Datasets:

kothasuhas
/

philosophy-textbooks-14

Dataset card Viewer Files Files and versions Community

Split (1)

train · 11.1k rows

text stringlengths 646 5.29k	textbook_name stringclasses 14 values
The Blackwell Companion to Philosophy Blackwell Companions to Philosophy This outstanding student reference series offers a comprehensive and authoritative survey of philosophy as a whole. Written by today’s leading philosophers, each volume provides lucid and engaging coverage of the key ﬁgures, terms, topics, and problems of the ﬁeld. Taken together, the volumes provide the ideal basis for course use, represent- ing an unparalleled work of reference for students and specialists alike. Already published in the series: 1 The Blackwell Companion to Philosophy, Second Edition Edited by Nicholas Bunnin and E. P. Tsui-James 2 A Companion to Ethics Edited by Peter Singer 3 A Companion to Aesthetics Edited by David Cooper 4 A Companion to Epistemology Edited by Jonathan Dancy and Ernest Sosa 5 A Companion to Contemporary Political Philosophy Edited by Robert E. Goodin and Philip Pettit 6 A Companion to Philosophy of Mind Edited by Samuel Guttenplan 7 A Companion to Metaphysics Edited by Jaegwon Kim and Ernest Sosa 8 A Companion to Philosophy of Law and Legal Theory Edited by Dennis Patterson 9 A Companion to Philosophy of Religion Edited by Philip L. Quinn and Charles Taliaferro 10 A Companion to the Philosophy of Language Edited by Bob Hale and Crispin Wright 11 A Companion to World Philosophies Edited by Eliot Deutsch and Ron Bontekoe 12 A Companion to Continental Philosophy Edited by Simon Critchley and William Schroeder 14 A Companion to Cognitive Science Edited by William Bechtel and George Graham 15 A Companion to Bioethics Edited by Helga Kuhse and Peter Singer 16 A Companion to the Philosophers Edited by Robert L. Arrington 17 A Companion to Business Ethics Edited by Robert E. Frederick 18 A Companion to the Philosophy of Science Edited by W. H. Newton-Smith 19 A Companion to Environmental Philosophy Edited by Dale Jamieson 20 A Companion to Analytic Philosophy Edited by A. P. Martinich and David Sosa 21 A Companion to Genethics Edited by Justine Burley and John Harris 22 A Companion to Philosophical Logic Edited by Dale Jacquette 23 A Companion to Early Modern Philosophy Edited by Steven Nadler Forthcoming A Companion to African American Philosophy Edited by Tommy Lott and John Pittman A Companion to African Philosophy Edited by Kwasi Wiredu A Companion to Ancient Philosophy Edited by Mary Louise Gill 13 A Companion to Feminist Philosophy Edited by Alison M. Jaggar and Iris Marion Young A Companion to Medieval Philosophy Edited by Jorge J. E. Gracia, Greg Reichberg, and Timothy Noone The Blackwell Companion to Philosophy SECOND EDITION Edited by NICHOLAS BUNNIN and E. P. TSUI-JAMES Copyright © 1996, 2003 Blackwell Publishers Ltd, a Blackwell Publishing company Editorial matter, selection and arrangement copyright © Nicholas Bunnin and Eric Tsui-James 1996, 2003 First edition published 1996 Reprinted 1996 (twice), 1998, 1999, 2002 Second edition published 2003 350 Main Street, Malden, MA 02148-5018, USA 108 Cowley Road, Oxford OX4 1JF, UK 550 Swanston Street, Carlton South, Victoria 3053, Australia Kurfürstendamm 57, 10707 Berlin, Germany The right of Nicholas Bunnin and Eric Tsui-James to be identiﬁed as the Authors of the Editorial Material in this work has been asserted in accordance with the UK Copyright, Designs and Patents Act 1988. 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, electronic, mechanical, photocopying, recording or otherwise, except as permitted by the UK Copyright, Designs and Patents Act 1988, without the prior permission of the publisher. Library of Congress Cataloging-in-Publication Data The Blackwell companion to philosophy / edited by Nicholas Bunnin and E. P. Tsui-James. –– 2nd ed. p. cm. –– (Blackwell companions to philosophy) Includes bibliographical references and index. ISBN 0–631–21907–2 –– ISBN 0–631–21908–0 (pbk.) I. Bunnin, Nicholas. 1. Philosophy. II. Tsui-James, E. P. III. Series. B21 .B56 2003 100––dc21 2002023053 A catalogue record for this title is available from the British Library.	Blackwell
Set in 10 on 121/2 pt Photina by SNP Best-set Typesetter Ltd, Hong Kong Printed and bound in the United Kingdom by T. J. International, Padstow, Cornwall For further information on Blackwell Publishing, visit our website: http://www.blackwellpublishing.com For Antonia and Oliver Bunnin and Jamie Perry Contents Preface to the Second Edition Preface to the First Edition Notes on Contributors Contemporary Philosophy in the United States – John R. Searle Contemporary Philosophy: A Second Look – Bernard Williams Part I Areas of Philosophy 1 Epistemology – A. C. Grayling 2 Metaphysics – Simon Blackburn, with a section on Time by Robin Le Poidevin 3 Philosophy of Language – Martin Davies 4 Philosophy of Logic – A. W. Moore 5 Philosophy of Mind – William G. Lycan 6 Ethics – John Skorupski 7 Aesthetics – Sebastian Gardner 8 Political and Social Philosophy – David Archard 9 Philosophy of Science – David Papineau 10 Philosophy of Biology – Elliott Sober 11 Philosophy of Mathematics – Mary Tiles 12 Philosophy of Social Science – Martin Hollis 13 Philosophy of Law – N. E. Simmonds 14 Philosophy of History – Leon Pompa 15 Philosophy of Religion – Charles Taliaferro 16 Applied Ethics – John Haldane 17 Bioethics, Genethics and Medical Ethics – Rebecca Bennett, Charles A. Erin, John Harris and Søren Holm 18 Environmental Ethics – Holmes Rolston, III 19 Business Ethics – Georges Enderle 20 Philosophy and Feminism – Jean Grimshaw and Miranda Fricker 21 Ethnicity, Culture and Philosophy – Robert Bernasconi ix x xii 1 23 35 37 61 90 147 173 202 231 257 286 317 345 375 403 428 453 490 499 517 531 552 567 CONTENTS Part II History of Philosophy 22 Ancient Greek Philosophy – Robert Wardy 23 Plato and Aristotle – Lesley Brown 24 Medieval Philosophy – Jorge J. E. Gracia 25 Bacon – Stephen Gaukroger 26 Descartes and Malebranche – Richard Francks and George Macdonald Ross 27 Spinoza and Leibniz – Richard Francks and George Macdonald Ross 28 Hobbes – Tom Sorell 29 Locke – R. S. Woolhouse 30 Berkeley – Howard Robinson 31 Hume – Peter Jones 32 Kant – David Bell 33 Hegel – Michael Inwood 34 Marx – Richard Norman 35 Bentham, Mill and Sidgwick – Ross Harrison 36 Pragmatism – Susan Haack 37 Frege and Russell – R. M. Sainsbury 38 Moore – Thomas Baldwin 39 Wittgenstein – David Pears 40 Nietzsche – David E. Cooper 41 Husserl and