run. We can look for more direct evidence in scientific tests. In such a scientific test we examine the reality the intuition is about; for example, we examine the paleontologist's white stone. These scientific examinations of reality, not intuitions about reality, are the primary source of evidence. The examinations may lead us to revise some of our initial intuitions. They will surely

Page 4 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Intuitions * show us that the intuitions are far from a complete account of the relevant bit of reality. Intuitions often play a role in “thought experiments”. Instead of real experiments that confront the expert with phenomena and ask her whether they are Fs, we confront her with descriptions of phenomena and ask her whether she would say that they were Fs.7 These thought experiments provide valuable clues to what the expert would identify as an F or a non‐F. They can do more: the descriptions that elicit the expert's response indicate the richer intuitions that, as we have already noted, can be a useful guide to the nature of Fs. Some experiments may be difficult, perhaps impossible, to perform other than in thought. Valuable and useful as thought experiments may be in practice, they are dispensable in principle: we can make do with real experiments. And thought experiments call on the same empirically based beliefs about the world as real experiments, and their results have the same empirical status. The view I have presented of the limited and theory‐laden role of intuitions does not need to be modified because of the special situation in the philosophy of language and linguistics, a situation where what we are investigating are the products of a human skill or competence. This situation arises elsewhere; for (p. 297) example, if we are (for whatever reason) investigating the nature of horseshoes, chess moves, touch typing, or thinking. Someone who has the relevant competence has ready access to a great deal of data that are to be explained. She does not have to go out and look for data because her competence produces them. Not only that, she is surrounded by similarly competent people who also produce them. As a result, she is in a good position to go in for some central‐processor reflection upon the data produced by herself and her associates. This reflection, often aided by appropriate education, can yield concepts and a theory about the data. And it can yield the capacity for sound intuitions, basic and richer, about the data. In brief, she can become an expert. But this is not to say that she will become an expert. A person can be competent and yet reflect little on the output of that competence. Or she can reflect a lot but make little progress. Bicycle riders typically fall into one of these two categories. It is a truism in sport that great players do not always make great coaches. The fact that they possess a competence to a superlative degree does not imply that they can articulate and communicate the elements of that competence. Knowledge‐how may not lead to knowledge‐that.8 Even if a competent person does become an expert, we should not assume that her opinions carry special authority simply because she is competent; her competence does not give her Cartesian access to the truth. She is privileged in her ready access to data, not in the conclusions she draws from the data; conclusions of the competent, just like those of the incompetent, are empirical responses to the phenomena and open to question; they arise from the empirical observation of data.

Page 5 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Intuitions * Touch typing provides a nice example of reflecting on the output of one's own competence. Ask a touch typist whether a ‘k’ should be typed with a middle finger and, very likely, he will think to himself, “How would I type a ‘k’?” He will attend as he goes through the actual or mental motions of doing so and respond immediately, “Yes”. Consider also this report: If a skilled typist is asked to type the alphabet, he can do so in a few seconds and with very low probability of error. If, however, he is given a diagram of his keyboard and asked to fill in the letters in alphabet order, he finds the task difficult. It requires several minutes to perform and the likelihood of error is high. Moreover, the typist often reports that he can only obtain the visual location of some letters by trying to type the letter and then determining where his finger would be. (Posner 1973: 25)9 (p.298) The only privilege enjoyed by the typist's judgment about which finger should be used to type a ‘k’, or about where a letter is placed on the keyboard diagram, is the privilege of being based on what is surely a good datum: on how he, a good touch typist, types.

Although these typist's judgments are slow relative to his typing, they would probably be fast enough for us to count them as intuitive. And they are likely to be sound, for it is fairly easy to think about typing. Contrast this with reflecting upon the outputs of another, much more important, human competence, the competence to think. We all have this competence to some degree or other: we can move in a somewhat rational way from one thought to another. Most of us reflect a bit on this and have some intuitions about what follows from what. Still, these intuitions are likely to be sparse and many of them are surely not sound. Thinking about thinking is so hard. Now it is, of course, possible that the typist has somewhere in his mind a prior representation of the keyboard which controls his typing and leads to his sound judgment about how to type a ‘k’. But why believe this? Set aside whether we need to posit this representation to explain his typing. We surely do not need the posit to explain his judgment. The much more modest explanation I have just given, making do with cognitive states and processes that we are already committed to, seems perfectly adequate for the job. Positing the prior representation is explanatorily unnecessary. Finally, when we turn to the case of thinkers, such positing would seem worse than unnecessary. The idea would have to be that the thinker's mind contains a representation of the “laws of thought” which controls her thinking and which leads her to, say, the intuition that ‘q’ follows from ‘if p then q’ and ‘p’. But, as Lewis Carroll's famous example of Achilles and the Tortoise demonstrates, this view of thinking would lead to an infinite regress. The modest explanation is the only plausible one: a person's thinking is governed by rules that she does not represent and

Page 6 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Intuitions * her few intuitive judgements about thinking are the result of reflecting on the performances of herself and others. On the modest picture of intuitions I am presenting, what should we make of philosophical and linguistic intuitions? And whose intuitions should we most trust?

(p.299) 3. Philosophical and Linguistic Intuitions Let us start with philosophical intuitions in general and the characteristic “armchair” method of philosophy. The traditional explanation of this method is that philosophers are conducting thought experiments that probe their concepts to yield a priori rational intuitions; they are doing “conceptual analysis”. The naturalistic explanation accepts that philosophers are conducting thought experiments but construes these differently. The philosophers are not probing concepts but rather intuitions about kinds. This is just as well because knowledge of concepts, being a species of semantic knowledge, is very hard to come by. In contrast, philosophers have acquired considerable knowledge of many kinds over a lifetime of acquaintance with them. The philosophers' intuitions that draw on this knowledge are not a priori but empirical. The philosophers are conducting thought experiments of the sort described earlier, counting themselves as experts about the kind in question. Thus, in a famous example of the method, “the analysis of knowledge”, the philosopher, as expert as anyone in identifying cases of knowledge, confronts descriptions of epistemic situations and considers whether the situations are cases of knowledge. On the basis of these empirical intuitions about cases she constructs an empirical theory about the nature of knowledge. The naturalist does not deny armchair intuitions a role in philosophy but does deny that their role has to be seen as a priori: the intuitions reflect an empirically based expertise in the identification of kinds. Turn next to linguistics and the philosophy of language. The competent speaker has ready access to a great deal of linguistic data just as the competent typist has to a great deal of typing data and the competent thinker has to a great deal of thinking data: the competent speaker and her competent fellows produce linguistic data day in and day out.10 So she is surrounded by tokens that may, as a matter of fact, be grammatical, be ambiguous, have to corefer with a certain noun phrase, and so on.11 So she is in a position to have well‐based (p.300) opinions about language by reflecting on these tokens. This is not to say that she will reflect. Indeed, a totally uneducated person may reflect very little and hence have few if any intuitive judgments about her language. Still it is clear that the normal competent speaker with even a little education does reflect on linguistic reality just as she reflects on many other striking aspects of the world she lives in. And this education will usually provide the terms and concepts of folk linguistics, at least. As a result she is likely to be able to judge in a fairly immediate and unreflective way that a token is grammatical, is ambiguous, does Page 7 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Intuitions * have to corefer with a certain noun phrase, and so on. Such intuitive opinions are empirical central‐processor responses to linguistic phenomena. They have no special authority: although the speaker's competence gives her ready access to data it does not give her Cartesian access to the truth about the data.12 Still, are these intuitive judgments likely to be right? Are the folk expert enough? They surely are if we stick to the simplest intuitions involving vocabulary, particularly syntactic and semantic vocabulary, that we suppose the folk have mastered well enough. So we can usually be confident about judgments that a string of words is or is not “acceptable” and that a pronoun might “refer” to x or to y. We can often be confident about judgments that one expression does or does not “mean the same” as another, and perhaps even that a string is or is not “grammatical”. And we can be very confident of judgments in “minimal‐pair” experiments where ordinary speakers are asked to say simply which of two word strings, differing only in that one fails a certain hypothesized syntactic constraint, is “worse”.13 In sum, we have good reason to suppose that the core judgments of folk linguistics, reflecting the “linguistic wisdom of the ages”, are good evidence for linguistic theories. This having been said, the intuitions that linguistics, at least, should mostly rely on are those of the linguists themselves because the linguists are the most expert. This is particularly so when we get beyond the simple cases to theoretically interesting ones like ‘The horse raced past the barn fell’ and ‘Who do you wanna kiss you this time?’ The linguists' skill at identifying items with and without a syntactic property like, say, the biologist's skill at identifying items with and without a biological property, is likely to be better than the folk's because their theories are better. Thus linguists have firm, and surely correct, intuitions about the acceptability of many sentences, and about some (p.301) matters of co‐reference, that the folk do not.14 Linguistic theory is, as linguists are fond of pointing out, in good shape, far better shape than semantic theory. As a result of their incessant observation of language, guided by a good theory, linguists are reliable indicators of syntactic reality; analogously, biologists are reliable indicators of biological reality. To say that intuitions, whether those of the linguists or the folk, are good evidence is not to say that they are the only, or even the primary, evidence. Indeed, we can look for more direct less theory‐laden evidence by studying what the intuitions are about, the linguistic reality itself. In fact, despite what the literature often suggests,15 there are many other sources of evidence. Still, linguistic intuitions, particularly those of the linguists, are often good evidence. So, they should be used. Intuitions are often a very convenient shortcut in theorizing.

Page 8 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Intuitions * In light of this modest explanation of intuitions and their reliability, I think that we should reject the standard view of linguistic intuitions and Stich's analogous suggestion (which he does not endorse) for referential intuitions. It is of course possible that the competent speaker's intuitions are reliable because they are derived from her mental representations of linguistic and referential principles. It has been argued, mistakenly in my view (2006d), that we need to posit such representations to explain language acquisition and use. In any case, I urge now that we surely do not need the posit to explain the reliability of linguistic and referential intuitions. Consider the analogous phenomena for typing and thinking. We can explain the reliability of intuitions about those processes without positing representations of the rules that govern the processes. Our explanations of these intuitions make do with cognitive states and processes that we are already committed to. These modest explanations seemed perfectly adequate for the job and, indeed, much more plausible than their representational rivals. So do the similarly modest explanations in the linguistics case. Language is a very striking and important part of the human environment. It is not surprising that empirical reflection on linguistic data, aided by some education, should make people fairly reliable detectors of the (p.302) most obvious facts about language. We are surely similarly reliable about other striking and important parts of the environment, for example, the physical, biological, and psychological parts.

4. Conclusion Philosophers typically take ordinary intuitions to be good evidence for a theory because they think that they are the result of a priori insight into concepts. Linguists typically take ordinary intuitions to be good evidence for a grammar because they think that they are derived by a rational process from a representation of linguistic principles. I have argued for a different view that has the great advantage of being theoretically modest: it treats these intuitions like intuitions in general as empirical central‐processor responses to phenomena. The view accommodates the evidential role that intuitions play in philosophy and linguistics without resort to the a priori or the mental representation of linguistic principles. Notes:

(*) First delivered at the International Ontology Congress VI, “From the Gene to Language: the State of the Art”, held in San Sebastian in Sept. 2004, and published in its proceedings, Victor Gomez Pin, Jose Ignacio Galparaso, and Gotzon Arrizabalaga (eds.), Ontology Studies Cuadernos de Ontologia: Proceedings of VI International Ontology Congress (San Sebastian, 2004): Devitt 2006c. Reprinted with kind permission from Universidad del Pais Vasco. (1) The best reason for rejecting a priori knowledge, I have argued (1996a; 1998, which is Ch. 12 in the present volume; 2009c, which is Ch. 13), is that we do not have even the beginnings of an account of what it is. We are simply told what it Page 9 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Intuitions * isn't, namely empirical knowledge. Bealer (1998) and BonJour (1998) are vigorous defenders of rational intuitions; see also Sosa 1998. For an exchange on the subject, see BonJour 2005a, b, c and Devitt 2005a, b. (2) Consider the following passages, for example: it seems reasonably clear, both in principle and in many specific cases, how unconscious knowledge issues in conscious knowledge . . . it follows by computations similar to straight deduction. (Chomsky 1986: 270; see also Pateman 1987: 100; Dwyer and Pietroski 1996: 342) we cognize the system of mentally‐represented rules from which [linguistic] facts follow. (Chomsky 1980: 9: the facts are expressed in intuitive judgments) Our ability to make linguistic judgments clearly follows from our knowing the languages that we know. (Larson and Segal 1995: 10; see also Pylyshyn 1984: 122; Baker 1995: 20) (3) I draw on earlier works (1994; 1996a: 48–86). See also Kornblith 1998. For a much more thorough treatment of linguistic intuitions see my 2006d: ch. 7, and 2006e. (4) In calling the intuitions “empirical” I am claiming simply that they must be justified “by experience”. Should any justified belief be entirely innate, which I doubt, then beliefs of that sort must have been justified somehow by the experiences (broadly construed) of our distant ancestors, and we must have inherited that justification via natural selection. [2009 addition]. See Ch. 13, sec. 1, in the present volume for a discussion. (5) “. . . intuition is the condensation of vast prior analytic experience; it is analysis compressed and crystallized. . . . It is the product of analytic processes being condensed to such a degree that its internal structure may elude even the person benefiting from it . . .” (Goldberg 2005: 150). (6) I owe this nice example to Kim Sterelny. Gladwell 2005 has other good examples: of art experts correctly judging an allegedly 6th‐century Greek marble statue to be a fake; of the tennis coach, Vic Braden, correctly judging a serve to be a fault before the ball hits the ground. (7) There are other things we might ask—e.g. “What would happen?”—but these are beside out concerns. Gendler 2003 is a nice summary of views about thought experiments. (8) “Highly skilled performers are often unable to reflect on or talk about how they achieve their skilled performance” (Carlson 2003: 38).

Page 10 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Intuitions * (9) And consider this report (Sun et al. 2001). Subjects were placed in front of a computer with the task of navigating a submarine through a minefield using sonar. After some episodes, “subjects were asked to step through slow replays of selected episodes and to verbalize what they were thinking during the episode” (p. 219). The experimenters sum up the results as follows: “The subject at first performed the task on an “instinctual” basis, without conscious awareness of any particular rules or strategies. Gradually, through “doing it” and then looking at the results, the subject was able to figure out the action rules explicitly. The segment suggested implicit procedural learning at the bottom level and the gradual explication of implicitly learned knowledge” (p. 226). (10) As Chomsky says, competent speakers “can easily construct masses of relevant data and in fact are immersed in such data” (1988: 46). (11) This presupposes a realism about the linguistic entities that, according to my 2003 and 2006d, are the subject matter of linguistics. This realism is curiously denied by some. I think that this denial is a mistaken reaction to two facts: first, that the properties in virtue of which something is a linguistic token are all relational; second, that tokens of the one linguistic expression can appear in a variety of physical forms, a variety of sounds, a variety of inscriptions, and so on. Yet something can really have a certain linguistic property just as something can really have a certain nationality even though neither have these properties intrinsically and even though other things that have them can differ greatly in their physical forms. [2009 addition] See my 2008a for further discussion. (12) I emphasize that this is a modest explanation of the origins of a speaker's intuitions about her language. It is emphatically not an explanation of the origins of her linguistic competence and is neutral about the extent to which that competence is innate. (13) See Crain and Thornton 1998 for a helpful discussion of experiments of this sort. (14) Subjects in an experiment (Spencer 1973) considered 150 sentences that linguists had categorized as clearly acceptable or unacceptable. The subjects disagreed with the linguists over 73 of these, either finding them unclear or giving them an opposite categorization. In another experiment (Gordon and Hendrick 1997), naïve subjects found co‐reference between a name and a pronoun that preceded it unacceptable even where the pronoun did not c‐ command the name. This is one of several experiments where folk intuitions were discovered to be at odds with the linguists' and with Binding Theory.

Page 11 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Intuitions * (15) Liliane Haegeman claims that “all the linguist has to go by . . . is the native speaker's intuitions” (1994: 8) but two pages later she allows, somewhat grudgingly, an evidential role for usage. Andrew Radford opens his book (1988) with an extensive discussion of the evidential role of intuitions. The first mention of the use of the “corpus of utterances” as data does not come until p. 24. Robert Fiengo starts an interesting paper on linguistic intuitions: “Intuitions, with the contents that they have, are the data of Linguistics” (2003: 253).

Access brought to you by:

Page 12 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

On Determining What There Isn't *

Putting Metaphysics First: Essays on Metaphysics and Epistemology Michael Devitt

Print publication date: 2009 Print ISBN-13: 9780199280803 Published to Oxford Scholarship Online: February 2010 DOI: 10.1093/acprof:oso/9780199280803.001.0001

On Determining What There Isn't * Michael Devitt (Contributor Webpage)

DOI:10.1093/acprof:oso/9780199280803.003.0016

Abstract and Keywords Stich points out that typical arguments for eliminativism about the mind presuppose a description theory of reference. His gloomy view of theories of reference leads him to a gloomy view of ontological issues in general. What has gone wrong? Stich thinks that the mistake came in looking to such theories to settle ontological issues. This chapter argues that Stich is right about that, but for the wrong reasons. We should not share his gloom about theories of reference, taken as proto-science. Still, an appropriately modest view of the accomplishments of this proto-science counts against using it to settle ontological issues. We should put metaphysics first. How then are we to settle ontological issues? Stich's answer is very much along the right lines. We do not have any principles adequate to help us with the difficult cases and it may be that some of these are indeterminate. Keywords:   Stich, eliminativism, mind, description theories of reference, ontology, proto-science, ontological issues, indeterminacy

In his engaging essay, “Deconstructing the Mind” (1996: 3–90), Stephen Stich raises some very good questions and gives some pretty good answers. My aim in this paper is to give some answers of my own, drawing on earlier work, and to compare these answers with Stich's.