Heidegger – Taylor Carman 42 Sartre, Foucault and Derrida – Gary Gutting Glossary Appendix Index 583 585 601 619 634 644 658 671 682 694 709 725 741 750 759 774 790 805 811 827 842 860 875 893 905 Preface to the Second Edition We thank readers for their gratifying response to the ﬁrst edition of the Companion. The second edition provides new chapters on Philosophy of Biology; Bioethics, Genethics and Medical Ethics; Environmental Ethics; Business Ethics; Ethnicity, Culture and Philosophy; Plato and Aristotle; Francis Bacon; Nietzsche; Husserl and Heidegger; and Sartre, Foucault and Derrida. There are signiﬁcant revisions or extensions to chapters on Metaphysics, Philosophy of Language, Philosophy of Mind, Political and Social Philosophy, Philosophy of Religion, Philosophy and Feminism, and Hobbes. The discussion of Descartes, Spinoza and Leibniz is now divided between two chapters, and in a new section Malebranche is considered along with Descartes in the ﬁrst of these. A longer chapter on Medieval Philosophy replaces the chapter by C. F. J. Martin, who was unavailable to extend his work. We welcome our new contributors and hope that readers will continue to be challenged and delighted by the Companion as a whole. Nicholas Bunnin E. P. Tsui-James Preface to the First Edition This Companion complements the Blackwell Companions to Philosophy series by presenting a new overview of philosophy prepared by thirty-ﬁve leading British and American philosophers. Introductory essays by John Searle and Bernard Williams, which assess the changes that have shaped the subject in recent decades, are followed by chapters exploring central problems and debates in the principal subdisciplines of philosophy and in specialized ﬁelds, chapters concerning the work of great historical	Blackwell
ﬁgures and chapters discussing newly developing ﬁelds within philosophy. Throughout the course of its chapters, the Companion examines the views of many of the most widely inﬂuential ﬁgures of contemporary philosophy. Although wide-ranging, the Companion is not exhaustive, and emphasis is placed on developments in Anglo-American philosophy in the latter part of the twentieth century. A premise underlying the Companion is that major participants in philosoph- ical debate can provide accounts of their own ﬁelds that are stimulating, accessible, stylish and authoritative. In its primary use, the Companion is an innovative textbook for introductory courses in philosophy. Teachers can use the broad coverage to select chapters in a ﬂexible way to support a variety of courses based on contemporary problems or the historical devel- opment of the subject. Specialist chapters can be used selectively to augment standard introductory topics or to prepare students individually for term papers or essays. Chap- ters include initial summaries, boxed features, cross-references, suggestions for further reading, references and discussion questions. In addition, terms are marked for a common glossary. These features and the problem-setting nature of the discussions encourage students to see the subject as a whole and to gain conﬁdence that explo- rations within philosophy can lead to unexpected and rewarding insights. In this aspect, the Companion reﬂects the contributors’ experience of small group teaching, in which arguments and perspectives are rigorously tested and in which no solution is imposed. In its secondary use, the Companion will accompany students throughout their undergraduate careers and will also serve the general reader wishing to understand the central concepts and debates within philosophy or its constituent disciplines. Students are unlikely to read the whole volume in their ﬁrst year of study, but those continuing with philosophy will ﬁnd their appreciation of the work deepening over time PREFACE TO THE FIRST EDITION as they gain insight into the topics of the more advanced chapters. The Companion will help them to formulate questions and to see connections between what they have already studied and new terrain. In its ﬁnal use, the Companion bears a special relationship to the Blackwell Com- panions to Philosophy series. Many readers will wish to read the integrated discussions of the chapters of the present Companion for orientation before turning to the detailed, alphabetically arranged articles of the volumes in the Companion series. Although con- ceived as a separate volume, the Companion to Philosophy will serve as a useful guide to the other excellent Companions in what amounts to a comprehensive encyclopedia of philosophy. The general reader might begin with the introductory essays and turn to chapters on Epistemology, Metaphysics, Ethics and Political and Social Philosophy, or to histori- cal chapters from Ancient Greek Philosophy to Hume. Cross-references and special interests will lead readers to other chapters. Cross-references in the text are marked in small capitals followed by a chapter number or page numbers in parentheses: Ethics (chapter 6) or Probability (pp. 308–11). We have used our judgement in marking terms appearing many times in the text for cross-references, and hope that we have supplied guidance without distracting readers. The Companion also provides a glossary of 210 terms and a comprehensive index. Both appear at the end of the volume, and readers are advised to use them reg- ularly for help in reading the chapters. When an author does not refer to a book by its ﬁrst edition, a recent publication is cited in the text, and the original date of publica- tion (or in some cases of composition) will appear in square brackets in the references. As editors, we are fully aware of our good fortune in attracting superb contributors.	Blackwell
The complexity of their insights and the clarity of their presentations are the chief attractions of the Companion. We appreciate their care in making the difﬁcult not only accessible but delightful as well. We also wish to thank the Departments of Philosophy at the University of Essex and the University of Hong Kong for their support through- out the preparation of this volume. We are especially grateful to Laurence Goldstein, Tim Moore and Frank Ciofﬁ for their comments and advice. A version of the Com- panion is published in Chinese by the Shandong Academy of Social Sciences, and we appreciate the friendly co-operation of our Chinese co-editors. Our cover illustration, R. B. Kitaj’s philosophically resonant If Not, Not, is a work by an American artist working in London during the period that provides the main focus of our volume. Nicholas Bunnin E. P. Tsui-James xi Notes on Contributors David Archard is Reader in Moral Philosophy and Director of the Centre for Ethics, Philosophy and Public Affairs at the University of St Andrews. He is the author of Sexual Consent (1998) and co-editor of The Moral and Political Status of Children: New Essays (2002). Thomas Baldwin is Professor of Philosophy at the University of York. He previously taught at the University of Cambridge (where he was Fellow of Clare College) and at Makerere University. He has published G. E. Moore (1990) and Contemporary Philoso- phy: Philosophy in English since 1945 (2001) in addition to many articles on issues in metaphysics and the philosophy of language. David Bell is Professor of Philosophy at the University of Shefﬁeld. He is the author of works on Frege, Husserl and Kant. His interests include the foundations of arithmetic, solipsism and the nature and origins of the analytic tradition. Rebecca Bennett is Lecturer in Bioethics at the Centre for Social Ethics and Policy, School of Law, University of Manchester. She edited (with Charles Erin) HIV and AIDS: Testing, Screening and Conﬁdentiality (1999). Robert Bernasconi is Moss Professor of Philosophy at the University of Memphis. He is the author of The Question of Language in Heidegger’s History of Being (1985) and Heidegger in Question (1993) as well as numerous articles on Hegel and on twentieth- century European philosophy. He has edited collections of essays on Derrida and on Levinas and most recently Race (2001). Simon Blackburn is Professor of Philosophy at the University of Cambridge. A former editor of the journal Mind, he has written Ruling Passions (1998), Spreading the Word (1984), Essays in Quasi Realism (1993) and The Oxford Dictionary of Philosophy (1994). His current work concerns problems of realism and its alternatives as they have emerged in historical and contemporary work. Lesley Brown is Tutorial Fellow in Philosophy, Somerville College, University of Oxford. She has written on Plato, especially his metaphysics and epistemology, and on ancient philosophy of language. CONTRIBUTORS Nicholas Bunnin is Director of the Philosophy Project at the Institute for Chinese Studies, University of Oxford and previously taught at the University of Glasgow and the University of Essex. He compiled (with Jiyuan Yu) the Dictionary of Western Philos- ophy: English–Chinese (2001) and edited (with Chung-ying Cheng) Contemporary Chinese Philosophy (2002). His main interests are in metaphysics, the philosophy of mind and political philosophy. Taylor Carman is Assistant Professor of Philosophy at Barnard College, Columbia University. He is co-editor of The Cambridge Companion to Merleau-Ponty (forthcoming) and the author of Heidegger’s Analytic: Interpretation, Discourse, and Authenticity in ‘Being and Time’ (forthcoming), and of other articles on Husserl, Heidegger, and Merleau-Ponty. David E. Cooper is Professor of Philosophy at the University of Durham and Director of the Durham Institute of Comparative Ethics. His books include Metaphor (1986),	Blackwell
Existentialism: A Reconstruction (2nd revd edn 2000), World Philosophies: An Historical Introduction (2nd revd edn 2002) and The Measure of Things: Humanism, Humility and Mystery (2002). Martin Davies is Professor of Philosophy in the Research School of Social Sciences, Australian National University. He was formerly Wilde Reader in Mental Philosophy at the University of Oxford. He has published widely in the areas of philosophy of language, mind and psychology. Georges Enderle is Arthur and Mary O’Neil Professor of International Business Ethics at the University of Notre Dame. His books include Business Students Focus on Ethics (1993), translated into Portuguese (1997) and Chinese (2001). Charles A. Erin is Senior Lecturer in Applied Philosophy and Fellow of the Institute of Medicine, Law and Bioethics at the University of Manchester. He has written widely on topics in bioethics and edited (with Rebecca Bennett) HIV and AIDS: Testing, Screening and Conﬁdentiality (1999). Richard Francks is Director of Undergraduate Studies in Philosophy at the University of Leeds. His main interests are in epistemology, the history of philosophy and the philosophy of history. Miranda Fricker is Lecturer in Philosophy at Birkbeck College, University of London and was previously Lecturer in Philosophy and British Academy Postdoctoral Fellow at Heythrop College, University of London. She has published articles in epistemology, ethics and social philosophy, and edited (with Jennifer Hornsby) The Cambridge Companion to Feminism in Philosophy (2000). Her current work focuses on the idea of an ethics of epistemic practice. Sebastian Gardner is Reader in Philosophy at University College, London. He is the author of Irrationality and the Philosophy of Psychoanalysis (1993) and Kant and the Critique of Pure Reason (1999). His interests lie in aesthetics, psychoanalysis and the history of philosophy. xiii CONTRIBUTORS Stephen Gaukroger is Professor of History of Philosophy and History of Science at the University of Sydney. He is the author of Explanatory Structures (1978), Cartesian Logic (1989), Descartes: An Intellectual Biography (1995), Francis Bacon and the Transformation of Early Modern Philosophy (2000) and Descartes’ System of Natural Philosophy (2001). He has also edited four collections of essays and published translations of Descartes and Arnaud. Jorge J. E. Gracia is a State University of New York Distinguished Professor and holds the Samuel F. Capon Chair in the Department of Philosophy, State University of New York, University at Buffalo. He has written widely on medieval philosophy, metaphysics, philosophical historiography, philosophy of language and philosophy in Latin America. His books include Introduction to the Problem of Individuation in the Early Middle Ages (2nd revd edn 1988), Individuality: An Essay on the Foundations of Metaphysics (1988), Philosophy and Its History: Issues in Philosophical Historiography (1992), A Theory of Textuality: The Logic and Epistemology (1995), Texts: Ontological Status, Identity, Author, Audience (1996) and Metaphysics and Its Task: The Search for the Categorical Foundation of Knowledge (1999). A. C. Grayling is Reader in Philosophy at Birkbeck College, London, and Supernu- merary Fellow at St Anne’s College, Oxford. Among his books are An Introduction to Philosophical Logic (3rd edn 1992), The Refutation of Scepticism (1985), Berkeley: The Central Arguments (1986), Wittgenstein (1988), Russell (1993), Moral Values (1998), The Quarrel of the Age (2000) and The Meaning of Things (2001). He has edited Philosophy: A Guide Through the Subject (1995) and Philosophy: Further Through the Subject (1998). Jean Grimshaw taught Philosophy and Women’s Studies at the University of the West of England, Bristol. She is the author of Feminist Philosophers: Women’s Perspectives on Philosophical Traditions (1986) and a number of articles, mainly on feminism and phi-	Blackwell