Page 1 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

On Determining What There Isn't * 1. Introduction In his misspent youth Stich was attracted to arguments for eliminativism about the mind—arguments against “intentional realism”. The arguments ran along the following lines: Beliefs, etc., are posits of folk psychology; Folk psychology is a seriously mistaken theory; So, beliefs, etc., do not exist. Yet, as he now points out, such an argument needs an extra premise. How does the falsity of a theory establish the nonexistence of the entities it purports to be about? Implicitly or explicitly, the extra premise is usually an application of the description theory of reference; for example, the premise that if ‘belief’ refers, it refers to whatever is picked out by the description associated with it in folk psychology. Now, if folk psychology is seriously mistaken that associated description does not pick out anything. So ‘belief’ does not refer. So beliefs do not exist. Stich was tempted by such arguments. But then along came William Lycan to wake Stich from his “dogmatic slumbers” (5).1 Lycan (1988) (p.304) pointed out that the description theory of reference is not the only theory of reference in town: one might, for example, adopt a causal‐historical theory of the sort made popular by Saul Kripke (1980) and Hilary Putnam (1975). A causal‐historical theory does not require a word that refers to x to be embodied in a theory that is largely true of x. Indeed, the theory was born out of “arguments from ignorance and error” that showed that description theories of some words placed far too heavy an epistemic burden on competent users of those words. According to the causal‐historical theory, a word that has the right sort of causal link to x will refer to x however wrong the theory of xs. So, if a causal‐historical theory applied to ‘belief’ it could refer even though folk psychology was seriously mistaken. So beliefs could exist despite our mistakes about them.

I like another reason for being dubious of the use of description theories in realism debates: even if such theories are true for some words, they cannot be true for all words. Description theories explain the reference of a word in terms of the reference of the other words in the description with which the word is associated; the reference of those other words determine what the description picks out. How then are we to explain the reference of those other words? Perhaps we can appeal to a description theory again but clearly such appeals must ultimately come to an end, for description theories simply pass the referential buck. The buck must stop with a different sort of theory, one that explains the reference of a word in terms of direct, presumably causal, relations to reality. Perhaps the words relevant to a realism debate are of that sort. So I think that Lycan and Stich are doing a service in emphasizing the way description theories of reference are often, and inappropriately, simply taken for granted in reaching antirealist conclusions. The most regrettable examples of this procedure, in my view (1984/1991b; 2005c, which is Ch. 4 in the present volume), are to be found in discussions of Putnam's “pessimistic meta‐induction” in the philosophy of science (1978). This meta‐induction is an argument against scientific realism: briefly, the unobservables posited by past theories do not Page 2 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

On Determining What There Isn't * exist; so, probably the unobservables posited by present theories do not exist. Since past theories are clearly mistaken—that's why they are past—a description theory of reference makes the premise of this argument seem rather obvious. But a causal theory (or perhaps a part‐causal part‐description theory) of reference is often more plausible for the words of past theories. If we adopt a causal theory of reference then the premise of the meta‐induction is likely to seem overstated, at least.2 (p.305) Although a description theory of reference should certainly not be taken for granted, I think that a description theory is plausible for terms like ‘belief’. So, I think that if intentional realism is to be saved we are likely to have to reject the view that folk psychology is seriously mistaken. In this respect, I differ from Lycan (1988: 31–2). This immediately raises a question. How should we settle such differences of opinion over reference? As Stich points out, there is nothing special about this epistemic question about reference: similar epistemic questions arise in theorizing about anything. Still, pondering his awakening, Stich thinks that reference raises some special metaphysical/methodological questions (pp. 6, 37–8). What makes a theory of reference true or false? What is a theory of reference supposed to do? Stich's discussion of these questions (pp. 37–51) yields a gloomy conclusion about reference and hence a gloomy conclusion about the importance and interest of the eliminativism issue. Worse still, as John Searle pointed out to Stich, this gloomy conclusion generalizes to other issues of existence, for example, of black holes and the big bang (pp. 51–4). Surely these central scientific issues are not unimportant and uninteresting. Something has clearly gone wrong. But what? Stich decides that the mistake came in the “semantic ascent” at the beginning, in the proposal “that questions about the existence of entities posited by false theories could be productively addressed by focusing on what the theory of reference tells us about the terms used in such theories” (p. 55). Stich concludes (pp. 60–82) by considering some nonsemantic approaches to the issue of eliminativism. I think that Stich is absolutely right that it was a mistake to begin with semantic ascent. Indeed, my Realism and Truth (1984/1991b) is an extended argument for tackling metaphysical/ontological issues before semantic ones;3 we should adopt the slogan, “put metaphysics first”. However, my reason for resisting semantic ascent is a bit different from Stich's. His reason is his gloomy conclusion about the theory of reference. I do not share his gloom. Still, I think that an appropriately modest view of the accomplishments of the theory of reference goes decisively against the semantic ascent approach to metaphysical/ ontological issues. I shall discuss this in section 4.

Page 3 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

On Determining What There Isn't * But, first, I want to discuss the theory of reference. Although my view of this theory is not rosy, it is much more so than Stich's. I shall discuss this in (p.306) the next two sections. Finally, in section 5, I shall briefly consider nonsemantic approaches to eliminativism, taking a view very like Stich's.

2. The Theory of Reference as Folk Semantics The deep problem with the theory of reference is not, Stich thinks, the usual epistemic one but rather a special metaphysical/methodological one. But is there really such a problem? One might think not because, at first glance, his metaphysical/methodological questions that pose the problem seem to have obvious answers. What makes a theory of reference true or false? Well, the nature of the reference relation does. What is a theory of reference supposed to do? Well, characterize that nature. So where is the special problem? Perhaps these obvious answers are a bit naive. We shall soon see. Stich thinks that there are “two quite different stories to be told about what a theory of reference is up to”: “the proto‐science account” and “the folk semantic account” (p. 6). Stich offers criticisms of both accounts whilst remaining fairly neutral on the choice between them. I shall begin by rejecting the folk semantic account. The inspiration for the folk semantic account comes from our heavy reliance on ordinary intuitions in theorizing about reference. According to the folk semantic account, “the theory of reference is attempting to capture the details of a commonsense theory about the link between words and the world” (p. 6), where that theory involves, at least, a generalization of the intuitions. Stich thinks that this account is “favored, albeit tacitly, by most philosophers” (p. 6). I think he is right about that. Still, this common account is puzzling. I have argued against it at some length elsewhere, particularly in Coming to Our Senses (1996a: 48–86; also 1994), and so will be brief here. According to my obvious answer above, the task of the theory of reference is to characterize reference; that is, the task is to characterize a certain word–world relation. If we start from this view, surely as good a starting place as one could have, why take the task to be to capture the folk theory of this relation? That would seem to be appropriate only if we assume that the folk must be right about reference. But why assume that? Why think that the folk have infallible insight into the nature of this particular word–world relation? We don't suppose that they are authorities on physics, biology, or economics, why suppose that they are on semantics? Stich has a neat suggestion (p. 40). Folk semantics might be viewed the way linguists standardly view folk linguistics. Their standard view is that speakers (p.307) derive their grammatical intuitions about their language from a representation of the grammatical principles of the language; the intuitions are, as I put it, “the voice of competence”. We might then take a similar view of Page 4 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

On Determining What There Isn't * referential intuitions: speakers derive them from a representation of referential principles. So, just as, according to the linguists, the true grammar that they seek to discover is already stored in the mind of every speaker, so too, according to the folk semanticists, is the true theory of reference. Linguistic intuitions, whether about syntax or reference, are not the result of the sort of empirical investigation that judgments of the world usually require. Rather, we might say, speakers have a “Cartesian access” to facts about their language simply in virtue of being competent in it and thus embodying representations of its principles. This would support the widely held view that semantics is a priori. But why should we accept the standard linguistic view? I have argued that the idea that the grammar is represented in the minds of speakers is implausible and unsupported. And even if the standard linguistic view were right, we would still need an independent reason to accept the folk semanticist view; mere analogy is not enough. I don't think that there is any such reason.4 If we do not view folk semantics the way linguists view folk linguistics, what are we to say about folk semantics? As Stich points out, we might see it as analogous to folk physics. But then, “there will be no guarantee that our internalized folk semantic theory is correct”. Our attempt to describe it “may be an interesting bit of psychology, but there is no reason to suppose that it will tell us much about reference” (43). Exactly! In my view, Stich has described just the attitudes and opinions we should have about folk semantics. We should see the semantic theory and intuitions of the folk as fallible and incomplete empirical responses to the phenomena. We should have the same attitudes to them as we have to folk theories and intuitions about physics, biology, psychology, or whatever.5 This having been said, I do think that there is a small truth underlying the folk semantics account. The truth is that folk intuitions about reference play a role at the very beginning of the theory of reference. The first stage of a theory of any property F or relation R involves identifying some apparently uncontroversial examples where F or R is instantiated and some apparently (p.308) uncontroversial examples where it is not instantiated. These examples can then be examined in the second stage to discover what is common and peculiar to F or R in the hope of determining its nature. In that first stage we should consult those most expert at identifying cases of F or R. If we are concerned with, say, being a gene, being an echidna, or being an isotope of, then we can look to scientists for the identification. But when our concern is with being referred to by, it is doubtful that anyone is more expert at identification than the folk.6 So these most basic folk intuitions about reference, intuitions that identify paradigm cases of it, play a role in the first stage of the theory of reference. But this role does not go against what I have just claimed about folk intuitions. Even these basic intuitions are not infallible. Theorizing at the second stage can lead to the rejection of results at the first stage: apparently uncontroversial examples turn out to be controversial; whales are not fish after all; tomatoes are not Page 5 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

On Determining What There Isn't * vegetables; unacceptable strings of words turn out to be grammatical. There is even less reason to think that any richer folk intuitions or theories about the nature of reference must be true. So, despite the undoubted popularity of the folk semantic account, I regard it as a nonstarter. I shall therefore set it aside without considering Stich's two criticisms of it: that the account makes reference indeterminate; and that it makes reference “idiosyncratic and uninteresting” (pp. 47–8). Without more ado, I turn to the proto‐science way of responding to Stich's metaphysical/ methodological problem about the theory of reference.

3. The Theory of Reference as Proto‐Science Stich describes the proto‐science account as follows: “the theory of reference is attempting to characterize a word‐world mapping that will be useful in one or another empirical discipline such as linguistics, cognitive psychology, or the history of science” (p. 6). On this account, “there is no saying what reference is until we have made some progress at building a science in which a reference‐ like word–world mapping plays a role” (p. 46). Stich's description of the proto‐science account blurs a distinction that is, I think, important in thinking about the theory of reference. According to his description, the task of a theory of reference is to characterize some word–world mapping that is scientifically useful. But if we go with my earlier (p.309) obvious answer, the task is to characterize reference in particular, not just some word–world relation. That task leaves, of course, the task of showing that the word–world relation that is reference is scientifically useful (and, perhaps, showing that no other word–world relation is as scientifically useful). But that is another task. We need to distinguish the task of characterizing reference from the task of showing its scientific utility. And although a lack of progress on the latter task might well make us lose interest in the former task it would not prevent progress on it; it would not prevent us “saying what reference is”. Even scientifically uninteresting properties like being a restaurant and relations like being to the left of have natures open to explanation. This distinction is commonly blurred. Why? The simple and obvious claim that the task of the theory of reference is to characterize reference strikes people as naive because there is thought to be a special problem about identifying reference. Which relation is it the task to characterize? To answer that question we must discover which word–world relations are scientifically interesting. So discovering this becomes part of the task of characterizing reference. I think that this line of thought is misguided because there is no special problem about identifying reference. The alleged problem was made vivid by Putnam's famous “model‐theoretic argument” (1978: 123–7; 1981: 29–48).7 We can pose it as follows: “Why identify reference with one word–world relation rather than with any of the others?” This Page 6 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

On Determining What There Isn't * asks why the relation between ‘Jemima’ and Jemima, ‘cat’ and cats, and . . . (continue on through the paradigms of reference) rather than that between, say, ‘Jemima’ and Fido, ‘cat’ and dogs, and . . . (continue on in any way that appeals) is to be identified as reference. But there is no special problem about reference here. Consider an analogous question about the relation being the father of. Why is the relation between George H. Bush and George W. Bush, Prince Philip and Prince Charles, and . . . (continue on through the paradigms of fathering) rather than the relation between, say, George H. Bush and Bill Clinton, Prince Philip and Baroness Thatcher, and . . . (continue on in any way that appeals) to be identified as being the father of? Or the analogous question about the property cathood. Why is the property of Jemima, Nana, and so on (continue on through the paradigm cats) rather than the property of, say, Fido, George W. Bush, and . . . (continue on in any way that appeals) to be identified as cathood? Insofar as we have any problem here it could be solved by the above‐ mentioned first stage in theorizing about any property F or relation R: we go back to those most expert at identifying instances of being (p.310) a father of and cathood. These experts would, of course, confirm what we surely already know. We need say no more than that the specified relation just is being the father of and the specified property just is cathood. Similarly, if there really were any doubt about which word–world relation was reference we could go back to the experts. They would confirm our view that the specified relation just is reference and we need say no more. At the beginning of theorizing about any F or R we may indeed have an identification problem but, as a matter of fact, we do not now have one with being a father of and cathood. No more do we now have one with reference. And if we did, we could solve it in the same way we solve it for any F or R. There is no problem identifying reference and certainly no special problem. If this is right, we can stand by our obvious answer: the task of the theory of reference is to characterize reference. No more need be said in describing that task. But it would indeed be naive to think that this should remove all metaphysical/methodological worries about reference. We need, at least, to accomplish this task: we need to discover what is common and peculiar to the reference relation. This is the above‐mentioned second stage in theorizing about reference. Accomplishing this is not enough, however. We need to say why reference is theoretically interesting. Not everything that we could characterize is interesting enough to be worth characterizing; for example, being a Virgo, being a witch, and being a restaurant may not be. If reference is to be worth characterizing there has to be something scientifically useful about this particular word–world relation, perhaps even something that makes it more useful than any other word–world relation. The point of my distinction is that showing that reference is scientifically useful is a different task from characterizing reference.

Page 7 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

On Determining What There Isn't * Having distinguished the two tasks we can distinguish two ways to get gloomy about the theory of reference. We could get gloomy because science is not providing a characterization of reference. And we could get gloomy because science does not give reference a significant role. These are clearly very different matters as we can see by considering genes. The success of Mendelian genetics left no basis for gloom about the role of genes in science but in the decades before the rise of molecular genetics there was certainly a basis for concern, even if not gloom, about the characterization of genes. Stich seems to be gloomy about the theory of reference in both ways. However, he does not sharply distinguish the two ways and has very little to say about characterizing reference (pp. 43, 46, 54). Almost all his discussion is about the problems of finding a scientific role for reference. Should we be gloomy about the characterization of reference? Certainly there is no established science that is casting light on this. Almost all the significant (p. 311) work on reference is being done by philosophers. From my naturalistic perspective, that alone is not sufficient reason for gloom. Philosophers have come up with a range of promising ideas to explain reference. Description theories seem plausible for some words; for example, ‘bachelor’ and ‘vixen’. Still, as we noted, description theories pass the referential buck and so in the end we will need other theories that explain the ultimate connections between words and the world. Attempted explanations have appealed to one or more of three causal relations between representations and reality. First, already alluded to in our discussion of Lycan, there is the historical cause of a particular token, a causal chain going back to the dubbing of the token's referent.8 Theorists interested in this have emphasized the “reference borrowing” links in the chain: in acquiring a word or concept from others we borrow their capacity to refer, even if we are ignorant of the referent (Donnellan 1972; Putnam 1975; Burge 1979; Kripke 1980). Second, there is the reliable cause of tokens of that type: a token refers to objects of a certain sort because tokens of that type are reliably correlated with the presence of those objects. The token “carries the information” that a certain situation obtains in much the same way that tree rings carry information about the age of a tree (Dretske 1981; Fodor 1990). Third, there is the teleological cause or function of tokens of that type, where the function is explained along Darwinian lines: the function is what tokens of that type do that explains why they exist, what the type has been “selected for” (Millikan 1984; Papineau 1987; Neander 1995). I would be the last to claim that any of these developments has yielded a thoroughly persuasive characterization of reference.9 Still, much progress has been made. We have reason to hope that the reference of any word will be explained by some such “ultimate” theory, or by a description theory, or by a theory that combines elements of both. Much work remains to be done but there is no cause for gloom.

Page 8 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

On Determining What There Isn't * Whether or not Stich is gloomy about characterizing reference he is certainly gloomy about finding a role for reference in science. In his initial description of the proto‐science account, he mentions the following sciences as possibly giving reference a role: linguistics, cognitive psychology, and the history of science (p. 6). Later he mentions anthropology, history, and sociology (p. 43). His gloom has two bases. First, he is “inclined to be more than a bit skeptical about the claim that any of these areas of inquiry make genuinely explanatory (p.312) use of a reference‐like word–world relation” (p. 44); “the relevant sciences have not yet determined which word–world relation will be of use to them” (p. 7). So, there is no sign that reference is scientifically useful. Second, “there is no a priori reason to suppose that the proto‐scientific project . . . will yield a unique result” (p. 45). So, even if reference were useful in some science, other word–world relations might be useful in other sciences. Is Stich right to be gloomy about finding a role for reference? If we were to set aside linguistics, I would be inclined to think that he was. All these other social sciences rather clearly have an explanatory role for the meaning/content (henceforth, simply “meaning”) of utterances and thoughts, This role is to be found in a widespread practice taken over by these social sciences from the folk: the sciences make ascriptions of the form, ‘X said that p’, ‘X believes that p’, ‘X hopes that p’, and so on, to explain intentional behavior. In these ascriptions, ‘that p’ ascribes a meaning to an utterance or thought.10 Consider a little bit of recent history as an example. Why did most Americans applaud the invasion of Iraq? Part of the explanation is that Bush had led them to believe that Saddam Hussein had weapons of mass destruction, was involved in 9/11, and so on. So part of the explanation is that Americans had certain mental states with certain meanings. Since these sorts of explanations are so common in the social sciences, meanings clearly have a big role in science.11 But, of course, meaning is one thing, reference another. And it does not look as if these social sciences have a direct role for reference. Nor does it look as if they have a direct role for other word–world relations. These relations just do not seem to be the concern of these sciences. But what about linguistics? Broadly construed, linguistics is the science of language. This includes semantics—the theory of meaning—studied in philosophy departments as well as linguistics departments. And we should take semantics to be concerned not only with the meanings of utterances but also with the meanings of thoughts; semantics includes psychosemantics. Now, prima facie, semantics does have a role for reference because the most popular theories of meaning are truth‐referential. Thus, such a theory might claim that the meaning of the sentence ‘Bush is deceitful’ is largely explained by the sentence's being true iff there is something that ‘Bush’ designates and ‘deceitful’ (p.313) applies to. So, the meaning of the sentence is partly explained in terms of two referential relations, designation and application. If such truth‐referential theories are along the right lines, then we have found a clear place for reference Page 9 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

On Determining What There Isn't * in science: reference is part of the scientific explanation of meaning and meaning has a place throughout the social sciences.12 In light of this, we might say that reference has an indirect role throughout those sciences. There are, of course, a range of ways to resist this conclusion. Eliminativists argue that meanings don't really play an explanatory role. Friends of narrow meaning argue that the meanings that play an explanatory role are all narrow and so are to be explained without any mention of reference, perhaps in terms of conceptual role. I have argued elsewhere (1996a) that the case for truth‐ referentialism is in fact strong and that these ways of resisting it fail. But there remains a final and, in my view, more serious threat to truth‐referentialism: deflationism. In giving reference a key role in the explanation of meaning, truth‐referentialists suppose that it is a substantial and explanatorily significant relation. Deflationists like Robert Brandom (1984) and Paul Horwich (1990) reject this supposition. They think, roughly, that reference isn't anything; in particular, it isn't anything that could play the explanatory role required of it in a truth‐ referential semantics.13 If they are right, then we need some other account of meaning (unless we abandon meaning altogether). That is a serious problem that deflationists had not tackled until recent works by Brandom (1994) and Horwich (1998, 2005). Brandom's account of meaning rests on unexplained normativity, as he frankly acknowledges. This feature of his account is, of course, very troubling to the naturalistically inclined. Horwich's account is a “use theory” inspired by Wittgenstein and does not have this feature. Indeed, he thinks that normativity is not intrinsic to meaning. However, in my view (2002d, 2009d), Horwich's theory is open to strong criticism. The arguments from ignorance and error that were devastating for description theories of reference for proper names and natural kind terms can be adapted to provide a strong case against Horwich's use theory. More generally, we need a lot more details of this theory before it can be taken as a serious rival to truth referentialism. If all this is so, why is deflationism such a threat to truth‐referentialism? First, deflationists argue very plausibly that the role of the reference term can be (p. 314) accounted for without taking it to stand for a substantial explanatory relation. Similarly, the truth term need not stand for a substantial explanatory property. Second, although I think we should be hopeful about the theory of reference, it is proving hard to come up with a convincing theory of the ultimate referential links between language and the world. This could be a sign that the deflationists are right and there is no substantial reference relation in need of a theory. It is time to take up Stich's reasons for gloom about the theory of reference. First, he claims that there is no clear sign that reference has a role in science. Well, if I am right, there is a very clear sign that it has a role in semantics, which Page 10 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

On Determining What There Isn't * is part of the science of linguistics widely construed. Of course, the sign might be misleading. I do think that deflationism should make the truth‐referentialist pause. But Stich does not cite deflationism as a reason for his gloom. Second, Stich wonders whether science will favor reference over other word– world relations. There are two issues to consider here. (1) Does science favor reference over other word–world relations in the explanation of meaning? (2) Whether or not science does, do other word–world relations have roles elsewhere in science? The short answer to (1) is that truth‐referentialist explanations of meaning involve reference not any other relation. Any property of a word that was explained in terms of some other word–world relation would not be a meaning, would not play the explanatory role that is definitive of being a meaning. Any explanation of a property of ‘Bush is deceitful’ in terms of a relation that ‘Bush’ has to, say, Fido or to Clinton, and a relation that ‘deceitful’ has to, say, being hungry or being bald would not be an explanation of that sentence's meaning (Devitt 1997b: 116–18). And the friends of reference can be very relaxed about (2). It might be the case that word–world relations other than reference have roles elsewhere in science. I doubt that it is the case but it remains to be seen. But even if it is the case, that does not reflect on the role that reference has in explaining meaning. It is time to sum up this discussion of the theory of reference. The view that this theory is a study of folk semantics is a nonstarter because we have no good reason to suppose that the folk are experts on reference. The proto‐scientific view is much more promising. This view, properly conceived, yields two tasks: first, characterizing reference; second, showing that reference has a scientific role. I have argued that Stich is wrong to be gloomy about the prospects of accomplishing these tasks. The first is certainly proving difficult but there are various ideas on the table that have led to progress and give us hope of more. Accomplishing the second task starts with the ubiquitous role of meanings in the social sciences. Those meanings then have to be explained. The most (p.315) promising explanations are truth‐referential ones. Those explanations give reference a central role in explaining meanings. But there is a dark cloud in the sky: deflationism.

4. Metaphysics Before Semantics Stich's gloom about the theory of reference leads him to the conclusion that it is a mistake to start an ontological investigation with “semantic ascent”: we should abandon the idea of using the theory of reference to address questions about the existence of entities posited by false theories (p. 55). I think that this conclusion is absolutely right, as I have already noted. But we don't need to share his gloom to see this. We simply have to be appropriately modest about the present accomplishments of the theory of reference.

Page 11 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

On Determining What There Isn't * Consider how the argument from semantic ascent goes. It rests on connections of the following sort: (x) (Fx iff ‘F’ applies to x) (x) (a = x iff ‘a’ designates x) The argument from semantic ascent then attempts to show, for example, either that there is nothing that ‘F’ applies to and so there are no Fs; or, that there is something that ‘F’ applies to and so there are Fs. But it is important to notice that the argument could just as well go in the other direction. We could attempt to show either that there are no Fs and so ‘F’ does not apply to anything; or, that there are Fs and so ‘F’ does apply to something. This raises the crucial question: which direction is better? Should we argue from semantics to ontology or vice versa? “The linguistic turn” in philosophy has always favored the direction from semantics to ontology. In my view, appropriate modesty about semantics should lead us in the other direction: we know much more about ontology than about semantics.

Consider two arguments from semantic ascent. (1) It is natural to think that the ancients lived on and had beliefs about the Earth, the very same planet that we live on and have beliefs about. Yet, some philosophers influenced by Thomas Kuhn claim that this is not really so. For, the descriptions that the ancients associated with their names for the heavenly body on which they lived, descriptions like ‘flat’, ‘the center of the universe’, etc., do not, from our theoretical perspective, pick out anything. Applying a description theory of reference, the philosophers may then say one of two things. (i) They may say that the purported referent of those ancient names does not exist and that (p. 316) all the beliefs that the ancients would express using those names were about nothing (and hence not about the Earth). (ii) Or they may say that those descriptions do pick out a planet and the ancient names referred to it; so, the ancients lived on that planet and had beliefs about it. But—this is where the story gets really nasty—that planet existed only relative to the ancient theories not absolutely; the ancients did not live on the same planet as we do. (2) Science tells us that phlogiston does not exist. Yet a philosopher enamored of a causal theory of reference might claim that phlogiston does exist because it is, say, oxygen. For, according to that causal theory, the term ‘phlogiston’ is causally related to the world in such a way that it refers to oxygen.14 Such arguments should give us pause.15 Theories of reference are being used to undermine widely held ontological views and, in the case of (1)(ii), to promote a bizarre metaphysics (criticized in Chs. 5 and 6 of the present volume). Why should we have such confidence in these theories? How can we be confident that that particular description theory is appropriate for the names that the ancients had for the heavenly body on which they took themselves to live and that that particular causal theory is appropriate for ‘phlogiston’. A Moorean response seems in order: the ontological views are much more firmly based than the theories of reference that are thought to undermine them. We have started the argument in the wrong place: rather than using theories of reference against the Page 12 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

On Determining What There Isn't * ontological views we should use the ontological views against the theories of reference. Indeed what support are these theories of reference supposed to have? Not usually the empirical support of the claims of science. Rather, the theories are supposed to be known a priori. Reflecting from the comfort of armchairs, philosophers decide what reference is and from this infer what the world must be like: A priori semantics → a priori metaphysics.16 The Moorean response alone casts doubt on this procedure and the philosophical method it exemplifies, the a priori method of “First Philosophy”. But we can do better: the doubt is confirmed by the sorts of considerations adduced by Quine (1952: pp. xi–xvii; 1961: 42–6). These considerations should lead us to reject a priori knowledge and embrace “naturalism”, the view that there (p.317) is only one way of knowing, the empirical way that is the basis of science (whatever that may be); in Quine's vivid metaphor, the web of belief is seamless.17 From the naturalistic perspective, philosophy becomes continuous with science. And theories of reference have no special status: they are simply some among many empirical hypotheses about the world we live in. As such, they do not compare in evidential support with the posits of successful sciences. Experience has taught us a great deal about the world of trees, cats, atoms, and muons but rather little about how we refer to this world. So semantics is just the wrong place to start the argument.18 Instead, we should start with an empirically based metaphysics and use that as evidence in an empirical study of reference: semantics itself becomes part of science, it becomes “naturalized”: Empirical metaphysics → empirical semantics.19 The semantic ascent arguments not only proceed in the wrong direction, they typically use the wrong methodology.20

When we proceed empirically in the right direction, we should reject any theory of reference that is at odds with scientifically well‐established metaphysical/ ontological views. So the right response to the semantic ascent argument (1) about the Earth is: “So much the worse for that description theory of reference”. It is, of course, possible that we should have sufficient confidence in some theory of reference for a term to allow it to settle a metaphysical issue. However, appropriate modesty about the present achievements of the theory of reference should make this seem a very remote possibility. Quine is fond of a vivid image taken from Otto Neurath. He likens our knowledge —our “web of belief”—to a boat that we continually rebuild whilst staying afloat on it. We can rebuild any part of the boat but in so doing we must take a stand on the rest of the boat for the time being. So we cannot rebuild it all at once. Similarly, we can revise any part of our knowledge but (p.318) in so doing we Page 13 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

On Determining What There Isn't * must accept the rest for the time being. So we cannot revise it all at once. And just as we should start rebuilding the boat by standing on the firmest parts, so also should we start rebuilding our web. Semantics is among the weakest places to stand.

5. How Then do we Do the Metaphysics? If we are not to appeal to theories of reference to settle ontological issues how are we to settle them? As Stich says: “How are we to go about deciding whether or not the entities posited by any false theory exist?” He wonders whether we can find “principles of rational ontological inference” and describes one strategy for finding them (p. 63). This is the “normative‐naturalist” strategy which considers actual cases in the history of science and attempts to abstract the principles (pp. 63–5). Stich is skeptical that this strategy “will uncover principles of rational ontological inference that are rich enough to tell us, in lots of the most interesting cases, what ontological conclusions we ought to draw when we come to believe that some previously accepted theory is seriously mistaken”. He suggests that “in many historical cases the resolution of ontological questions can be explained in part by the personalities of those involved or by social and political factors in the relevant scientific community or in the surrounding society” (p. 70). This leads him, enfant terrible that he likes to be, to wonder whether he is embracing social constructivism. I don't think that he is embracing it and he certainly shouldn't (1991b; 2001a, which is Ch. 5 in the present volume). Still, I think that his skepticism about finding principles that will settle these tricky ontological questions is fairly well‐based. Here, briefly, is my view on these questions, a view very much along the same lines as Stich's. Consider how, in general, we argue directly for the nonexistence of Fs. On the basis of the established view of Fs, we start, implicitly if not explicitly, with an assumption about the nature of being F: something would not be an F unless it were G; being G is part of the very essence of being F. Then we argue that nothing is G. So, there are no Fs. But suppose that someone responds by denying the essentialist assumption upon which the argument rests. “Fs do not have to be G, they are just mistakenly thought to be G. So the argument proves nothing.” Stich has a very nice discussion of witches that illustrates this response: he describes people who believe in witches but claim that “witches make no pact with the devil, cast no evil spells, and do not practice black magic (or ride on broomsticks!)” (pp. 68–9). How do we settle disagreements over an essentialist assumption? (p.319) The difficulty of doing so is apparent when we consider how we determine the nature of being F where Fs clearly do exist. As I pointed out (sec. 2), we identify some uncontroversial examples of Fs and non‐Fs and examine them to see what is common and peculiar to the Fs. Our present difficulty is that we obviously cannot examine any Fs if there aren't any to examine! There were, and are, plenty of people around whom some people call “witches” but we witch deniers obviously cannot discover the nature of witches

Page 14 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

On Determining What There Isn't * by examining these people. For us, there are no uncontroversial examples of witches. So how do we discover what it is to be a witch? We clearly do not have principles that are adequate to settle such issues and I think that Stich is right to be a bit dubious about our finding them. How worried should we be about this? I think we should be a little worried, but not very. And we certainly should not conclude that there is no fact of the matter about ontological issues (let alone sink into constructivism). There are, after all, many generally agreed claims about what exists and what does not. The fact that we do not have epistemological principles adequate to support these claims should not persuade us that they are ill‐based. For, we do not have epistemological principles adequate to support most of what we agree on about the world. This having been said, there are many ontological claims that we do not agree on and I think we may have to accept that some of these disagreements do not concern matters of fact. Furthermore, the intentional realism issue seems particularly troublesome because it involves semantics, a notoriously vague area. It is usual to think that beliefs are essentially things with meanings (or contents). But then, meanings in what sense? Must they be truth‐conditional, for example? These are tricky issues (1996a: 5.1). If folk psychology is really so wrong (which I doubt that it is) then I think that it will be very difficult to settle the intentional realism issue.

6. Conclusion Stich is absolutely right that it was a mistake to tackle ontological issues, including that of intentional realism, by appeal to the theory of reference. However, he is wrong about why it was a mistake. He thinks it was a mistake because, whether we adopt the folk semantic or the proto‐science account of the theory of reference, all we can see is the dark at the end of the tunnel. I have not challenged this gloom over the folk semantic account because, I argue, the account is a nonstarter: we have no good reason to suppose that (p.320) the folk are so expert about reference. I have challenged the gloom over the much more promising proto‐scientific account. This account, properly conceived, yields two tasks: first, characterizing reference; second, showing that reference has a scientific role. The first is certainly proving difficult but there are various ideas on the table that should give us hope. Accomplishing the second task starts with the ubiquitous role of meanings in the social sciences. Those meanings then have to be explained. Despite the threat of deflationism, the most promising explanations are truth‐referential ones. Those explanations give reference a central role in explaining meanings and hence in science. So I have argued that gloom about the theory of reference is not a good reason for resisting semantic ascent. Still, there is a good reason. It arises from being modest about the accomplishments of the theory of reference. Given how little we know about reference, a Moorean response is appropriate to semantic ascent Page 15 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

On Determining What There Isn't * arguments: rather than drawing ontological conclusions from claims about reference, we should be drawing conclusions about reference from claims about ontology. We should “put metaphysics first”. This Moorean response is good on its own but when supported by naturalism it is formidable. From the naturalistic perspective these claims about reference cannot be supported a priori and they do not come close to having the empirical support enjoyed by the ontological claims that they seek to undermine. The semantic ascent arguments use the wrong methodology and proceed in the wrong direction. How then are we to settle ontological issues? I think that Stich's answer is very much along the right lines. We do not have any principles adequate to help us with the difficult cases and it may be that some of these are indeterminate. And intentional realism may be a particularly difficult issue if folk psychology is very wrong. However, we should not lose track of the fact that many ontological issues are not difficult. And, most important of all, we should not lapse into the despair of constructivism, thinking that these issues are really nothing but matters of politics and personalities. Notes:

(*) First published in Dominic Murphy and Michael A. Bishop (eds.), Stich and his Critics (Devitt 2009b). Reprinted with kind permission from Wiley‐Blackwell. (1) All such parenthetical numerals are references to pages in Stich 1996. (2) [2009 addition] See Ch. 4, sec. 4.2, of the present volume for further discussion. (3) “Maxim 3 Settle the realism issue before any epistemic or semantic issue” (1984/1991b: 4). Semantic ascent is one feature of “the linguistic turn” in philosophy. Another is the conflation of metaphysical and semantic issues. Realism and Truth argues against that too: “Maxim 2 Distinguish the metaphysical (ontological) issue of realism from any semantic issue” (1984/ 1991b: 3). [2009 addition] See also Chs. 3, 5, and 7 in the present volume. (4) I argue against the standard linguistic view in Devitt and Sterelny 1989 and in much more detail in my 2006d and e. I argue against the view that a person competent in a language must have propositional knowledge about the meaning and reference of its expressions in my 1981: 95–110; 1996a: 48–86, 1991b: 270– 5. (5) [2009 addition] See Ch. 14 in the present volume for a discussion of intuitions. (6) [2009 addition] This grants the folk too much. I think they are probably expert enough but philosophers and others specializing in language are more expert. Page 16 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

On Determining What There Isn't * (7) My discussion here draws on my 1997b. The alleged problem is prominent in the debate about Putnam's argument; see e.g. LePore and Loewer 1988. (8) I have argued that a term is usually grounded in its referent not only at a dubbing: it is multiply grounded in the referent (1981). This is important if the causal‐historical theory is to handle reference change and many cases of reference confusion. (9) Devitt and Sterelny 1999 has a detailed discussion of the state of play with the theory of reference. (10) These ascriptions of meaning have another very important role (although not obviously a scientific role): they enable us to use the thoughts and utterances of others as guides to reality (1996a: 53–62). (11) The idea here is that a thought has a certain effect in virtue of its meaning. It is worth noting that a thought also has a certain cause in virtue of its meaning. Thus, I might acquire a belief that Bush is deceitful as a result of perceiving Bush's behavior. Part of the explanation of the formation of that particular belief is that it has the meaning it has; a belief that, say, Clinton is chaste would not have been caused by that perceptual experience. Meanings have a role in the social sciences both as causes and effects (1997b: 117). (12) Stich cites a passage in Devitt and Sterelny 1987 as the “most explicit endorsement” of the view that reference has a role in linguistics (86 n.). That endorsement seems rather inadequate to me now (1996a). (13) [2009 addition] For more on the metaphysics of deflationism, see Ch. 8 in the present volume. (14) Enc mentions, although he does not endorse, such a line of argument (1976: 267). (15) [2009 addition] The discussion that follows in this section is similar to that of underdetermination and extreme skepticism in Ch. 3, secs. 4–6, and of constructivism in Ch. 5, sec. 6, of the present volume. (16) This procedure is, of course, analogous to that of traditional epistemology: inferring what the world must be like from a conclusion about the nature of knowledge. (17) A particularly important consideration against the a priori, in my view (1998, which is Ch. 12 in the present volume; 2005a; 2009c, which is Ch. 13), is the lack of anything close to a satisfactory explanation of a nonempirical way of knowing. We are told what this way of knowing is not—it is not the empirical way of deriving knowledge from experience—but we are not told what it is.