losophy. She has edited (with Jane Arthurs) Women’s Bodies: Discipline and Transgression (1999). Gary Gutting is Professor of Philosophy at the University of Notre Dame. He is the author of Religious Belief and Religious Skepticism (1982), Michel Foucault’s Archaeology of Knowledge (1989) and French Philosophy in the Twentieth Century (2001). Susan Haack formerly Professor of Philosophy at the University of Warwick, currently Professor of Philosophy at the University of Miami, is the author of Deviant Logic (1974), Philosophy of Logic (1978), Evidence and Inquiry: Towards Reconstruction in Epistemology (1993) and Manifesto of a Passionate Moderate: Unfashionable Essays (1998). Her main areas of interest are the philosophy of logic and language, episte- mology and metaphysics and pragmatism. She is a past President of the Charles Peirce Society. John Haldane is Professor of Philosophy and formerly Director of the Centre for Philosophy and Public Affairs at the University of St Andrews. He has published widely in the philosophy of mind, the philosophy of value and the history of philoso- phy. He is co-author with J. J. C. Smart of Atheism and Theism (1996) in the Blackwell Great Debates in Philosophy series. xiv CONTRIBUTORS John Harris is Sir David Alliance Professor of Bioethics, Institute of Medicine, Law and Bioethics, University of Manchester. He is a member of the United Kingdom Human Genetics Commission and of the Ethics Committee of the British Medical Association. He was a Founder Director of the International Association of Bioethics and a founder member of the Board of the journal Bioethics. Among his books are The Value of Life (1985) and Clones, Genes and Immortality (1998) (a revised edition of Wonderwoman and Superman, 1992), and he is editor of Bioethics (2001) in the Oxford Readings in Phi- losophy series. Ross Harrison teaches philosophy at the University of Cambridge, where he is also a Fellow of King’s College. Among his publications are Bentham (1983), Democracy (1993) and (as editor and contributor) Henry Sidgwick (2001). Martin Hollis was Professor of Philosophy at the University of East Anglia, Norwich. He specialized in the philosophy of social science, especially in topics to do with ratio- nality. Among his books are Models of Man (1977), The Cunning of Reason (1987), The Philosophy of Social Science (1994), Reason in Action (1995), Trust Within Reason (1998) and Pluralism and Liberal Neutrality (1999). The last two volumes were published after his untimely death in 1998. Søren Holm is Professor of Clinical Bioethics at the University of Manchester. He is the author of Ethical Problems of Clinical Practice: The Ethical Reasoning of Health Care Professionals (1997) and has edited (with Inez de Beaufort and Medard Hilhorst) In the Eye of the Beholder: Ethics and Medical Change of Appearance (1996) and (with John Harris) The Future Of Human Reproduction: Ethics, Choice and Regulation (1998). Michael Inwood is Tutorial Fellow in Philosophy at Trinity College, Oxford. He has published several books on Hegel. His other interests include ancient philosophy and Heidegger. He is especially interested in the interconnections between Greek and German philosophy. Peter Jones was Professor of Philosophy and Director of the Institute for Advanced Studies in the Humanities at the University of Edinburgh. He is the author of numer- ous works, including Hume’s Sentiments (1982). Robin Le Poidevin is Professor of Metaphysics at the University of Leeds, where he was Head of the School of Philosophy 1988–2001. He is the author of Change, Cause and Contradiction: A Defence of the Tenseless Theory of Time (1991) and Arguing for Atheism: An Introduction to the Philosophy of Religion (1996) and has edited Questions of Time and Tense (1998) and (with Murray MacBeath) The Philosophy of Time (1993). William G. Lycan is William Rand Kenan, Jr Professor of Philosophy at the University	Blackwell
of North Carolina. He has published a number of books, including Consciousness (1987), Judgement and Justiﬁcation (1988) and Consciousness and Experience (1996). He is the editor of Mind and Cognition (1990). His interests are in the philosophy of mind, the philosophy of language and epistemology. A. W. Moore is Tutorial Fellow in Philosophy at St Hugh’s College, Oxford. He is the author of The Inﬁnite (2nd edn 2001) and Points of View (1997). He has also edited two collections of essays: Meaning and Reference (1993) and Inﬁnity (1993). xv CONTRIBUTORS Richard Norman is Professor of Moral Philosophy at the University of Kent. His pub- lications include The Moral Philosophers (1983), Free and Equal (1987) and Ethics, Killing and War (1995). David Papineau is Professor of Philosophy of Science at King’s College, London. He has published widely in epistemology, the philosophy of mind and the philosophy of science. His books include Reality and Representation (1987), Philosophical Naturalism (1993), Introducing Consciousness (2000) and Thinking About Consciousness (2002). David Pears is Emeritus Professor of Philosophy at the University of Oxford. His most recent publications are The False Prison: A Study in the Development of Wittgenstein’s Phi- losophy (2 vols, 1987 and 1988) and Hume’s System: An Examination of Book I of the Treatise (1991). His other interests include entomology and the visual arts. Leon Pompa was Professor of Philosophy at the University of Birmingham. His research interests include the history of philosophy and the philosophy of history. He has published a number of articles on the problems of fact, value and narrative in history and on Descartes, Vico, Kant, Hegel, Marx, Collingwood and Wittgenstein. He co-edited with W. H. Dray Substance and Form in History: Essays in Philosophy of History (1981), was editor and translator of Vico: A Study of the ‘New Science’ (2nd edn 1990) and is the author of Human Nature and Historical Knowledge: Hume, Hegel and Vico (1990). Howard Robinson is Professor of