Page 17 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

On Determining What There Isn't * (18) “Linguistics . . . is a tiny twig on the tree of science. Yet if [scientists cannot draw ontological conclusions until linguists tell them what reference is], then this twig on the tree of science . . . gets to play a fundamental role in determining” what exists. Stich is surely dead right to find this a “hopelessly implausible scenario” (54). (19) Of course, one might wonder how an empirical semantics should proceed. I think that it is very difficult to say. My attempt is 1996a: ch. 2. (20) There is, of course, a truth underlying semantic ascent: to determine whether the posits of a theory exist we have to know what those posits are and for that we have to understand the language of the theory (1991b: 50–3). But understanding a language is a practical skill that does not require theoretical knowledge about the language, else we would understand very little; see n. 4 of this chapter.

Access brought to you by:

Page 18 of 18

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References

Putting Metaphysics First: Essays on Metaphysics and Epistemology Michael Devitt

Print publication date: 2009 Print ISBN-13: 9780199280803 Published to Oxford Scholarship Online: February 2010 DOI: 10.1093/acprof:oso/9780199280803.001.0001

(p.321) References Bibliography references: Alston, W. 1958. “Ontological Commitments”. Philosophical Studies, 9: 8–17. Antony, L. 2004. “A Naturalized Approach to the A Priori”. In E. Sosa and E Villanueva (eds.), Epistemology: Philosophical Issues 14, 2004 (Cambridge, Mass.: Blackwell Publishers), 1–17. Appiah, Antony. 1991. “Representations and Realism”. Philosophical Studies, 61: 65–74. Armstrong, D. M. 1978a. Nominalism and Realism: Universals and Scientific Realism, i (New York: Cambridge University Press). —— 1978b. A Theory of Universals: Universals and Scientific Realism, ii (New York: Cambridge University Press). —— 1980. “Against ‘Ostrich Nominalism’: A Reply to Michael Devitt”. Pacific Philosophical Quarterly, 61: 440–9. Repr. in Mellor and Oliver 1997a: 101–11. —— 2004. Truth and Truthmakers (Cambridge: Cambridge University Press). Aune, Bruce. 1987. “Conceptual Relativism”. In Tomberlin 1987: 269–85. Auxier, Randall E., and Lewis Edwin Hahn, eds. 2007. The Philosophy of Michael Dummett (Chicago: Open Court Publishing Co.). Ayer, Alfred Jules. 1952. Language, Truth and Logic (New York: Dover Publications, Inc.).

Page 1 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References Azzouni, Jody. 2007. “Ontological Commitment in the Vernacular”. Noûs, 41: 204–26. Baker, C. L. 1995. English Syntax (2nd edn. Cambridge, Mass.: MIT Press; 1st edn. 1989). Båve, Arvid. 2009. “Deflationism and the Primary Truth Bearer”. Synthese, forthcoming. Bealer, G. 1992. “The Incoherence of Empiricism”. The Aristotelian Society, suppl. 66: 99–138. —— 1998. “Intuition and the Autonomy of Philosophy”. In DePaul and Ramsey 1998: 201–39. —— 1999. “A Theory of the A Priori”. Philosophical Perspectives, 13. Epistemology, 1999, 29–55. Berkeley, George. 1710. Principles of Human Knowledge. London. Biello, David. 2007. “Life's Bar Code: Genetic Tests Unveil 15 New Species of North American Birds”. Science and Technology at Scientific American.com. 22 Feb. Blackburn, Simon. 1980. “Truth, Realism, and the Regulation of Theory”. In French et al. 1980: 353–71. —— 1984. Spreading the Word: Groundings in the Philosophy of Language (Oxford: Clarendon Press). —— 1993a. Essays in Quasi‐Realism (New York: Oxford University Press). —— 1993b. “Realism, Quasi, or Queasy?” In Haldane and Wright 1993a: 365–83. —— 1999. “Introduction to the Realism Debates”. In Steven D. Hales (ed.), Metaphysics: Contemporary Readings (Belmont, Calif.: Wadsworth Publishing Co.), 47–51. (p.322) Blackburn, Thomas. 1988. In French et al. 1988: 179–94. Boghossian, Paul. 1990a. “The Status of Content”. Philosophical Review, 99: 157–84. —— 1990b. “The Status of Content Revisited”. Pacific Philsophical Quarterly, 71: 264–78. —— 2000. “Knowledge of Logic”. In Boghossian and Peacocke 2000: 229–54.

Page 2 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References —— 2001. “How are Objective Epistemic Reasons Possible?” Philosophical Studies, 106: 1–40. —— and Christopher Peacocke. 2000. New Essays on the A Priori (Oxford: Clarendon Press). —— and David J. Velleman. 1989. “Colour as a Secondary Quality”. Mind, 98: 81– 103. BonJour, L. 1998. In Defense of Pure Reason: A Rationalist Account of A Priori Justification (Cambridge: Cambridge University Press). —— 2001a. Precis of In Defense of Pure Reason. Philosophy and Phenomenological Research, 63: 625–31. —— 2001b. “Replies”. Philosophy and Phenomenological Research, 63: 673–98. —— 2005a. “In Defense of A Priori Reasons”. In Sosa and Steup 2005: 98–105. —— 2005b. “Reply to Devitt”. In Sosa and Steup 2005: 115–18. —— 2005c. “Last Rejoinder”. In Sosa and Steup 2005: 120–2. Boyd, Richard N. 1973. “Realism, Underdetermination and a Causal Theory of Evidence”. Noûs, 7: 1–12. —— 1984. “The Current Status of Scientific Realism”. In Leplin 1984a: 41–82. —— 1985. “Lex Orandi est Lex Credendi”. In Churchland and Hooker 1985: 3– 34. —— 1988. “How to be a Moral Realist”. In Sayre‐McCord 1988a: 181–228. —— 1999. “Homeostasis, Species, and Higher Taxa”. In Wilson 1999a: 141–85. Brandom, Robert. 1984. “Reference Explained Away”. Journal of Philosophy, 81: 469–92. —— 1988. “Pragmatism, Phenomenalism, and Truth Talk”. In French et al. 1988: 75–93. —— 1994. Making it Explicit: Reasoning, Representing, and Discursive Commitment (Cambridge, Mass.: Harvard University Press). —— 1997. “From Truth to Semantics: A Path through Making it Explicit”. In Villanueva 1997: 141–54. Brown, Curtis. 1988. “Internal Realism: Transcendental Idealism?” In French et al. 1988: 145–55. Page 3 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References Brown, James Robert. 1994. Smoke and Mirrors: How Science Reflects Reality (New York: Routledge). Burge, Tyler. 1979. “Individualism and the Mental”. In P. A. French, T. E. Uehling Jr., and H. K. Wettstein (eds), Midwest Studies in Philosophy, 4 Studies in Metaphysics, 73–121. —— 1998. “Computer Proof, Apriori Knowledge, and Other Minds”. In J. Tomberlin (ed.), Philosophical Perspectives, 12. Language, Mind, and Ontology, 1998 (Cambridge, Mass.: Blackwell Publishers), 1–38. (p.323) Busch, Jacob. 2006a. “Does the Issue of Response‐Dependence have Any Consequences for Realism?” Croatian Journal of Philosophy, 6: 27–39. —— 2006b. “Entity Realism Meets the Pessimistic Meta‐Induction: The World is Not Enough”. Sats—Nordic Journal of Philosophy, 7: 106–26. Caplan, Arthur L. 1980. “Have Species Become Declassé?” PSA 1: 71–82. Carlson, Richard A. 2003. “Skill Learning”. In Nadel 2003: iv. 36–42. Carnap, Rudolf. 1963. “Intellectual Autobiography”. In The Philosophy of Rudolf Carnap, ed. P. A. Schilpp (La Salle, Ill.: Open Court), 1–84. Carroll, Sean B. 2005. Endless Forms Most Beautiful: The New Science of Evo Devo and the Making of the Animal Kingdom (New York: W. W. Norton). Cartright, Nancy. 1983. How the Laws of Physics Lie (Oxford: Clarendon Press). Chomsky, Noam. 1980. “Rules and Representations”. Behavioral and Brain Sciences, 3: 1–14. —— 1986. Knowledge of Language: Its Nature, Origin, and Use (New York: Praeger Publishers). —— 1988. Language and Problems of Knowledge: The Managua Lectures (Cambridge, Mass.: MIT Press). —— 1995. “Language and Nature”. Mind, 104: 1–61. Churchland, Paul M. 1979. Scientific Realism and the Plasticity of Mind (Cambridge: Cambridge University Press). —— 1985. “The Ontological Status of Observables: In Praise of the Superempirical Virtues”. In Churchland and Hooker 1985: 35–47. —— and Clifford A. Hooker, eds. 1985. Images of Science: Essays on Realism and Empiricism, with a Reply from Bas C. van Fraassen (Chicago: University of Chicago Press). Page 4 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References Claridge, M. F., H. A. Dawah, and M. R. Wilson, eds. 1997. The Units of Biodiversity (London: Chapman & Hall). Clendinnen, F. J. 1989. “Realism and the Underdetermination of Theory”. Synthese, 81: 63–90. Cracraft, Joel. 1983. “Species Concepts and Speciation Analysis”, in R. Johnston (ed.), Current Ornithology (New York: Plenum Press), 159–87. Repr. in Ereshefsky 1992a: 93–120. (Citations are to Ereshefsky.) Crain, Stephen, and Rosalind Thornton. 1998. Investigations in Universal Grammar: A Guide to Experiments on the Acquisition of Syntax and Semantics (Cambridge, Mass.: MIT Press). Cushing, James T., and Ernan McMullin, eds. 1989. Philosophical Consequences of Quantum Theory (Notre Dame, Ind.: University of Notre Dame Press). —— A. Fine, and S. Goldstein, eds. 1996. Bohmian Mechanics and Quantum Theory: An Appraisal, Boston Studies in the Philosophy of Science, 184 (Dordrecht: Kluwer). Darwin, Charles. 1859. The Origin of Species by Means of Natural Selection (1st edn. 1859; New York: Barnes & Noble Classics, 2004). David, Marian. 1994. Correspondence and Disquotation: An Essay on the Nature of Truth (Oxford: Oxford University Press). (p.324) DePaul, M. R., and W. Ramsey, eds. 1998. Rethinking Intuition: The Psychology of Intuition and its Role in Philosophical Inquiry (London: Rowman & Littlefield Publishers). De Queiroz, Kevin. 1999. “The General Lineage Concept of Species and the Defining Properties of the Species Category”. In Wilson 1999a: 49–89. Devitt, Michael. 1979. “Against Incommensurability”. Australasian Journal of Philosophy, 57: 29–50. —— 1980. “ ‘Ostrich Nominalism’ or ‘Mirage Realism’?” Pacific Philosophical Quarterly, 61: 433–39. Repr. in Mellor and Oliver 1997a: 93–100. (Chapter 1 in the present volume.) —— 1981. Designation (New York: Columbia University Press). —— 1983a. “Dummett's Anti‐Realism”. Journal of Philosophy, 80: 73–99. —— 1983b. “Realism and the Renegade Putnam” (a critical study of Putnam 1978). Noûs, 17: 291–301.

Page 5 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References —— 1983c. “Realism and Semantics” (Part II of a critical study of French et al. 1980). Noûs, 17: 669–81. —— 1984. Realism and Truth (Oxford: Basil Blackwell). —— 1987. “Does Realism Explain Success?” Revue Internationale de Philosophie: New Trends in Realism: An Australian Perspective, 29–44. —— 1988a. “Realism Without Truth: A Response to Bertolet”. Analysis, 48: 198– 203. —— 1988b. “Rorty's Mirrorless World”. In French et al. 1988: 157–77. —— 1990. “Transcendentalism about Content”. Pacific Philosophical Quarterly, 71: 247–63. —— 1991a “Aberrations of the Realism Debate”. Philosophical Studies, 61: 43– 63. Repr. in German in H. J. Sandkuhler and D Patzold (eds.), Dialektic: Die Wirklichkeit der Wissenschaft: Probleme des Realismus (Hamburg: Meiner, 1991), 113–37. (Chapter 2 in the present volume.) —— 1991b. Realism and Truth (2nd edn. Oxford: Basil Blackwell). —— 1991c. “Minimalist Truth: A Critical Notice of Paul Horwich's Truth”. Mind and Language, 6: 273–83. —— 1991d. “Realism Without Representation: A Response to Appiah”. Philosophical Studies, 61: 75–7. —— 1993. “A Critique of the Case for Semantic Holism”. In Fodor and Lepore 1993: 17–60. An earlier version, without the “Postscript”, appeared in Philosophical Perspectives, 7. Language and Logic, ed. James E. Tomberlin (Atascadero, Calif.: Ridgeview Publishing Co.), 281–306. —— 1994. “The Methodology of Naturalistic Semantics”. Journal of Philosophy, 91: 545–72. —— 1996a. Coming to our Senses: A Naturalistic Defense of Semantic Localism (New York: Cambridge University Press). —— 1996b. “The Metaphysics of Nonfactualism”. In Philosophical Perspectives, 10. Metaphysics, 1996, ed. James E. Tomberlin (Cambridge, Mass.: Blackwell Publishers), 159–76. (Chapter 7 in the present volume.) (p.325) —— 1997a. “Afterword”. In a repr. of Devitt 1991b (Princeton: Princeton University Press), 302–45. Page 6 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References —— 1997b. “On Determining Reference”. In Alex Burri (ed.), Sprache und Denken/ Language and Thought (New York: Walter de Gruyter), 112–21. —— 1997c. “Responses to the Maribor Papers”. In Dunja Jutronic (ed.), The Maribor Papers in Naturalized Semantics (Maribor: Pedagoska fakulteta Maribor), 353–411. —— 1998. “Naturalism and the A Priori”. Philosophical Studies, 92: 45–65. (Chapter 12 in the present volume.) —— 1999a. “A Naturalistic Defense of Realism”. In Steven D. Hales (ed.), Metaphysics: Contemporary Readings (Belmont, Calif.: Wadsworth Publishing Co.), 90–103. —— 1999b. “Postscript to ‘A Naturalistic Defense of Realism’ ”. In Steven D. Hales (ed.), Metaphysics: Contemporary Readings (Belmont, Calif.: Wadsworth Publishing Co.), 104–5. —— 2001a. “Incommensurability and the Priority of Metaphysics”. In P. Hoyningen‐Huene and H. Sankey (eds.), Incommensurability and Related Matters (Dordrecht: Kluwer Academic Publishers), 143–57. (Chapter 5 in the present volume.) —— 2001b. “The Metaphysics of Truth”. In Michael Lynch (ed.), The Nature of Truth (Cambridge, Mass.: MIT Press), 579–611. (Chapter 8 in the present volume.) —— 2002a. “Underdetermination and Realism”. In Ernest Sosa and Enrique Villanueva (eds.), Realism and Relativism: Philosophical Issues 12, 2002 (Cambridge, Mass.: Blackwell Publishers), 26–50. —— 2002b. “The Metaphysics of Deflationary Truth”. In Richard Schantz (ed.), What is Truth? (Berlin and New York: Walter de Gruyter), 60–78. —— 2002c. “Moral Realism: A Naturalistic Perspective”. Croatian Journal of Philosophy, 4: 1–15. Repr. in Spanish in Arete, 16 (2004), 185–206. (Chapter 9 in the present volume.) —— 2002d. “Meaning and Use”. Philosophy and Phenomenological Research, 65: 106–21. —— 2003. “Linguistics is Not Psychology”. In Alex Barber (ed.), Epistemology of Language (Oxford: Oxford University Press), 107–39.

Page 7 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References —— 2004. “Pourquoi il est si difficile de faire un monde: Contre la reponse‐ dependence globale”. Tr. into French by Olivier Massin. In Jean‐Maurice Monnoyer (ed.), La Structure du monde: Objets, propriétés, états de choses. Renouveau de la metaphysique dans L’École Australienne (Paris: Vrin), 421–44. English version, “Worldmaking Made Hard: Rejecting Global Response Dependency”. Croatian Journal of Philosophy, 6 (2006): 3–25. —— 2005a. “There is No A Priori”. In Sosa and Steup 2005: 105–15. —— 2005b. “Reply to BonJour”. In Sosa and Steup 2005: 118–20. —— 2005c. “Scientific Realism”. In Frank Jackson and Michael Smith (eds.), The Oxford Handbook of Contemporary Philosophy (Oxford: Oxford University Press), 767–91. Repr. in Greenough and Lynch 2006: 100–24. (Chapter 4 in the present volume.) —— 2005d. “Rigid Application”, Philosophical Studies, 125: 139–65. —— 2006a. “Responses to the Rijeka Papers”. Croatian Journal of Philosophy, 6: 97–112. (p.326) Devitt, Michael. 2006b. “The Pessimistic Meta‐Induction: A Response to Jacob Busch”. Sats—Nordic Journal of Philosophy, 7: 127–35. —— 2006c. “Intuitions”. In Victor Gomez Pin, Jose Ignacio Galparaso, and Gotzon Arrizabalaga (eds.), Ontology Studies Cuadernos de Ontologia: Proceedings of VI International Ontology Congress (San Sebastian, 2004) (San Sebastian: Universidad del Pais Vasco), 169–76. (Chapter 14 in the present volume.) —— 2006d. Ignorance of Language (Oxford: Oxford University Press). —— 2006e. “Intuitions in Linguistics”. British Journal for the Philosophy of Science, 57: 481–513. —— 2008a. “Explanation and Reality in Linguistics”. Croatian Journal of Philosophy, 8: 203–31. —— 2008b. “Resurrecting Biological Essentialism”. Philosophy of Science, 75: 344–82. (Chapter 11 in the present volume.) —— 2009a. “Biological Realisms”. In Heather Dyke (ed.), From Truth to Reality: New Essays in Logic and Metaphysics (London: Routledge & Kegan Paul), 43–65. —— 2009b. “On Determining What There Isn't”. In Dominic Murphy and Michael A. Bishop (eds.), Stich and his Critics (Oxford: Wiley‐Blackwell), 46–61. (Chapter 15 in the present volume.) Page 8 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References —— 2009c. “No Place for the A Priori”. In Michael J. Shaffer and Michael Veber (eds.), New Perspectives on A Priori Knowledge and Naturalism (La Salle, Ill.: Open Court, forthcoming). (Chapter 13 in the present volume.) —— 2009d. “Meaning: Truth‐Referential or Use”. In Richard Schantz (ed.), Prospects for Meaning (Berlin: Walter de Gruyter, forthcoming). —— 2009e. “Natural Kinds and Biological Realisms”. In Joseph Campbell, Michael O'Rourke, and Matthew Slater (eds.), Carving Nature at its Joints: Topics in Contemporary Philosophy, viii (Cambridge, Mass.: MIT Press, forthcoming). (Chapter 10 in the present volume.) —— and Georges Rey. 1991. “Transcending Transcendentalism: A Response to Boghossian”. Pacific Philosophical Quarterly, 72: 87–100. —— and Kim Sterelny. 1987. Language and Reality: An Introduction to the Philosophy of Language (Cambridge, Mass.: MIT Press). —— and —— 1989. “Linguistics: What's Wrong with ‘the Right View’ ”. In Philosophical Perspectives, 3. Philosophy of Mind and Action Theory, ed. James E. Tomberlin (Atascadero, Calif.: Ridgeview Publishing Co.), 497–531. —— and —— 1999. Language and Reality: An Introduction to the Philosophy of Language (2nd edn. Cambridge, Mass.: MIT Press). Donnellan, Keith S. 1972. “Proper Names and Identifying Descriptions”. In Donald Davidson and Gilbert Harman (eds.), Semantics of Natural Language (Dordrecht: Reidel), 356–79. Doris, John M., and Alexandra Plakias. 2008. “How to Argue about Disagreement: Evaluative Diversity and Moral Realism”. In Sinnott‐Armstrong 2008b: 303–31. Dretske, Fred I. 1981. Knowledge and the Flow of Information (Cambridge, Mass.: MIT Press). (p.327) Duhem, P. 1906. The Aim and Structure of Physical Theory, trans. P. Wiener (Princeton: Princeton University Press, 1954). Dummett, Michael. 1973. Frege: Philosophy of Language (London: Duckworth). —— 1977. Elements of Intuitionism (Oxford: Oxford University Press). —— 1978. Truth and Other Enigmas (Cambridge, Mass.: Harvard University Press). —— 1991. Logical Basis of Metaphysics (London: Duckworth).

Page 9 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References —— 1993. The Seas of Language (Oxford: Clarendon Press). —— 2007. “Reply to Putnam”. In Auxier and Hahn 2007: 168–84. Dupré, John. 1981. “Natural Kinds and Biological Taxa”. Philosophical Review, 90: 66–90. —— 1993. The Disorder of Things (Cambridge, Mass.: Harvard University Press). —— 1999. “On the Impossibility of a Monistic Account of Species”. In Wilson 1999a: 3–22. Dwyer, Susan, and Paul Pietroski. 1996. “Believing in Language”. Philosophy of Science, 63: 338–73. Ellis, Brian. 1979. Rational Belief Systems (Oxford: Basil Blackwell). —— 1985. “What Science Aims to Do”. In Churchland and Hooker 1985: 48–74. Eldredge, N., and J. Cracraft. 1980. Phylogenetic Patterns and the Evolutionary Process (New York: Columbia University Press). Eldredge, Niles, and Stephen J. Gould. 1972. “Punctuated Equilibria: An Alternative to Phyletic Gradualism”. In T. J. M Schopf (ed.), Models in Paleobiology (San Francisco: Freeman, Cooper, & Co.), 82–115. Enc, B. 1976. “Reference of Theoretical Terms”. Noûs, 10: 261–82. Ereshefsky, Marc. 1991. “Species, Higher Taxa, and the Units of Evolution”. Philosophy of Science, 58: 84–101. Repr. in Ereshefsky 1992: 381–98. (Citations are to Ereshefsky.) —— (ed.) 1992a. The Units of Evolution: Essays on the Nature of Species (Cambridge, Mass.: MIT Press). —— 1992b. “Introduction”. In Ereshefsky 1992a: pp. xiii–xvii. —— 1992c. “Introduction to Part II: Philosophical Issues”. In Ereshefsky 1992a: 187–98. —— 1998. “Species Pluralism and Anti‐Realism”. Philosophy of Science, 65: 103– 20. —— 1999. “Species and the Linnaean Hierarchy”. In Wilson 1999a: 285–305. —— 2001. The Poverty of the Linnaean Heirarchy: A Philosophical Study of Biological Taxonomy (Cambridge: Cambridge University Press).

Page 10 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References Fales, Evan. 1988. “How to be a Metaphysical Realist”. In French et al. 1988: 253–74. Feyerabend, Paul K. 1975. Against Method (London: New Left Books). —— 1978. Science in a Free Society (London: New Left Books). —— 1981a. Realism, Rationalism and Scientific Method: Philosophical Papers, i (Cambridge: Cambridge University Press). —— 1981b. Problems of Empiricism: Philosophical Papers, ii (Cambridge: Cambridge University Press). Field, Hartry. 1972. “Tarski's Theory of Truth”. Journal of Philosophy, 69: 347– 75. (p.328) Field, Hartry. 1973. “Theory Change and the Indeterminacy of Reference”. Journal of Philosophy, 70: 462–81. —— 1978. “Mental Representation”. Erkenntnis, 13: 9–61. —— 1986. “The Deflationary Conception of Truth”. In Graham MacDonald and Crispin Wright (eds.), Fact, Science, and Morality: Essays on A. J. Ayer's Language, Truth, and Logic (Oxford: Basil Blackwell), 55–117. —— 1994. “Disquotational Truth and Factually Defective Discourse”. Philosophical Review, 103: 405–52. —— 1996. “The A Prioricity of Logic”. Proceedings of the Aristotelian Society, 96: 1–21. —— 1998. “Epistemological Nonfactualism and the A Prioricity of Logic”. Philosophical Studies, 92: 1–21. Fiengo, Robert. 2003. “Linguistic Intuitions”. Philosophical Forum, 34: 253–65. Fine, Arthur. 1986a. The Shaky Game: Einstein, Realism, and the Quantum Theory (Chicago: University of Chicago Press). —— 1986b. “Unnatural Attitudes: Realist and Instrumentalist Attachments to Science”. Mind, 95: 149–77. —— 1991. “Piecemeal Realism”. Philosophical Studies, 61: 79–96. Fodor, Jerry. 1990. A Theory of Content and Other Essays (Cambridge, Mass.: MIT Press). —— and Ernest Lepore, eds. 1993. Holism: A Consumer Update (Grazer Philosophische Studien, 46; Amsterdam: Rodopi, B.V.). Page 11 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References French, Peter A., Theodore E. Uehling Jr., and Howard K. Wettstein, eds. 1980. Midwest Studies in Philosophy, v. Studies in Epistemology (Minneapolis: University of Minnesota Press). —— 1988. Midwest Studies in Philosophy, xii. Realism and Antirealism (Minneapolis: University of Minnesota Press). Gauker, Christopher. 2006. “Scientific Realism as an Issue in Semantics”. In Greenhaugh and Lynch 2006: 125–36. Geach, Peter, 1960. “Ascriptivism”. Philosophical Review, 69: 221–5. Gendler, Tamar Szabo. “Thought Experiments”. In Nadel 2003: iv. 388–94. Ghiselin, Michael T. 1974. “A Radical Solution to the Species Problem”. Systematic Zoology, 47: 350–83. Repr. in Ereshefsky 1992a: 279–91. (Citations are to Ereshefsky.) —— 1987. “Species Concepts, Individuality, and Objectivity”, Biology and Philosophy, 2: 127–143. Repr. in Ereshefsky 1992a: 363–80. (Citations are to Ereshefsky.) Gibbard, Alan, 1990. Wise Choices, Apt Feelings: A Theory of Normative Judgment (Cambridge Mass.: Harvard University Press). Gladwell, M. 2005. Blink: The Power of Thinking Without Thinking (New York: Little, Brown). Glymour, Clark. 1984. “Explanation and Realism”. In Leplin 1984: 173–92. Godfrey‐Smith, Peter. 2008. “Recurrent Transient Underdetermination and the Glass Half Full”. Philosophical Studies, 137: 141–8. Goldberg, Elkhonon. 2005. The Wisdom Paradox: How Your Mind Can Grow Stronger as Your Brain Grows Older (New York: Gotham Books). (p.329) Goldman, A. (1999). “A Priori Warrant and Naturalistic Epistemology”. In J. Tomberlin (ed.), Philosophical Perspectives, xiii. Epistemology, 1999 (Cambridge, Mass.: Blackwell Publishers), 1–28. Goodman, Nelson. 1978. Ways of Worldmaking (Indianapolis: Hackett Publishing Co.). Gopnik, Alison, and Eric Schwitzgebel. 1998. “Whose Concepts are They, Anyway? The Role of Philosophical Intuitions in Empirical Psychology”. In DePaul and Ramsey 1998: 75–91.

Page 12 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References Gordon, P.C., and R. Hendrick. “Intuitive Knowledge of Linguistic Co‐reference”. Cognition, 62: 325–70. Greenough, Patrick, and Michael Lynch, eds. 2006. Truth and Realism (Oxford: Oxford University Press). Griffiths, Graham C. D. 1974. “On the Foundations of Biological Systematics”, Acta Biotheoretica, 23: 85–131. Griffiths, Paul. 1999. “Squaring the Circle: Natural Kinds with Historical Essences”, in Wilson 1999a: 209–28. —— 2002. “What is Innateness?” Monist, 85: 70–85. Grover, Dorothy, L., Joseph L. Camp, Jr., and Nuel D. Belnap, Jr. 1975. “A Prosentential Theory of Truth”. Philosophical Studies, 27: 73–125. Gutting, Gary. 1985. “Scientific Realism versus Constructive Empiricism: A Dialogue”. In Churchland and Hooker 1985: 118–31. Haack, Susan. 1987. “ ‘Realism’ ”. Synthese, 73: 275–99. Hacking, Ian. 1983. Representing and Intervening: Introductory Topics in the Philosophy of Natural Science (Cambridge: Cambridge University Press). Haegeman, Liliane. 1994. Introduction to Government and Binding Theory (2nd edn. Oxford: Blackwell Publishers; 1st edn. 1991). Haidt, Jonathan, and Frederik Bjorklund. 2008. “Social Intuitionists Answer Six Questions about Moral Psychology”. In Sinnott‐Armstrong 2008b: 181–217. Haldane, John, and Crispin Wright, eds., 1993a. Reality, Representation, and Projection (New York: Oxford University Press). —— 1993b. “Introduction”. In Haldane and Wright 1993a: 3–12. Hale, Bob. 1993. “Can there be a Logic of Attitudes?” In Haldane and Wright 1993a: 337–63. Harman, Gilbert. 1977. The Nature of Morality (New York: Oxford University Press). —— 1986. “Moral Explanations of Natural Facts—Can Moral Claims be Tested Against Moral Reality?” Southern Journal of Philosophy, 24: 57–68. —— 1988. “Ethics and Observation”. In Sayre‐McCord 1988a: 119–24 (repr. of The Nature of Morality (New York: Oxford University Press, 1977), ch. 1). Page 13 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References —— 1999. Reasoning, Meaning, and Mind (Oxford: Clarendon Press). —— and Judith Jarvis Thomson. 1996. Moral Relativism and Moral Objectivity (Cambridge, Mass.: Blackwell Publishers). Hauser, Marc D., Liane Young, and Fiery Cushman. 2008. “Reviving Rawl's Linguistic Analogy: Operative Principles and the Causal Structure of Moral Actions”. In Sinnott‐Armstrong 2008b: 107–43. (p.330) Hawkes, Terence. 1977. Structuralism and Semiotics (London: Methuen & Co.). Heil, John. 1989. “Recent Work in Realism and Anti‐Realism”. Philosophical Books, 30: 65–73. Heller, Mark. 1988. “Putnam, Reference, and Realism”. In French et al. 1988: 113–27. Hennig, Willi. 1966. Phylogenetic Systematics (Urbana, Ill.: University of Illinois Press). Hesse, Mary. 1967. “Laws and Theories”. In The Encylopedia of Philosophy, ed. Paul Edwards (New York: Macmillan), iv. 404–10. Hoefer, C., and A. Rosenberg. 1994. “Empirical Equivalence, Underdetermination, and Systems of the World”. Philosophy of Science, 61: 592– 607. Hooker, Clifford A. 1974. “Systematic Realism”. Synthese, 51: 409–97. Horgan, Terence, and Mark Timmons. 1990/1. “New Wave Moral Realism Meets Moral Twin Earth”. Journal of Philosophical Research, 16: 447–65. —— 1992a. “Troubles on Moral Twin Earth: Moral Queerness Revived”. Synthese, 92: 221–60. —— 1992b “Troubles for New Wave Moral Semantics: The ‘Open Question Argument’ Revived”. Philosophical Papers, 21: 153–75. Horwich, Paul. 1990. Truth (2nd edn. 1998; Oxford: Clarendon Press). —— 1997. “Deflationary Theory and the Problem of Aboutness”. In Villanueva 1997: 95–106. —— 1998. Meaning (Oxford: Clarendon Press). —— 2005. Reflections on Meaning (Oxford: Clarendon Press).

Page 14 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References Hoyningen‐Huene, P. 1993. Reconstructing Scientific Revolutions: Thomas S. Kuhn's Philosophy of Science, trans. A. T. Levine from German edn. 1989 (Chicago: University of Chicago Press). —— Eric Oberheim and Hanne Andersen. 1996. “On Incommensurability”. Studies in the History and Philosophy of Science, 27: 131–41. Hull, David L. 1965. “The Effects of Essentialism on Taxonomy: Two Thousand Years of Stasis”. British Journal for the Philosophy of Science, 15: 314–26; 16: 1– 18. Repr. in Ereshefsky 1992a: 199–225. (Citations are to Ereshefsky.) —— 1978. “A Matter of Individuality”, Philosophy of Science, 45: 335–60. Repr. in Ereshefsky 1992a: 293–316. (Citations are to Ereshefsky.) —— 1997. “The Ideal Species Concept—And Why We Can't Get It”. In Claridge et al. 1997: 357–80. —— 1999. “On the Plurality of Species: Questioning the Party Line”. In Wilson 1999a: 23–48. Jackson, Frank. 1998. From Metaphysics to Ethics (Oxford: Clarendon Press). —— and Michael Smith, eds. 2005. The Oxford Handbook of Contemporary Philosophy (Oxford: Oxford University Press). Jameson, Frederic. 1972. The Prison‐House of Language (Princeton: Princeton University Press). Jennings, Richard. 1989. “Scientific Quasi‐Realism”. Mind, 98: 223–45. Johnston, Mark. 1989. “Dispositional Theories of Value”. Proceedings of the Aristotelian Society, suppl. 63: 139–74. (p.331) —— 1993. “Objectivity Refigured: Pragmatism without Verificationism”. In Haldane and Wright 1993: 85–130. —— 1998. “Are Manifest Qualities Response‐Dependent?” The Monist, 81: 3–43. Kant, Immanuel. 1783. Prolegomena to Any Future Metaphysics. Keil, Frank C. 1989. Concepts, Kinds, and Cognitive Development (Cambridge, Mass.: MIT Press). Khlentzos, Drew. 2004. Naturalistic Realism and the Anti‐Realist Challenge (Cambridge, Mass.: MIT Press). Kirkham, Richard. 1992. Theories of Truth: A Critical Introduction (Cambridge MA: MIT Press).

Page 15 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References Kitcher, Philip. 1984. “Species”, Philosophy of Science, 51: 308–33. Repr. in Kitcher 2003: 113–34. Repr. in Ereshefsky 1992a: 317–41. (Citations are to Kitcher 2003.) —— 1989. “Some Puzzles about Species”. In Kitcher 2003: 135–58. —— 1993. The Advancement of Science: Science without Legend, Objectivity without Illusions (New York: Oxford University Press). —— 2003. In Mendel's Mirror: Philosophical Reflections on Biology (New York: Oxford University Press). Kitts, David B., and David J. Kitts. 1979. “Biological Species as Natural Kinds”. Philosophy of Science, 46: 613–22. Kornblith, Hilary. 1993. Inductive Inference and its Inductive Ground: An Essay in Naturalistic Epistemology (Cambridge, Mass.: MIT Press). —— 1998. “The Role of Intuition in Philosophical Inquiry: An Account with No Unnatural Ingredients”. In DePaul and Ramsey 1998: 129–41. —— 2000. “The Impurity of Reason”. Pacific Philosophical Quarterly, 81: 67–89. Kripke, Saul A. 1980. Naming and Necessity (Cambridge, Mass.: Harvard University Press). Kuhn, Thomas S. 1962. The Structure of Scientific Revolutions (Chicago: Chicago University Press; 2nd edn. 1970). —— 1979. “Metaphor in Science”. In A. Ortony (ed.), Metaphor and Thought (Cambridge: Cambridge University Press), 409–19. —— 1983. “Commensurability, Comparability, Communicability”. In Proceedings of the 1982 Biennial Meeting of the Philosophy of Science Association, ed. P. D. Asquith and T. Nickles (East Lancing: Philosophy of Science Association), 669– 88. Kukla, Andre. 1993. “Laudan, Leplin, Empirical Equivalence, and Underdetermination”. Analysis, 53: 1–7. —— 1998. Studies in Scientific Realism (New York: Oxford University Press). Lance, Mark. 1997. “The Significance of Anaphoric Theories of Truth and Reference”. In Villanueva 1997: 181–98. Lange, Marc. 1995. “Are there Natural Laws Concerning Particular Biological Species?” Journal of Philosophy, 92: 430–51.

Page 16 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References —— 2002. “Baseball, Pessimistic Inductions, and the Turnover Fallacy”. Analysis, 62: 281–5. LaPorte, Joseph. 2004. Natural Kinds and Conceptual Change (Cambridge: Cambridge University Press). (p.332) Larson, Richard, and Gabriel Segal. 1995. Knowledge of Meaning: An Introduction to Semantic Theory (Cambridge, Mass.: MIT Press). Latour, Bruno, and Steve Woolgar. 1986. Laboratory Life: The Construction of Scientific Facts (2nd edn. Princeton: Princeton University Press; 1st edn. 1979). Laudan, Larry. 1981. “A Confutation of Convergent Realism”. Philosophy of Science, 48: 19–49. Repr. in Leplin 1984a. —— 1984. Science and Values (Berkeley, Calif.: University of California Press). —— 1996. Beyond Positivism and Relativism: Theory, Method and Evidence (Boulder, Colo.: Westview Press). —— and Jarrett Leplin. 1991. “Empirical Equivalence and Underdetermination”. Journal of Philosophy, 88: 449–72. Repr. in Laudan 1996. (Page references are to the latter.) Leeds, Stephen. 1978. “Theories of Reference and Truth”. Erkenntnis, 13: 111– 29. Leplin, Jarrett, ed. 1984a. Scientific Realism (Berkeley, Calif.: University of California Press). —— 1984b. “Introduction”. In Leplin 1984a: 1–7. —— 1987. “Surrealism”. Mind, 96: 519–24. —— 1997. A Novel Defense of Scientific Realism (New York: Oxford University Press). —— and Larry Laudan. 1993. “Determination Underdeterred: Reply to Kukla”. Analysis, 53: 8–15. Lepore, Ernest, and Barry Loewer. 1988. “A Putnam's Progress”. In French et al. 1988: 459–73. Levin, Michael. 1984. “What Kind of Explanation is Truth?” In Leplin 1984a: 124–39. Levin, Joseph. 1993. “Intentional Chemistry”. In Fodor and Lepore 1993: 103–34.