Philosophy, Central European University, Budapest. He was previously Soros Professor of Philosophy at the Eötvös Loránd University, Budapest and Reader in Philosophy at the University of Liverpool. His main interests are in the philosophy of mind and in idealism. He is the author of Matter and Sense (1982) and Perception (1994), and co-author (with John Foster) of Essays on Berkeley (1985). He edited Objections to Physicalism (1991) and is currently editing Berkeley’s Principles and Three Dialogues for Oxford University Press’s World Classics series. Holmes Rolston, III is University Distinguished Professor and Professor of Philosophy at Colorado State University. He has written seven books, most recently Genes, Genesis and God (1999), Philosophy Gone Wild (1986), Environmental Ethics (1988), Science and Religion: A Critical Survey (1987) and Conserving Natural Value (1994). He gave the Gifford Lectures, University of Edinburgh, 1997–8, has lectured on seven continents, is featured in Joy A. Palmer’s (ed.) Fifty Key Thinkers on the Environment and is past and founding president of the International Society for Environmental Ethics. George MacDonald Ross is Senior Lecturer in the Department of Philosophy at the University of Leeds and Director of the Philosophical and Religious Studies Subject Centre of the Learning and Teaching Support Network. He has written extensively on Leibniz and other seventeenth- and eighteenth-century philosophers, and is the author of Leibniz (1984). R. M. Sainsbury is Stebbing Professor of Philosophy at King’s College, London and was editor of the journal Mind for several years until 2000. He has published Russell (1979), Paradoxes (1995) and Logical Forms (2000). His main interests are in philosophical logic and the philosophy of language. xvi CONTRIBUTORS John R. Searle is Mills Professor of Mind and Language at the University of California	Blackwell
where he has been a faculty member since 1959. Before that, he was a lecturer at Christ Church, Oxford, and he received all his university degrees from Oxford. Most of his work is in the philosophy of mind, the philosophy of language, and social philosophy. His most recently published books are Rationality in Action (2001) and Mind, Language and Society (1998). He is the author of several other important books, including Speech Acts: An Essay in the Philosophy of Language (1969), Expression and Meaning: Studies in the Theory of Speech Acts (1985), Intentionality (1983), Minds, Brains and Science, the 1984 Reith Lectures (1989), The Rediscovery of Mind (1992) and The Construction of Social Reality (1995). N. E. Simmonds is a Fellow of Corpus Christi College, Cambridge where he lectures in law. His interests include the philosophy of law and political philosophy. He has pub- lished The Decline of Judicial Reason (1984), Central Issues in Jurisprudence (1986) and numerous articles on the philosophy of law. John Skorupski is Professor of Moral Philosophy at the University of St Andrews. He is the author of John Stuart Mill (1989) and English-Language Philosophy 1750–1945 (1993). His most recent book is Ethical Explorations (1999). Elliott Sober is Hans Reichenbach Professor and Henry Vilas Research Professor at the University of Wisconsin, Madison and Centennial Professor at the London School of Economics. He is the author of The Nature of Selection (1984), Reconstructing the Past (1988), Philosophy of Biology (1993) and (with David S. Wilson) Unto Others: Evolution and Psychology of Unselﬁsh Behaviour (1988). Tom Sorell is Professor of Philosophy at the University of Essex. He is the author of Hobbes (1986), Descartes (1987), Moral Theory and Capital Punishment (1987), Scientism (1991), (with John Hendry) Business Ethics (1994) and Moral Theory and Anomaly (2000). He is the editor of The Rise of Modern Philosophy (1993), The Cambridge Com- panion to Hobbes (1995), Health Care, Ethics, and Insurance (1998), Descartes (1999), and (with John Rogers) Hobbes and History (2000). Charles Taliaferro is Professor of Philosophy at St Olaf College, Northﬁeld, Minnesota and the author of Consciousness and the Mind of God (1994), Contemporary Philosophy of Religion (1999) and the co-editor of A Companion to Philosophy of Religion (1998). Mary Tiles is Professor of Philosophy at the University of Hawaii at Manoa. Her inter- ests include the history and philosophy of mathematics, science and technology and their interactions with culture (European and Chinese). She has published Living in a Technological Culture (with Hans Oberdiek) (1995), An Introduction to Historical Episte- mology (with James Tiles) (1993), Mathematics and the Image of Reason (1991) and Bachelard: Science and Objectivity (1984). Eric P. Tsui-James studied as a postgraduate at Oriel College, Oxford. He taught phi- losophy at St Hilda’s College, Oxford, for two years before moving to the University of Hong Kong in 1990. He has published work on the metaphysics of mathematics, but his research interests now centre around the work of William James, especially the nineteenth-century psychological and physiological contexts of his radical empiricism. xvii CONTRIBUTORS Robert Wardy teaches philosophy and classics at St Catharine’s College, Cambridge. He has published in the ﬁelds of ancient Greek philosophy and rhetoric, Latin litera- ture, the philosophy of language and Chinese philosophy. Bernard Williams is Monroe Deutsch Professor of Philosophy, University of Califor- nia, Berkeley, and was White’s Professor of Moral Philosophy and a Fellow of Corpus Christi College, Oxford. His works include Morality (1972), Problems of the Self (1973), Descartes: The Project of Pure Enquiry (1978), Moral Luck (1981), Ethics and the Limits of Philosophy (1985), Shame and Necessity (1993) and Making Sense of Humanity (1995). R. S. Woolhouse is Professor of Philosophy at the University of York. He is the author	Blackwell