Page 17 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References —— 1998. “Troubles with Moderate Localism”. In Dunja Jutronic (ed.), The Maribor Papers in Naturalized Semantics (Maribor: Pedagoska fakulteta Maribor), 59–72. Lewis, David. 1983. “New Work for a Theory of Universals”. Australasian Journal of Philosophy, 61: 343–77. Repr. in Mellor and Oliver 1997a: 188–227. —— 1984. “Putnam's Paradox”. Australasian Journal of Philosophy, 62: 221–36. Lewis, Peter J. 2002. “Why the Pessimistic Induction is a Fallacy”. Synthese, 129: 371–80. Lipton, Peter. 1991. Inference to the Best Explanation (London: Routledge). Luntley, Michael. 1988. Language, Logic and Experience: The Case for Anti‐ Realism (La Salle, Ill.: Open Court). Lycan, William G. 1988. Judgement and Justification (Cambridge: Cambridge University Press). Lynch, Michael P. 1998. Truth in Context: An Essay on Pluralism and Objectivity (Cambridge, Mass.: MIT Press). McAllister, J. W. 1993. “Scientific Realism and Criteria for Theory‐Choice”. Erkenntnis, 38: 203–22. Mackie, John, 1977. Ethics: Inventing Right and Wrong (New York: Penguin). (p.333)

—— 1988. “The Subjectivity of Values”. In Sayre‐McCord 1988a: 95–118. (Repr. of Ethics: Inventing Right and Wrong (New York: Penguin, 1977), ch. 1.) McMichael, Alan. 1988. “Creative Ontology and Absolute Truth”. In French et al. 1988: 51–74. McMullin, Ernan. 1984. “A Case for Scientific Realism”. In Leplin 1984a: 8–40. —— 1991. “Comment: Selective Anti‐Realism”. Philosophical Studies, 61: 97– 108. —— Mallet, James. 1995. “A Species Definition for the Modern Synthesis”. Trends in Ecology and Evolution, 10: 294–9. Margolis Joseph. 1986. Pragmatism without Foundations: Reconciling Realism and Relativism (Oxford: Basil Blackwell). Martin, Michael. 1971. “Referential Variance and Scientific Objectivity”. British Journal for the Philosophy of Science, 22: 17–26. Page 18 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References Mason, H. E. 1988. “Realistic Interpretations of Moral Questions”. In French et al. 1988: 413–32. Matheson, Carl. 1989. “Is the Naturalist Really Naturally a Realist?” Mind, 98: 247–58. Matthen, Mohan. 1998. “Biological Universals and the Nature of Fear”. Journal of Philosophy, 95: 105–32. Maxwell, Grover. 1962. “The Ontological Status of Theoretical Entities”. In Minnesota Studies in the Philosophy of Science, iii. Scientific Explanation, Space and Time, ed. H. Feigl and G. Maxwell (Minnepolis: University of Minnesota), 3– 27. Mayr, Ernst. 1961. “Cause and Effect in Biology”. Science, 134: 1501–6. —— 1963. “Species Concepts and their Application”, from Populations, Species, and Evolution (Cambridge, Mass.: Harvard University Press), ch. 2. (Repr. in Ereshefsky 1992a: 15–25. Citations are to Ereshefsky.) —— 1969. Principles of Systematic Zoology (New York: McGraw Hill). —— 1982. The Growth of Biological Thought (Cambridge, Mass.: Harvard University Press). Mellor, D. H., and Alex Oliver, eds. 1997a. Properties (Oxford: Oxford University Press). —— 1997b. “Introduction”. In Mellor and Oliver 1997a: 1–33. Miller, Alexander. 2003. “The Significance of Semantic Realism”. Synthese, 136: 191–217. Miller, Richard W. 1987. Fact and Method: Explantion, Confirmation and Reality in the Natural and Social Sciences (Princeton: Princeton University Press). Millikan, Ruth. 1984. Language, Thought, and Other Biological Categories: New Foundations for Realism (Cambridge, Mass.: MIT Press). —— 2000. On Clear and Confused Ideas (Cambridge: Cambridge University Press). Mishler, Brent D. 1999. “Getting Rid of Species?” In Wilson 1999a: 307–15. —— and Michael J. Donoghue. 1982. “Species Concepts: A Case for Pluralism”, Systematic Zoology, 31: 491–503. Repr. in Ereshefsky 1992a: 121–37. Musgrave, Alan. 1985. “Realism versus Constructive Empiricism”. In Churchland and Hooker 1985: 197–221. Page 19 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References —— 1988. “The Ultimate Argument for Scientific Realism”. In R. Nola (ed.), Relativism and Realism in Science (Dordrecht: Kluwer Academic Publishers), 229–52. (p.334) Nadel, Lynn, ed. 2003. Encyclopedia of Cognitive Science (London: Nature Publishing Group). Nagel, Thomas, 1980. “The Limits of Objectivity”. In Sterling M. McMurrin (ed.), The Tanner Lectures on Human Values (Salt Lake City: University of Utah Press), 77–139. —— 1986. The View from Nowhere (Oxford: Oxford University Press). Neander, Karen. 1995. “Misrepresenting and Malfunctioning”. Philosophical Studies, 79: 109–41. Oberheim, Eric, and Paul Hoyningen‐Huene. “Incommensurability, Realism and Meta‐Incommensurability”. Theoria, 12: 447–65. O'Hara, Robert J. 1993. “Systematic Generalization, Historical Fate, and the Species Problem”. Systematic Biology, 42: 231–46. Okasha, Samir. 2002. “Darwinian Metaphysics: Species and the Question of Essentialism”. Synthese, 131: 191–213. Papineau, David. 1979. Theory and Meaning (Oxford: Clarendon Press). —— 1987. Reality and Representation (Oxford: Basil Blackwell). —— 1993. Philosophical Naturalism (Oxford: Blackwell Publishers). Patterson, Douglas. 2002. “Theories of Truth and Convention T”. Philosophers' Imprint. http://www.philosophersimprint.org/ Peacocke, Christopher. 2005. “The A Priori”. In F. Jackson and M. Smith (eds.), The Oxford Handbook of Contemporary Philosophy (Oxford: Oxford University Press), 739–63. Pettit, Philip. 1990. “The Reality of Rule‐Following”. Mind, 1–21. —— 1991. “Realism and Response‐Dependence”. Mind, 587–626. —— 1993. The Common Mind: An Essay on Psychology, Society and Politics (New York: Oxford University Press). —— 1998a. “Terms, Thing and Response‐Dependence”. European Review of Philosophy, 3: 55–66. ——. 1998b. “Noumenalism and Response‐Dependence”. Monist, 81: 112–32. Page 20 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References Plantinga, Alvin. 1993. Warrant and Proper Function (New York: Oxford University Press). Popper, Karl R. 1950. The Open Society and its Enemies (Princeton: Princeton University Press). —— 1968. Logic of Scientific Discovery (New York: Basic Books). Posner, M. I. 1973. Cognition: An Introduction (Glenville, Ill.: Scott, Foresman). Price, Huw. 1998. “Two Paths to Pragmatism II”. European Review of Philosophy, 3: 109–47. Psillos, Stathis. 1999. Scientific Realism: How Science Tracks Truth (New York: Routledge). Putnam, Hilary. 1975. Mind, Language and Reality: Philosophical Papers (Cambridge: Cambridge University Press), ii. —— 1978. Meaning and the Moral Sciences (London: Routledge & Kegan Paul). —— 1979. “Reflections on Goodman's Ways of World‐Making”. Journal of Philosophy, 76: 603–18. (p.335) —— 1981. Reason, Truth and History (Cambridge: Cambridge University Press). —— 1983. Realism and Reason: Philosophical Papers (Cambridge: Cambridge University Press), iii. —— 1985. “A Comparison of Something with Something Else”. New Literary History, 17: 61–79. —— 1987. The Many Faces of Realism (LaSalle, Ill.: Open Court). —— 2007. “Between Scylla and Charybdis: Does Dummett Have a Way Out?” In Auxier and Hahn 2007: 155–67. Pylyshyn, Zenon W. 1984. Computation and Cognition: Toward a Foundation for Cognitive Science (Cambridge, Mass.: MIT Press). Quine, W. V. 1952. Methods of Logic (London: Routledge & Kegan Paul). —— 1960. Word and Object (Cambridge, Mass.: MIT Press). —— 1961. From a Logical Point of View (2nd edn. Cambridge, Mass.: Harvard University Press; 1st edn. 1953).

Page 21 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References —— 1966. “The Scope and Language of Science”. In The Ways of Paradox and Other Essays (New York: Random House), 215–32. —— 1969. “Epistemology Naturalized”. In Ontological Relativity and Other Essays (New York: Random House), 69–90. —— 1970. Philosophy of Logic (Cambridge, Mass.: Harvard University Press). —— 1975a. “On Empirically Equivalent Systems of the World”. Erkenntnis, 9: 313–28. —— 1975b. “The Nature of Natural Knowledge”. In S. Guttenplan (ed.), Mind and Language (Oxford: Clarendon Press), 67–81. —— 1980. “Soft Impeachment Disowned”. Pacific Philosophical Quarterly, 61: 450–1. —— 1981. Theories and Things (Cambridge, Mass.: Harvard University Press). Radford, Andrew. 1988. Transformational Grammar: A First Course (Cambridge: Cambridge University Press). Railton, Peter. 1986. “Moral Realism”. Philosophical Review, 95: 163–207. —— 1993. “What the Non‐Cognitivist Helps Us to See the Naturalist Must Help Us to Explain”. In Haldane and Wright 1993: 279–300. —— 1996. “Moral Realism: Prospects and Problems”. In Walter Sinnott‐ Armstrong and Mark Timmons (eds.), Moral Knowledge: New Readings in Moral Epistemology (Oxford: Oxford University Press), 49–81. Rey, Georges. 1993. “The Unavailability of What We Mean: A Reply to Quine, Fodor and Lepore”. In Fodor and Lepore 1993: 61–101. —— 1997. “Devitt's Naturalism: A Priori Resistance to the A Priori?” In Dunja Jutronic (ed.), The Maribor Papers in Naturalized Semantics (Maribor: Pedagoska fakulteta Maribor), 141–54. —— 1998. “A Naturalistic A Priori”. Philosophical Studies, 92: 25–43. —— 2001. “Digging Deeper for the A Priori”. Philosophy and Phenomenological Research, 63: 649–56. Richard, Mark. 1997. “Deflating Truth”. In Villanueva 1997: 57–78. (p.336) Rorty, Richard. 1976. “Realism and Reference”. Monist, 59: 321–40. —— 1979. Philosophy and the Mirror of Nature (Princeton: Princeton University Press). Page 22 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References —— 1982. Consequences of Pragmaticism (Essays: 1972–1980) (Minneapolis: University of Minnesota Press). —— 2006. “Truth and Realism: Remarks at St. Andrews”. In Greenhaugh and Lynch 2006: 239–47. Rosen, Gideon. 1994. “Objectivity and Modern Idealism: What is the Question?” In Michaelis Michael and John O'Leary‐Hawthorne (eds.), Philosophy in Mind: The Place of Philosophy in the Study of Mind (Dordrecht: Kluwer Academic Publishers), 277–319. Rosenberg, Alex. 1985. The Structure of Biological Science (New York: Cambridge University Press). Ruse, Michael. 1987. “Biological Species: Natural Kinds, Individuals, or What?” British Journal for the Philosophy of Science, 38: 225–42. Repr. in Ereshefsky 1992a: 343–61. (Citations are to Ereshefsky.) Sankey, Howard. 1994. The Incommensurability Thesis (Brookfield: Avebury). —— 1997. “Incommensurability: The Current State of Play”. Theoria, 12: 425–45. —— 1998. “Taxonomic Incommensurabilty”. International Studies in the Philosophy of Science, 12: 7–16. Sayre‐McCord, Geoffrey, ed. 1988a. Essays on Moral Realism (Ithaca, NY: Cornell University Press). —— 1988b. “Preface”. In Sayre‐McCord 1988a: pp. ix–xii. —— 1988c. “Introduction: The Many Moral Realisms”. In Sayre‐McCord 1988a: 1–23. —— 1991. “Being Realist about Relativism (in Ethics)”. Philosophical Studies, 61: 155–76. Schaffer, Jonathan. 2008. “Truthmaker Commitments”. Philosophical Studies, 141: 7–19. Simpson, George G. 1945. “The Principles of Classification and a Classification of Mammals”. Bulletin of the American Museum of Natural History, 85: 1–350. —— 1961. Principles of Animal Taxonomy (New York: Columbia University Press). Sinnott‐Armstrong, Walter, ed. 2008a. Moral Psychology, i. The Evolution of Morality: Adaptations and Innateness (Cambridge, Mass.: MIT Press).

Page 23 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References —— ed. 2008b. Moral Psychology, ii. The Cognitive Science of Morality (Cambridge, Mass.: MIT Press). —— 2008c. “Framing Moral Intuitions”. In Sinnott‐Armstrong 2008b: 47–76. Slote, Michael A. 1971. “The Rationality of Aesthetic Value Judgments”. Journal of Philosophy, 68: 821–39. Sklar, Lawrence. 1981: “Do Unborn Hypotheses have Rights?” Pacific Philosophical Quarterly, 62: 17–29. Smart, J. J. C. 1963: Philosophy and Scientific Realism (London: Routledge & Kegan Paul). Soames, Scott. 1984. “What is a Theory of Truth?” Journal of Philosophy, 81: 411–29. —— 1997. “The Truth about Deflationism”. In Villanueva 1997: 1–44. (p.337) Sober, Elliott. 1980. “Evolution, Population Thinking and Essentialism”. Philosophy of Science, 47: 350–83. Repr. in Ereshefsky 1992a: 247–78. (Citations are to Ereshefsky.) —— 1993. Philosophy of Biology (Boulder, Colo.: Westview Press). Sokal, Robert, and Theodore J. Crovello. 1970. “The Biological Species Concept: A Critical Evaluation”. American Naturalist, 104: 127–53. Repr. in Ereshefsky 1992a: 27–55. (Citations are to Ereshefsky.) Sosa, Ernest. 1998. “Minimal Intuition”. In DePaul and Ramsey 1998: 257–69. —— and Mattias Steup, eds. 2005. Contemporary Debates in Epistemology (Cambridge, Mass.: Blackwell Publishers). Spencer, N. J. 1973. “Differences between Linguists and Nonlinguists in Intuitions of Grammaticality‐Acceptability”. Journal of Psycholinguistic Research, 2: 83–98. Stanford, P. Kyle. 1995. “For Pluralism and Against Realism about Species”. Philosophy of Science, 62: 70–91. Sterelny, Kim. 1999. “Species as Ecological Mosaics”. In Wilson 1999a: 119–38 —— and Paul Griffiths. 1999. Sex and Death (Chicago: University of Chicago Press). Stich, Stephen P. 1983. From Folk Psychology to Cognitive Science: The Case Against Belief (Cambridge, Mass.: MIT Press).

Page 24 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References —— 1996. Deconstructing the Mind (Oxford: Oxford University Press). Stove, D. C. 1991. The Plato Cult and Other Philosophical Follies (Oxford: Basil Blackwell). Sturgeon, Nicholas L. 1988. “Moral Explanations”. In Sayre‐McCord 1988a: 229– 55. Sun, Ron, Edward Merrill, and Todd Peterson. 2001. “From Implicit Skills to Explicit Knowledge: A Bottom–Up Model of Skill Learning”. Cognitive Science, 25: 203–44. Tarski, Alfred 1949. “The Semantic Conception of Truth and the Foundations of Semantics”. In H. Feigl and W. Sellars (eds.), Readings in Philosophical Analysis (New York: Appleton‐Century‐Crofts), 52–84. —— 1956. Logic, Semantics, Metamathematics: Papers from 1923–1938 (Oxford: Clarendon Press). Taylor, Barry. 1987. “The Truth in Realism”. Revue Internationale de Philosophie: New Trends in Realism: An Australian Perspective, 45–63. Tomberlin, James T., ed. 1987. Philosophical Perspectives, i. Metaphysics, 1987 (Atascadero, Calif.: Ridgeview Publishing Co.). Van Fraassen, Bas C. 1980. The Scientific Image (Oxford: Clarendon Press). —— 1985. “Empiricism in the Philosophy of Science”. In Churchland and Hooker 1985: 245–308. —— 1989. Laws and Symmetry (Oxford: Clarendon Press). —— 2000. “The False Hopes of Traditional Epistemology”. Philosophy and Phenomenological Research, 40: 253–80. Van Valen, Leigh. 1976. “Ecological Species, Multi‐Species and Oaks”, Taxon, 25: 233–9. Repr. in Ereshefsky 1992a: 69–77. (Citations are to Ereshefsky.) Villanueva, Enrique, ed. 1997. Truth: Philosophical Issues 8, 1997 (Atascadero, Calif.: Ridgeview Publishing Co.). (p.338) Wade, Nicholas. 2006. “Scientists Hope to Unravel Neanderthal DNA and Human Mysteries”. New York Times (21 July 2006): A14. Warenski, Lisa. 2009. “Naturalism, Fallibilism, and the A Priori”. Philosophical Studies, 142: 403–26.

Page 25 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

References Webster, Gerry, and Brian C. Goodwin. 1996. Form and Transformation: Generative and Relational Principles in Biology (Cambridge: Cambridge University Press). Wedgwood, Ralph. 1998. “The Essence of Response‐Dependence”. European Review of Philosophy, 3: 31–54. Whorf, Benjamin Lee. 1956. Language, Thought, and Reality, ed. and intro. John B. Carroll (Cambridge, Mass.: MIT Press). Wiggins, David. 1980. Sameness and Substance (Oxford: Blackwell). Williams, D. C. 1966. Principles of Empirical Realism: Philosophical Essays, with editorial assistance by Harry Ruja (Springfield, Ill.: Charles C. Thomas). Williams, Michael. 1993. “Realism and Scepticism”. In Haldane and Wright 1993a: 193–214. Wilson, Robert A., ed. 1999a. Species: New Interdisciplinary Essays (Cambridge, Mass.: MIT Press). —— 1999b. “Realism, Essence, and Kind: Resuscitating Species Essentialism?” In Wilson 1999a: 187–207. Winsor, Mary P. 2006. “The Creation of the Essentialism Story: An Exercise in Metahistory”. History and Philosophy of the Life Sciences, 28: 149–74. Wisniewski, E. J. 1998. “The Psychology of Intuitions”. In DePaul and Ramsey 1998: 45–58. Wolterstorff, Nicholas. “Are Concept‐Users World‐Makers?” In Tomberlin 1987: 233–67. Worrall, J. 1989. “Structural Realism: The Best of Both Worlds”. Dialectica, 43: 99–124. Wright, Crispin. 1988. “Realism, Antirealism, Irrealism, Quasi‐Realism”. In French et al. 1988: 25–49. —— 1993. “Realism: The Contemporary Debate—W(h)ither Now?” In Haldane and Wright 1993: 63–84. —— 2001. “On Basic Logical Knowledge: Reflections on Paul Boghossian's ‘How Are Objective Epistemic Reasons Possible?’ ” Philosophical Studies, 106: 41–85.

Access brought to you by: Page 26 of 26

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Index

Putting Metaphysics First: Essays on Metaphysics and Epistemology Michael Devitt

Print publication date: 2009 Print ISBN-13: 9780199280803 Published to Oxford Scholarship Online: February 2010 DOI: 10.1093/acprof:oso/9780199280803.001.0001

(p.339) Index abduction (inference to the best explanation) 4, 75–8, 91, 93–4, 279–80 aims of science 3, 36, 46–7, 72 n. 10 Alston, W. 24 anagenesis 237–8, 243, 248 analyticity 9, 254 n. 3, 284–7, 289 Andersen, Hanne 104, 114 Antony, Louise 273 n. 3 Appiah, Anthony 3, 35 n. 12, 45 n. 31, 48, 50–1 a priori knowledge 1, 3, 5, 8–9, 45, 54, 63–6, 77, 85, 109–12, 129, 130, 136, 180, 184, 188–9, 253–91, 292, 299, 302, 307, 316–17, 320 Aristotle 213–14, 216, 240–1, 243 n. 49, 248–9 armchair philosophy 1, 9, 63, 109–10, 276–7, 292, 299, 316 Armstrong, David 2, 13–30, 139, 154 n. 37, 164, 203 n. 12 artifacts and tools 42, 49–50, 101–2 n. 3, 107 n. 15, 125, 229, 231–2 Aune, Bruce 42 n. 24 Ayer A. J. 137 nn. 1–4 and 6, 138 n. 8, 152, 161 n. 11, 163 n. 14, 186 nn 7–10 and 12–13 Azzouni, Jody 21 n. 9 Baker, C. L. 293 n. 2 Barker, Matthew 249 n. Båve, Arvid 176 n. Bealer, George 9, 275–8, 281–2, 283 n. 14, 284 n. 15, 288–9, 290 n. 20, 292 n. 1 Berkeley, George 94 n. 56 Biello, David 219 n. 14, 240 n. 44 Bigelow, John 275 n. 6 biological essentialism 7–8, 213–49 biological explanation: historical 206–9, 212, 220–3, 239; structural 7, 206–10, 212, 220–3, 231–9, 243, 247 biological laws 222, 246–7, 249 biological realisms 7, 197–212 biological variation and change 239–47, 248–9 Bjorklund, Frederik 192 n. 18 Page 1 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Index Blackburn, Simon 3, 41 nn. 25, 27, 47 n. 39, 48, 51, 54–6, 137–41, 145 n. 22, 149 n. 29, 161, 163 n. 14, 164–5, 186 Blackburn, Thomas 40 n. 22 Boghossian, Paul 109 n. 19, 129 n., 136 n. 20, 137 nn. 1 and 4–6, 138 n. 8, 140 n. 14, 154 n., 161 nn. 9 and 11, 163 n. 14, 186 nn. 7, 10–13, 281, 290 n. 19 BonJour, Laurence 9, 277–8, 281–2, 283 n. 14, 289–90, 292 n. 1 Boyd, Richard N. 32 n. 3, 47 n. 39, 71 n. 7, 73, 76–8, 84, 183 nn. 1, 2, 191, 194, 195–6, 201, 203 n. 10, 205 n. 13, 209 n. 17, 214 n. 4, 216 n. 10, 219, 244 n. 54, 246 n. 57 Brandom, Robert 34 n. 9, 159–60, 163 n. 14, 167, 172 n. 25, 174–5, 179–81, 313 Brown, Curtis 41 n. 24 Brown, J. R. 32 n. 3, 72 n. 9, 183 n. 2 Bub, Jeffrey 91 n. 51 Burge, Tyler 288, 283, 311 Busch, Jacob 90 n. 50, 121 Busch, Lisa 154 n. 37 Campbell, Keith 154 n. Caplan, Arthur L. 241 Carlson, Richard A. 297 n. 8 Carnap, Rudolf 172 n. 26 Carroll, Lewis 298 Carroll, Sean B. 240 n. 45 Cartesian 9, 43, 46, 48, 63, 68, 95, 108, 142 n. 17, 254 n. 3, 258, 263 n. 14, 285–9, 293, 297, 300, 307 Cartright, Nancy 246 category problem 7–8, 199, 217, 225–8, 232–5, 237, 239, 248 Chomsky, Noam 9, 191 n., 293, 299 n. 10 Churchland, Paul 38 n. 16, 76 n. 20 circular arguments: premise‐ 280–1; rule‐ 8, 77, 279–83 Clendinnen, F. J. 76 n. 20 Commonsense Realism (about observables) 3, 33, 57–66, 67–9, 104 competence, conceptual, and linguistic 9, 45, 142 n. 17, 177 n. 34, 254 n. 3, 284–7, 293–4, 296–301, 307, 311 n. 20 (p.340) conspecificity problem 7–8, 219, 230, 232–5, 237, 239, 242, 245, 248 Constructivism 4–5, 10, 62 n. 7, 68–70, 101–14, 117–20, 121–2, 126, 144 n., 316 n. 15, 318–20, see also realism convergence in science 7, 47, 178 Cordero, Alberto 249 n. Cowie, Fiona 39 n. 19, 48 n. 43 Cracraft, Joel 7, 197, 208, 224 n. 26 Crain, Stephen 300 n. 13 Crovello, Theodore J. 198, 224 Cushing, James T., 69 n. 2 Cushman, Fiery 191 n. Darwell, Stephen 189 Darwin, Charles 7–8, 75, 198 n., 213–14, 217, 218–19, 222, 223 n. 24, 224 n. 26, 227, 239–42, 248–9, 291, 311 David, Marian 175–6, 181 n. 42 Davidson, Donald 41 n. 27 Page 2 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Index De Queiroz, Kevin 199 n. 6 Descartes, Rene 58–9, 63, 262; see also Cartesian Dickenson, Michael 235 n., 249 n. Donnellan, Keith S. 311 Donoghue, Michael J. 228 Doris, John M. 192–3 n. 19 Dretske, Fred 111 n. 23, 134, 311 Dudau, Radu 91 n. 51 Duhem, P. 79 n. 25, 274; Quine 72, 82, 113, 138 n. 12, 184, 194, 282, 292, 316–17 Dummett, Michael v, 3, 31, 33, 39, 41, 48, 51–4, 138 n. 12, 145 n. 22, 155, 180, 281 Dupré, John 199 nn. 5 and 6, 202, 205, 214 n. 1, 218, 224 n. 26, 232 n. 36 Dwyer, Susan 293 n. 2 Eldredge, N. 7, 197, 208, 243 eliminativism (antirealism) 137, 189, 305–6; epistemological 47; about meaning 6, 35, 51, 181, 313; about the mind 9, 88 n. 43, 303–5; about truth 155–6, 160, see also realism Ellis, Brian 32 n. 3, 33 n. 6, 34 n. 9, 43 n. 30, 47 n. 38, 71 n. 7, 82 n. 32, 183 n. 2 emotivism 137, 140, 148, 150 n., 152, 161, 162, 186, 187 empirical equivalence 4, 67, 79–86, 91–3, 95 Enc, B. 316 n. 14 epistemology: a priori, 3, 5, 54, 64–5, 110, 273 n. 3,; naturalized (empirical) 4, 46, 54, 64–5, 77, 90, 97, 110, 113, 255, 271 n. 1, 273 n. 3, 317–8; priority of metaphysics over 2, 3–5, 44–6, 48, 62–6, 109–12, see also naturalism, epistemological; skepticism equivalence schema and thesis 34, 37, 43–4, 71 n. 8, 150–1, 159–60, 168–9, 173–5, 177– 8 Ereshefsky, Marc 7, 197, 198, 199 n. 6, 200, 202–7, 208, 209 n. 18, 210–12, 215 n. 5, 219, 223, 228, 241 n. 47, 242, 244 n. 53, 249 n. 60 error theories 140, 180, 189 essentialism 214, 318–19; biological 7–8, 213–49; individual 40 n. 23, 214 n. 2, 216 n. 11 explanatory virtues 85–6 facts: see universals Fales, Evan 32 n. 3, 40 n. 22, 41 n. 29, 71 n. 7, 183 n. 2 fallibilism 255–6, 262, 273–4 n. 3, 283–4 n. 14, 293, 306–8 Feyerabend, Paul 4, 62 n. 6, 68, 99–114, 115, 257 Field, Hartry 8, 14 n. 9, 38 n. 16, 77 n. 22, 100, 110 n. 21, 148 n. 27, 154 n. 37, 171–2, 178 n., 181 n. 42, 184 n. 4, 253, 256, 262 n., 263–70, 277, 279 n. 9, 280 n. 11 Fiengo, Robert 301 n. 15 Fig‐Leaf Realism 32, 107 Fine, Arthur 3, 32 n. 4, 36, 39, 41 n. 25, 43 n. 29, 46 n. 35, 47 nn. 40 and 41, 48, 51, 55– 6, 69 n. 2, 72 n. 9, 75–6, 78 n. 23 First Philosophy 61–6, 110, 254–5, 316 Fodor, Jerry 311 Page 3 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Index folk theories 276, 295–6, 307; biological 213, 217, 221 n. 18; epistemological 33; moral 188; metaphysical 32–3, 50, 58, 62, 68; psychological 295, 303–5, 319; semantic and linguistic 9, 179, 292–4, 300–1, 306–8, 312, 314, 319–20 functionalism 131 Gauker, Christopher 50 n. 45, 69–70 n. 4, 71 n. 8, 73 n. 13 Geach, Peter 145 n. 22 Gendler, Tamar Szabo 296 n. Ghiselin, Michael T. 201 n. 9, 216, 218, 224 n. 27 Gibbard, Alan 148 n. 27 Gladwell, M. 295 n. Glymour, Clark 76 n. 19 (p.341) Godfrey‐Smith, Peter 91 n. 51, 96, 222 n. 22, 249 n. 60 Goldberg, Elkhonon 294 n. 5 God’s Eye View 39 n. 19, 45–6, 120 Goldstein S. 69 n. 2 Goodman, Nelson 31, 33, 41 n. 24, 62 n. 6, 102, 105 n. 10, 117–18 Goodwin, Brian C. 215 n. 8 Gordon, P.C., 301 n. 14 Gopnik, Alison 295 Gould, Stephen J. 243 Griffiths, G. C. D. 230 Griffiths, Paul 7, 197, 198–9, 204, 209–10, 213, 215–16, 218–21, 222 n. 19, 223 n. 23, 224, 226–8, 232, 238, 239, 240–1, 243 n. 51, 245, 246 n. 57, 249 n. 60 Grover, Dorothy L. 34 n. 9, 159–60, 163 n. 14, 166, 167 n. Gutting, Gary 76 n. 20 Haack, Susan 32 n. 2 Hacking, Ian 38 n. 15, 70 n. 5, 78 n. 23, 92 Haegeman, Liliane, 301 n. 15 Haidt, Jonathan 192 n. 18 Haldane, John 137 nn. 2–6, 138–9 n. 12, 161 n. 11, 186 nn. 8–13 Hale, Bob 137 nn. 2, 4 and 6, 138 n. 9, 145 n. 22, 161 n. 11, 163 n. 14, 186 nn. 8, 10, 12 and 14 Hales, Steven 109 n. 19 Harman, Gilbert 7, 149 n. 28, 192, 193–4, 278 n. Hauser, Mark 191 n. Hawkes, Terence 107 n. 16 Heil, John 1 Heller, Mark 40 n. 22 Hennig, Willi 238 Hendrick, R. 301 n. 14 Herbert, Paul 219 n. 14, 240 n. 44 Hesse, Mary 32 n. 3, 60 n. 5, 71 n. 7, 183 n. 2 higher taxa and categories 7, 197, 198, 203 n. 11, 207–12, 213–15, 225 Hoefer, C. 80 holism: confirmation 8, 253, 257–8, 267–8, 274–5, 282, 284, 288; Page 4 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Index meaning 179–80, 253 homeostatic cluster kinds 195, 203 n. 10, 219 Hooker, Clifford A. 32 n. 3, 71 n. 7, 183 n. 2 Horgan, Terence 185 n. 5 Horwich, Paul 71, 151 n. 32, 159–60, 164, 166, 167 n. 17, 173–4, 177 n. 35, 181 nn. 41 and 42, 313 Hoyningen‐Huene, Paul 4, 5, 62 n. 6, 99–117, 120 Hull, David 197 n. 3, 200 n. 7, 201 n. 9, 216, 219, 222 nn. 20 and 21, 230–1, 241, 242, 245 Hume, David 105 n. 10, 192 idealism 32, 36, 47–8, 58, 61–3, 106, 118–19, 121, see also realism ideas and sense‐data 32, 46, 58–9, 61–2, 69, 94 n. 56 ignorance and error arguments 304, 313 incommensurability 4–5, 99–120; meta‐ 114–17, 120 indeterminacy (vagueness) 10, 89–90, 192, 265, 308, 319–20; in biology 8, 217, 242–6, 248–9 innate beliefs, knowledge, and epistemic systems 8, 191 n., 254, 263, 264–6, 272–3, 294 instrumentalism 1, 5, 72 n. 11, 147–8, 154, 161, see also realism intentional realism 303, 305, 319–20 internal realism (Putnam’s) 33, 40, 102 interpretation and realism 3, 6, 33, 41, 52–3, 55, 69 n. 2, 72 n. 11, 140–4, 148 nn. 25 and 26, 151 n. 33, 153, 162, 165, 167, 187–8 intuitions 1–2, 9, 275–8, 281–2, 288, 292–302; linguistic 9, 293–4, 299–302, 306–7; about meanings 109 n. 20, 293, 300–1; metaphysical 88–9; moral 194–5; referential 9, 292–4, 300–1, 306–8 Jackson, Frank 13, 19, 91 n. 51, 136 nn. 19, 20 Jameson, Frederick 106 Jennings, Richard 32 n. 3, 71 n. 7, 183 n. 2 Johnston, Mark 105 n. 10, 122, 127 n. 10, 128 n. 12, 129–30 Kant, Immanuel, 4–5, 33, 47, 58, 68, 100 n., 101–14, 117–20, 121, 123, 127, 180, 189, 202 Karlsson, Mikael 91 n. 51 Keil, Frank C. 213 Khlenzos, Drew 53 n. 50 Kirk, Robert 286 n. 17 Kirkham, Richard, 156 n., 164 n., 172 n. 26, 174 nn. 31 and 32 Kitcher, Philip 7, 32 n. 3, 72 n. 9, 74, 75, 87, 183 n. 2, 198, 200, 202–7, 210, 211–12, (p. 342) 215, 216, 219, 221–2, 223, 224 n. 26, 226, 228, 238, 240 n. 44, 245 n. 55, 249 n. 60 Kitts, David B. 219 Kitts, David J. 219 Kornblith, Hilary 205 n., 219 n. 13, 244 n. 54, 276 n., 283–4 n., 294 n. 3 Kripke, Saul A. 88, 111 n. 23, 123, 180 n. 39, 213, 214 nn. 1 and 4, 218, 304, 311 Kuhn, Thomas 4, 33, 60 n., 62 n. 6, 68, 88 n. 43, 99–114, 115, 118 n.31, 126, 257, 274, 315 Page 5 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Index Kukla, Andre 70, 75 n. 18, 79, 80–2, 83, 84 n. 35, 85 n. 38, 86 n. 40, 91 n. 51 Lance, Mark 160 n, 169 n. 19, 178 n., 181 n. 42 Lange, Marc 89 n. 48, 246 n. 57 LaPorte, Joseph 230, 249n Larson, Richard 293 n. 2 Latour, Bruno 62 n. 6, 107 n. 16 Laudan, Larry, 39, 47 n. 39, 59 n. 4, 60 n., 73–5, 79 n. 26, 80, 84, 85 nn. 38, 39, 86 n. 40, 87, 90 n. 49, 92 Leeds, Stephen 34 n. 9, 35, 174–5 Lenin, V. I. 52 Leplin, Jarrett 32, 71 nn. 6 and 7, 75, 76 n. 19, 77 n. 22, 80, 82–4, 85 nn. 38 and 39, 86 nn. 40 and 41, 91 n. 51, 183 Lepore, Ernest 40 n. 22, 43–4, 309 n. Levin, Michael 39 n. 20, 73 n. 12, 249 n. Levine, Joseph 254 n. 3 Lewis, David 2, 28–9, 40 n. 21, 136 Lewis, Peter 75 n. 17 Lichtenberg, Judith linguistic realism 299 n. 11 linguistic turn, the 2, 36, 41, 129, 138, 142, 157, 184, 305 n., 315 Lipton, Peter 75, 77 n. 22 Locke, John 94 n. 56, 122, 129, 205 n. 13, 214 n. 4, 221 n. 17, 244 n. 54 Loewer, Barry 40 n. 22, 43–4, 309 n. logic 8–9, 63, 253–4, 257–70, 271, 273 n. 3, 274, 277–82, 285, 287, 289, 290, 291 Luntley, Michael 39 n. 18, 41 n. 27, 46 n. 34 Lycan, William G. 46 n. 36, 154 n., 181 n. 42, 303–5, 311 Lynch, Michael 5, 109 n. 18, 117–20, 163 n. 14 McAllister, J. W. 74 Mackie, John 7, 140 n., 149 n. 28, 183, 186, 189, 191–2, 194 McMichael, Alan 42 n. 24 McMullin, Ernan 69 n. 2, 74, 76 n. 19, 77 n. 22, 78 n. 23, 87 Mallet, James 208, 224 n. 26, 227 n. 30 Margolis, Joseph 3, 43 n. 29 Martin, Michael 99 Mason, H. E. 183 n. 1 mathematics 8, 41, 53–4 n. 49, 63, 253–4, 257–8, 268, 271, 274–5, 282, 291 Matheson, Carl 32 n. 3, 71 n. 7, 183 n. 2 Matthen, Mohan 218, 221, 223 n. 23, 230 n. 33, 234 n. 37 Maxwell, Grover 73 Mayr, Ernst 197 n. 3, 198–9, 206, 214 nn. 1 and 3, 221–2, 224–5, 232, 233, 240 Mellor, D. H. 2, 23–4 Merrill, Edward 297–8 n. 9 Metaphor Thesis (Dummett’s) 41, 53 n. 50 metaphysical realism (Putnam’s) 32 n. 4, 40 metaphysics: a priori 3, 5, 64–5, 110–11, 316–18; naturalized (empirical) 2, 3–4, 65, 110–11, 317–8; of mathematics 275; of nonfactualism 1, 5–6, 72 n. 11, 137–54;