of Locke’s Philosophy of Science and Knowledge (1971), Locke (1983), The Empiricists (1988) and Descartes, Spinoza, Leibniz: The Concept of Substance in Seventeenth-Century Philosophy (1993). xviii Contemporary Philosophy in the United States J O H N R. S E A R L E Philosophy as an academic discipline in America has considerably fewer practitioners than do several other subjects in the humanities and the social sciences, such as sociology, history, English, or economics; but it still shows enormous diversity. This variety is made manifest in the original research published by professional philosophers, whose differing points of view are expressed in the large number of books published each year, as well as in the many professional philosophy journals. There are over two thousand colleges and universities in the United States, of which nearly all have philosophy departments, and the number of professional philosophers is correspondingly large. Because of this diversity, any generalizations about the discipline as a whole, which I am about to make, are bound to be misleading. The subject is too vast and complex to be describable in a single essay. Furthermore, anyone who is an active participant in the current controversies, as I am, necessarily has a perspective conditioned by his or her own interests, commitments and convictions. It would be impossible for me to give an ‘objective’ account. I am not therefore in what follows trying to give a neutral or disinterested account of the contemporary philosophical scene; rather I am trying to say what in the current developments seems to me important. In spite of its enormous variety, there are certain central themes in contemporary American philosophy. The dominant mode of philosophizing in the United States is called ‘analytic philosophy’. Without exception, the best philosophy departments in the United States are dominated by analytic philosophy, and among the leading philosophers in the United States, all but a tiny handful would be classiﬁed as analytic philosophers. Practitioners of types of philosophizing that are not in the analytic tradition – such as phenomenology, classical pragmatism, existentialism, or Marxism – feel it necessary to deﬁne their position in relation to analytic philosophy. Indeed, analytic philosophy is the dominant mode of philosophizing not only in the United States, but throughout the entire English-speaking world, including Great Britain, Canada, Australia and New Zealand. It is also the dominant mode of philosophizing in Scandinavia, and it is also becoming more widespread in Germany, France, Italy and throughout Latin America. I personally have found that I can go to all of these parts of the world and lecture on subjects in contemporary analytic philosophy before audiences who are both knowledgeable and well trained in the techniques of the discipline. JOHN R. SEARLE 1 Analytic Philosophy What, then, is analytic philosophy? The simplest way to describe it is to say that it is primarily concerned with the analysis of meaning. In order to explain this enterprise and its signiﬁcance, we need ﬁrst to say a little bit about its history. Though the United States now leads the world in analytic philosophy, the origins of this mode of philosophizing lie in Europe. Speciﬁcally, analytic philosophy is based on the work of Gottlob Frege, Ludwig Wittgenstein, Bertrand Russell and G. E. Moore, as well as the work done by the logical positivists of the Vienna Circle in the 1920s and 1930s. Going further back in history, one can also see analytic philosophy as a natural descend- ant of the empiricism of the great British philosophers Locke, Berkeley and Hume, and of the transcendental philosophy of Kant. In the works of philosophers as far back as Plato and Aristotle, one can see many of the themes and presuppositions of the	Blackwell
methods of analytic philosophy. We can best summarize the origins of modern analytic philosophy by saying that it arose when the empiricist tradition in epistemology, together with the foundationalist enterprise of Kant, were tied to the methods of logical analysis and the philosophical theories invented by Gottlob Frege in the late nineteenth century. In the course of his work on the foundations of mathematics, Frege invented symbolic logic in its modern form and developed a comprehensive and profound philosophy of language. Though many of the details of his views on language and mathematics have been superseded, Frege’s work is crucial for at least two reasons. Firstly, by inventing modern logic, speciﬁcally the predicate calculus, he gave us a primary tool of philosophical analysis; and, secondly, he made the philosophy of language central to the entire philosophical enterprise. From the point of view of analytic philosophy, Frege’s work is the greatest single philo- logical sophical achievement of analysis were later augmented by the ordinary language analysis inspired by the work of Moore and Wittgenstein and are best exempliﬁed by the school of lin- guistic philosophy that ﬂourished in Oxford in the l950s. In short, analytic phi- losophy attempts to combine certain traditional philosophical themes with modern techniques. the nineteenth century. Fregean techniques of Analytic philosophy has never been ﬁxed or stable, because it is intrinsically self- critical and its practitioners are always challenging their own presuppositions and con- clusions. However, it is possible to locate a central period in analytic philosophy – the period comprising, roughly speaking, the logical positivist phase immediately prior to the 1939–45 war and the postwar phase of linguistic analysis. Both the prehistory and the subsequent history of analytic philosophy can be deﬁned by the main doctrines of that central period. In the central period, analytic philosophy was deﬁned by a belief in two linguistic distinctions, combined with a research programme. The two distinctions are, ﬁrstly, that between analytic and synthetic propositions, and, secondly, that between descriptive and evaluative utterances. The research programme is the traditional philosophical research programme of attempting to ﬁnd foundations for such philosophically prob- lematic phenomena as language, knowledge, meaning, truth, mathematics and so on. 2 CONTEMPORARY PHILOSOPHY IN THE UNITED STATES One way to see the development of analytic philosophy over the past thirty years is to regard it as the gradual rejection of these two distinctions, and a corresponding rejec- tion of foundationalism as the crucial enterprise of philosophy. However, in the central period, these two distinctions served not only to identify the main beliefs of analytic phi- losophy, but, for those who accepted them and the research programme, they deﬁned the nature of philosophy itself. 1.1 Analytic versus synthetic The distinction between analytic and synthetic propositions was supposed to be the distinction between those propositions that are true or false as a matter of deﬁnition or of the meanings of the terms contained in them (the analytic propo- sitions) and those that are true or false as a matter of fact in the world and not solely in virtue of the meanings of the words (the synthetic propositions). Examples of analytic truths would be such propositions as ‘Triangles are three-sided plane ﬁgures’, ‘All bachelors are unmarried’, ‘Women are female’, ‘2 + 2 = 4’ and so on. In each of these, the truth of the proposition is entirely determined by its meaning; they are true by the deﬁnitions of the words that they contain. Such propositions can be known to be true or false a priori, and in each case they express necessary	Blackwell