Page 6 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Index put first 2, 3–5, 9–10, 21, 24, 30, 44–6, 53–4, 62–6, 87–8, 96 n. 57, 99–117, 122–4, 136, 138–9, 141–3, 163, 184–5, 186–7, 304–5, 315–20; of truth 1, 6, 150–1, 155–81, see also realism Miller, Alexander 53 n. 50, 111 n. 22 Miller Richard W. 32 n. 3, 72 n. 9, 89 n. 47, 183 n. 2 Millikan, Ruth 111 n. 23, 219, 223 n. 23, 311 Mirage Realism 2, 13–30 Mishler, Brent D. 209–10 n. 19, 211, 215 n. 5, 228 Modern Synthesis, the 208, 219 monophyly 7, 209–12 Moore, G. E. 188; see also Moorean Moorean 3, 26, 62–3, 109–12, 113–14, 122–3, 136, 184–5, 188, 316, 320 moral realism 6–7, 72 n. 11, 139–40, 148–50, 153, 161–8, 171, 182–96 Musgrave, Alan 76 nn. 19 and 20, 78 n. 23 Nagel, Thomas 151 n. 34, 183 n. 1 naturalism, epistemological 1–2, 6–7, 8–9, 77, 253–91; and intuitions 1, 9, 292–302; and meaning 180–1, 312–15, 316–8; and morality 6–7, 182–96; and Commonsense and Scientific Realism 3, 31, 40, 42–8, 55, 62–6, 67, 78, 90, 109 n. 19, 110–4, 122–3, 130, 136, 163 n. 13; (p.343) and reference 310–15, 316–18, 320; and truth 6, 155, 167, 180–1, 313–14 naturalism, metaphysical 64 n. 9, 195–6, 254–5; and morality 6–7, 182–96 natural kinds 7, 197–212, 216 natural kind terms 214 nn. 2, 4, 313 Natural Ontological Attitude (Fine’s) 39 n. 19, 72 n. 9 Neander, Karen 249 n., 311 Neurath’s boat 8, 64, 77, 110–11, 143–4, 267, 268, 280–2, 317–18 nominal essences 205 n., 214 n. 4, 244 n. 54, see also essentialism nominalism 2, 13–30, 124, 139, 164, 187, 201, 216 noncognitivism 1, 5, 6, 72 n. 11, 137, 161–3, 165–7, 171, 174, 184, 186–9, see also moral realism nonfactualism 1, 5–6, 73 n. 11, 137–54, 156, 157, 161–8, 176–7, 186, 265, 269, see also realism; epistemic 265, 269 noumenal world 5, 101–2, 105–7, 113, 117–18, 121, 126 Oberheim, Eric 100 n. 1, 104, 106, 112, 114–17 O’Hara, Robert J. 198, 223, 228 Okasha, Samir 198–9, 214, 216, 218–19, 222 n. 21, 223–9, 232, 236, 240, 242, 245, 249 n. Oliver, Alex 2, 23–4 one over many problem 2, 13–19, 23–30, 201 ontological commitment 2, 14–25, 28, 29, 36, 41, 141–3 operationism 200 n. 7 see also realism Ostrich Nominalism 2, 13–30 Pap, Arthur 19 Page 7 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Index Papineau, David 32 n. 3, 71 n. 7, 91 n. 51, 183 n. 2, 281, 311 Pateman 293 n. 2 Patterson, Douglas 168 n. 17, 176 n., 177 n. 34 Peacocke, Christopher, 9, 287–8 Peirce, C. S. 43–4 pessimistic meta‐induction 4, 67, 72, 75, 86–91, 95–7, 304 Peterson, Todd 297–8 n. 9 Pettit, Philip 5, 122–36 phenomenal worlds 4–5, 33, 101–2, 106–7, 112–13, 118, 121–2 phenomenology 288, 299 physicalism 64 n. 9, 172, 180, 182–96, 254–5 Pietroski, Paul 181 n. 42, 293 n. Plakias, Alexander 192–3 n. 19 Plantinga, Alvin 288 Plato v, 2, 13, 23, 189, 259 pluralism 5, 109 n. 18, 117–20, see also realism; about species 7, 198–200, 202–6, 210, 211–12, 228 Popper, Karl 155, 213 Posner, M. I. 297 possible evidence, nature of 4, 83 n., 91–3 post‐structuralism and post‐modernism 33, 102, 117, 188 Price, Huw 128 n. 11 Prior, Elizabeth, 13 projectivism 137, 161, 186, 192 properties and relations: see universals proto‐science 9–10, 295, 306, 308–15, 319–20 Psillos, Stathis 70, 72 n. 11, 75 n. 18, 76 n. 20, 77, 79, 80 n. 27, 84 n. 36, 87, 91 n. 51, 281 Putnam, Hilary v, 3, 31, 32 n. 4, 33, 34 n. 10, 40, 43–4, 45–6, 48, 49, 51–4, 62 n. 6, 67, 72 n. 9, 73, 79 n. 25, 86, 102, 106, 111 n. 23, 117, 120, 123, 126, 213, 214 nn. 1, 4, 218, 289, 304, 309, 314 putting metaphysics first 2, 3–5, 9–10, 21, 24, 30, 44–6, 53–4, 62–6, 87–8, 96 n. 57, 99– 117, 122–4, 136, 138–9, 141–3, 163, 184–5, 186–7, 304–5, 315–20 Pylyshyn, Zenon W. 293 n. 2 quantum theory 69 n. 2, 268 quasi‐realism (Blackburn’s) 137, 139 n. 13, 140, 161, 186 Quine, W. V., on deflationary truth 34 n. 9, 175; Duhem‐ 72, 82, 113, 138 n.12, 184, 194, 282, 292, 316–17; on essentialism 213; on naturalism and the a priori 1, 63–4, 110–11, 143, 253–7, 264, 266–9, 271 n., 274, 280; on semantic eliminativism 35, 51, 181; on underdetermination 4, 59 n. 3, 60 n., 79 n. 25, 83 n., 91–2; on universals and the One over Many 2, 13–29, 120, 185, 201 Raatikainen, Panu 52 n. 49 Radford, Andrew, 301 n. 15 Railton, Peter 6, 138 n. 9, 149 n. 28, 163 n. 14, 183, 186 n. 14, 189, 195–6 (p.344) real essences 205 n., 214 n. 4, 218, 221 n. 17, see also essentialism realism 1–3, 9, 31–154, 155, 157, 197, 200–2, 244; Page 8 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Index and the aims of science 3, 36, 46–7, 72 n. 10; about artifacts and tools 42, 49–50, 125; in Australia 1, 46 nn. 34, 36, 104; biological 7, 197–212; Commonsense (about observables) 3, 33, 57–66, 67–9, 104; and convergence 7, 47; about entities classified by secondary properties (qualities) 124–9; and epistemic attitudes 69; epistemic doctrines of 70–1; existence dimension of 32–3, 35, 48–51, 55, 57–8, 61, 67–70, 141–3, 156 n., 185–6, 200; Fig‐Leaf 32–3, 107; and global response dependency 5, 121–36, 144 n. 19; and incommensurability 4–5, 99–120; independence dimension of 32–3, 35, 41–2, 48–50, 54, 55–6, 58, 61–2, 68–9, 124–9, 156 n., 185, 200; intentional 303, 305, 319–20; internal 33, 40, 102; and interpretation 3, 6, 33, 41, 52–3, 55, 69 n. 2, 72 n. 11, 140–4, 148 nn. 25 and 26, 151 n. 33, 153, 162, 165, 167, 187–8; issue distinct from semantics 2–3, 33–42, 48, 49–54, 71–2, 73 n. 12, 138–9, 141–3, 178–9, 182–8; about linguistic entities 299 n.11; as a metatheory 47–8, 55–6; Mirage 2, 13–30; moral 6–7, 72 n. 11, 139–40, 148–50, 153, 161–8, 171, 182–96; priority over epistemological and semantic issues 3–5, 44–6, 48, 53–4, 62–6, 87–8, 96 n. 57, 109–12, 122–4, 136, 138–9, 184–5, 304–5; Putnam’s metaphysical 32 n. 4, 40; quasi‐ 137, 139 n. 13, 140, 161, 186; about quantum theory 69 n. 2, 268; Scientific (about unobservables) 4, 33, 57, 66, 67–98, 105, 112; and social entities 125; Strong Scientific 37–8, 70–98; about truth 155–81; about universals 2, 13–30, 124, 139–40, 164–5, 174 n. 29, 185–7, 201, 203–5, see also abduction; Constructivism; eliminativism; idealism; instrumentalism; nonfactualism; operationism; pluralism; skepticism extreme (Cartesian); Worldmaking reduction 16–17, 64 n. 9, 139, 164–6, 168, 172, 173, 174 n. 29, 188, 254 reference, causal theories of 9, 35 n. 12, 40, 71, 88, 113 n., 150, 166, 171 n. 23, 181, 285, 304, 311, 316; deflationary theories of 6, 39, 51, 71, 73 n. 12, 87–8, 172, 174–5, 181, 313–15, 320; description theories of 9, 88, 108–11, 113 n., 114–15, 116, 133–4, 171 n. 23, 214 n. 4, 303–5, 311, 313, 315–17; intuitions about 9, 292–4, 300–1, 306–8, see also, truth; semantics relativism 5, 102, 103, 107, 108, 113, 121–2; moral 185, 189, see also moral realism reliablism 8, 253–63, 269, 272–4, 286, 291 n. 21 Page 9 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Index religion 48, 89, 105 n. 10, 108 n. 17, 189, 272 response dependency 5, 121–36, 144 n. 19 Rey, Georges 8, 48 n. 42, 77 nn. 21, 22, 108 n., 110 n. 21, 154 n. 37, 161 n. 9, 181 n. 42, 182 n. 4, 249 n. 60, 253–63, 268, 269, 273 n. 3, 274, 283 n.14, 290 n. 19, 291 n. 21 Richard, Mark 173 n. 28, 178 n. rigidity 134 n. 16, 169, 214 n. 2 Rorty, Richard 3, 39, 41 n. 27, 46, 48, 51, 55–6, 105 n. 10 Rosen, Gideon 127 n. 9 Rosenberg, A. 80, 218, 222 n. 21 Ruse, Michael 214 n. 1, 230, 241 Sankey, Howard 72 n. 10, 91 n. 51, 102 n. 4, 103 n. 6, 104, 106 n. 13 Sayre‐McCord, Geoffrey 6, 137 nn. 1, 3–6, 138 n. 9, 161 n. 11, 163 n. 14, 182–3, 186 Schaffer, Jonathan 22 n. 11, 30 n Schwartz, Stephen 249 n. Schwitzgebel, Eric 295 Scientific Realism (about unobservables) 4, 33, 57, 66, 67–98, 105, 112; strategy for defending 76, 81–2 secondary properties (qualities) 122, 124–30, 144 n. 19 Segal, Gabriel 293 n. 2 selective realism 139–40, 164, 203–5 Sellars, Wilfred 180 semantic ascent 9–10, 34, 62 n. 7, 109 n. 18, 184, 305, 315–8, 320 semantics, distinct from realism issue 2–3, 33–42, 48, 49–54, 71–2, 73 n. 12, 138–9, 141–3, 178–9; 182–8; priority of metaphysics over 2, 3–5, 9–10, 21, 24, (p.345) 30, 44–6, 48, 53–4, 87–8, 96 n. 57, 109–12, 122–4, 136, 138–9, 141–3, 163, 184–5, 186–7, 304–5, 315–20; truth‐referential 161–2, 179, 253, 312–5, 320; verificationist 43–5, 53–4, 72 n. 11, 138 n. 12, 152–3, 155, 156 n. 3, 162, 179, see also reference; truth, Simpson, George G. 238, 243 Sinnott‐Armstrong, Walter 190 n., 195 n. skepticism, extreme (Cartesian) 3–4, 37, 42–3, 48, 57–66, 68, 76, 79–82, 85–6, 92 n. 54, 93–4, 109 n. 18, 279, 116 n. 15 Sklar, Laurence 95 Slote, Michael 149 n. 28, 154 n. Smart, J. J. C. 73 Soames, Scott 153, 161 n. 9, 163 n. 14, 172–3, 174 n. 31 Sober, Elliott 187 n. 3, 209 n. 18, 213, 214 n. 3, 218–19, 220, 221, 222–3, 224 n. 26, 239–44 Sokal, Robert 198, 224 Sosa, Ernest 292 n. speciation 203 n. 11, 221 n. 18, 224, 226–7, 230 species as individuals 201 n. 9, 216 species concepts 7–8, 197–200, 202–3, 205, 223–33, 239, 247–8 Spencer, N. J., 301 n. 14 Stanford, Kyle 4, 94–8, 199 n. 6, 200, 208 n. 15, 232 n. 36, 244 n. 53 Sterelny, Kim 7, 29 n. 19, 34 n. 9, 40–1 n. 24, 45, 89 n. 44, 103 n. 6, 124 n. 5, 161 n. 13, 174 n. 30, 180 n. 39, 181, 197,198–9, 201, 204, 208–10, 215–16, 218–21, 224, 226–8,

Page 10 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Index 232, 238, 239 n. 43, 243 n. 51, 245, 273 n. 1, 275 n. 6, 295 n. 6, 307 n. 4, 311 n. 9, 313 n. 12 Stich, Stephen P. 9–10, 38 n. 16, 88 n. 43, 249 n. 60, 273 n. 2, 292–3, 301, 303–20 Stove, D. C. 42 n. 24, 48 n. 42, 105 n. 10, 108 n. 17 structuralism 33, 102, 106 Sturgeon, Nicholas L. 149 n. 28, 193–4 subjectivism, moral 150 n., 185, 189, see also moral realism success of science 4, 37, 39, 67, 73–8, 87–90, 96–8, 105 n. 8, 151, 166, 178 Sun, Ron 297–8 n. 9 supervention 145, 165, 188, 192–4 surrealism 75 Tarski, Alfred 6, 34 n. 10, 155, 170–3 taxon problem 7, 199, 217, 225–39, 248 Taylor, Barry 41 n. 27 theory ladenness, epistemic 113, 115–16, 194, 276, 294–6, 301; semantic 113 n. Thomson, Judith Jarvis 192 Thornton, Rosalind 300 n. 13 thought experiments 276–7, 292, 296, 299 Timmons, Mark 185 n. 5 truth, as the aim of science 3, 46–7, 72 n. 10; approximate 37, 71, 73–5, 78, 86–7, 95; correspondence theories of 3, 6, 32, 33–40, 45, 47, 53–4, 71–2, 73 n. 12, 155–81, 183–4; deflationary theories of 5–6; 21–2 n. 10, 34–5, 36, 39, 47, 71, 72 n. 10, 73 n. 12, 150–1, 152–3, 155–81, 184, 313–15, 320; eliminativism about 155–6, 160; epistemic theories of 43–5, 53–4, 72 n. 11, 138 n. 12, 152–3, 155, 156 n. 3, 162, 179, see also semantics; reference truth terms 6, 151 n. 31, 153 n. 36, 156–63, 167, 169–78, 183–4, 314 unconceived alternatives 4, 94–8 underdetermination 3–4, 57–68, 72, 79–86, 91, 95, 109 n. 18, 262, 316 n. 15 universals 2, 13–30, 124, 139–40, 164–5, 174 n. 29, 185–7, 201, 203–5 use/mention distinction 6, 20, 157, 158, 169 n. 20, 170–6 Van Fraassen, Bas C. 3, 4, 31, 33, 39 n. 19, 46 n. 37, 47 n. 40, 48 n. 43, 70 n. 4, 71 n. 6, 72 n. 10, 75–6, 77 n. 22, 79 n. 25, 80–1, 83 n. 33, 91, 271 n., 279 Van Valen, Leigh 198–9, 224 Vasiliou, Iakovos 249 n. Velleman, David J. 129 n. 13 Wade, Nicholas 219 Warenski, Lisa 273–4 n. 3 Webster, Gerry 215 n. 8 Wedgwood, Ralph 129 n. 13 Whorf, Benjamin Lee 102, 107 n. 16 Wiggins, David 213–14 Wilkins, John 215 n. 7, 249 n. Williams, D. C. 55, 253 n. 2 Williams, Michael 3, 48–50, 141–2 Williamson, Timothy 84 n. 34 Page 11 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.

Index Wilson, Robert 199 n. 5, 203 n. 10, 214 n. 3, 219, 226, 228, 232–4, 237, 239, 249 n. Winsor, Mary P. 214 n. 1 Wisniewski, E. J. 296 (p.346) Wittgenstein, Ludwig 109 n. 18, 123, 180, 313 Wolterstorff, Nicholas 42 n. 24, 105 n. 10 Woolgar, Steve 62 n. 6, 107 n. 16 Worldmaking 5, 109 n. 18, 121–36, 205, 242, 244 Worrall, J. 87 Wright, Crispin 39 n. 18, 122 n., 137 nn. 2–6, 138–9 nn. 9 and 12, 145 n. 22, 161 n. 11, 163 n. 14, 186 n. 8–12, 14, 281 Young, Liane 191 n.

Access brought to you by:

Page 12 of 12

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.