truths. Indeed, it was a characteristic feature of the analytic philosophy of this central period that terms such as ‘analytic’, ‘necessary’, ‘a priori’ and ‘tautological’ were taken to be co-extensive. Contrasted with these were synthetic propositions, which, if they were true, were true as a matter of empirical fact and not as a matter of deﬁnition alone. Thus, propositions such as ‘There are more women than men in the United States’, ‘Bachelors tend to die earlier than married men’ and ‘Bodies attract each other according to the inverse square law’ are all said to be synthetic propositions, and, if they are true, they express a posteriori empirical truths about the real world that are independent of language. Such empirical truths, according to this view, are never necessary; rather, they are contingent. For philosophers holding these views, the terms ‘a posteriori’, ‘synthetic’, ‘contingent’ and ‘empirical’ were taken to be more or less co-extensive. It was a basic assumption behind the logical positivist movement that all meaning- ful propositions were either analytic or empirical, as deﬁned by the conceptions that I have just stated. The positivists wished to build a sharp boundary between meaningful propositions of science and everyday life on the one hand, and nonsensical propositions of metaphysics and theology on the other. They claimed that all meaningful proposi- tions are either analytic or synthetic: disciplines such as logic and mathematics fall within the analytic camp; the empirical sciences and much of common sense fall within the synthetic camp. Propositions that were neither analytic nor empirical propositions, and which were therefore in principle not veriﬁable, were said to be nonsensical or meaningless. The slogan of the positivists was called the veriﬁcation principle, and, in a simple form, it can be stated as follows: all meaningful propositions are either analytic or synthetic, and those which are synthetic are empirically veriﬁable. This slogan was sometimes shortened to an even simpler battle cry: the meaning of a proposition is just its method of veriﬁcation. 3 JOHN R. SEARLE 1.2 The distinction between evaluative utterances and descriptive utterances Another distinction, equally important in the positivist scheme of things, is the dis- tinction between those utterances that express propositions that can be literally either true or false and those utterances that are used not to express truths or falsehoods, but rather, to give vent to our feelings and emotions. An example of a descriptive statement would be, ‘The incidence of crimes of theft has increased in the past ten years’. An instance of the evaluative class would be ‘Theft is wrong’. The positivists claimed that many utterances that had the form of meaningful propositions were used not to state propositions that were veriﬁable either analytically or synthetically, but to express emo- tions and feelings. Propositions of ethics look as if they are cognitively meaningful, but they are not; they have only ‘emotive’ or ‘evaluative’ meaning. The propositions of science, mathematics, logic and much of common sense fall in the descriptive class; the utterances of aesthetics, ethics and much of religion fall in the evaluative class. It is important to note that on this conception evaluative propositions are not, strictly speak- ing, either true or false, since they are not veriﬁable as either analytic or empirical. The two distinctions are crucially related in that all of the statements that fall on one side or the other of the analytic–synthetic distinction also fall within the descriptive class of the descriptive–evaluative distinction. The importance that these two distinctions had for deﬁning both the character of the philosophical enterprise and the relationships between language and reality is hard	Blackwell
to exaggerate. One radical consequence of the distinction between descriptive and evaluative propositions was that certain traditional areas of philosophy, such as ethics, aesthetics and political philosophy, were virtually abolished as realms of cognitive meaningfulness. Propositions in these areas were, for the most part, regarded as non- sensical expressions of feelings and emotions, because they are not utterances that can be, strictly speaking, either true or false. Since the aim of philosophers is to state the truth, and since evaluative utterances cannot be either true or false, it cannot be one of the aims of philosophy to make any evaluative utterances. Philosophers might analyse the meaning of evaluative terms, and they might examine the logical rela- tionships among these terms, but philosophers, qua philosophers, can make no ﬁrst- order evaluations in aesthetics, ethics or politics, as these ﬁrst-order evaluations are not, strictly speaking, meaningful. They may have a sort of secondary, derivative meaning, called ‘emotive meaning’, but they lack scientiﬁcally acceptable cognitive meaning. If the task of philosophy is to state the truth and not to provide evaluations, what then is the subject matter of philosophy? Since the methods of philosophers are not those of empirical science – since their methods are a priori rather than a posteriori – it cannot be their aim to state empirical truths about the world. Such propositions are the propositions of the special sciences. The aim of philosophers, therefore, is to state analytic truths concerning logical relations among the concepts of our language. In this period of philosophy, the task of philosophy was taken to be the task of conceptual analysis. Indeed, for most philosophers who accepted this view, philosophy and con- ceptual analysis were the same. Where traditional philosophers had taken their task to be the discussion of the nature of the good, the true, the beautiful and the just, the 4 CONTEMPORARY PHILOSOPHY IN THE UNITED STATES positivist and post-positivist analytic philosophers took their task to be the analysis of the meaning of concepts such as ‘goodness’, ‘truth’, ‘beauty’ and ‘justice’. Ideally the analysis of these and other philosophically interesting concepts, such as ‘knowledge’, ‘certainty’ and ‘cause’, should give necessary and sufﬁcient conditions for the applica- tion of these concepts. They saw this as being the legitimate heir of the traditional phi- losophical enterprise, but an heir purged of the metaphysical nonsense and confusion that had discredited the traditional enterprise. If we combine the assumption that philosophy is essentially a conceptual, analytic enterprise with the assumption that its task is foundational – that is, its task is to provide secure foundations for such things as knowledge – then the consequence for the posi- tivists is that philosophical analysis tends in large part to be reductive. That is, the aim of the analysis is to show, for example, how empirical knowledge is based on, and ulti- mately reducible to, the data of our experience, to so-called sense data. (This view is called ‘phenomenalism’.) Similarly, statements about the mind are based on, and there- fore ultimately reducible to, statements about external behaviour (behaviourism). Nec- essary truth is similarly based on conventions of language as expressed in deﬁnitions (conventionalism); and mathematics is based on logic, especially set theory (logicism). In each case, the more philosophically puzzling phenomenon is shown to have a secure foundation in some less puzzling phenomenon, and indeed, the ideal of such analysis was to show that the puzzling phenomena could be entirely reduced to less puzzling phenomena. ‘Phenomenalism’ supposedly gave science a secure foundation because science could be shown to be founded on the data of our senses. Since the form of the	Blackwell
reduction was analytic or deﬁnitional, it had the consequence that statements about empirical reality could be translated into statements about sense data. Similarly, accord- ing to behaviourism, statements about mental phenomena could be translated into statements about behaviour. Within the camp of analytic philosophers who thought the aim of philosophy was conceptual analysis, there were two broad streams. One stream thought ordinary language was in general quite adequate, both as a tool and as a subject matter of philosophical analysis. The other stream thought of ordinary language as hopelessly inadequate for philosophical purposes, and irretrievably confused. The philosophers of this latter stream thought that we should use the tools of modern mathematical logic both for analysing traditional philosophical problems and, more importantly, for creat- ing a logically perfect language, for scientiﬁc and philosophical purposes, in which certain traditional confusions could not even arise. There was never a rigid distinction between these two streams, but there were certainly two broad trends: one which emphasized ordinary language philosophy and one which emphasized symbolic logic. Both streams, however, accepted the central view that the aim of philosophy was con- ceptual analysis, and that in consequence philosophy was fundamentally different from any other discipline; they thought that it was a second-order discipline analysing the logical structure of language in general, but not dealing with ﬁrst-order truths about the world. Philosophy was universal in subject matter precisely because it had no special subject matter other than the discourse of all other disciplines and the discourse of common sense. A further consequence of this conception was that philosophy became essentially a linguistic or conceptual enterprise. For that reason, the philosophy of language was 5 JOHN R. SEARLE absolutely central to the philosophical task. In a sense, the philosophy of language was not only ‘ﬁrst philosophy’; all of philosophy became a form of philosophy of language. Philosophy was simply the logical investigation of the structure of language as it was used in the various sciences and in common life. 2 The Rejection of These Two Distinctions and the Rejection of Foundationalism Work done in the 1950s and 1960s led to the overcoming of these two distinctions; and with the rejection of these two distinctions came a new conception of analytic phi- losophy – a conception that emerged in the 1970s and 1980s and which is still being developed. The rejection of these two distinctions and of the foundationalist research programme led to an enormous upheaval in the conception of the philosophical enter- prise and in the practice of analytic philosophers. The most obvious problem with tra- ditional analytic philosophy was that the reductionist enterprise failed. In every case, the attempts to provide reductionist analyses of the sort proposed by the phenomenal- ists and behaviourists were unsuccessful, and by 1960 the lack of success was obvious. A series of important theoretical developments also took place at this time, but for the sake of simplicity I shall concentrate on only ﬁve of these: Quine’s rejection of the analytic–synthetic distinction, Austin’s theory of speech acts, Wittgenstein’s criticism of foundationalism, Rawls’s work in political philosophy and the changes in the philosophy of science due to Kuhn and others. 2.1 Quine’s attack on the analytic–synthetic distinction Perhaps the most important criticism of the analytic–synthetic distinction was made by W. V. O. Quine in a famous article entitled ‘Two dogmas of empiricism’ (Quine 1953). In this article, Quine claimed that no adequate, non-circular deﬁnition of analyticity had ever been given. Any attempt to deﬁne analyticity had always been made using notions that were in the same family as analyticity, such as synonymy and deﬁnition,	Blackwell
and consequently, the attempts to deﬁne analyticity were invariably circular. However, an even more important objection that emerged in Quine’s article was this: the notion of an analytic proposition is supposed to be a notion of a proposition that is immune to revision, that is irrefutable. Quine claimed that there were no propositions that were immune to revision, that any proposition could be revised in the face of recalcitrant evidence, and that any proposition could be held in the face of recalcitrant evidence, provided that one was willing to make adjustments in other propositions originally held to be true. Quine argued that we should think of the language of science as being like a complex network that was impinged upon by empirical veriﬁcation only at the edges. Recalcitrant experiences at the edges of science can produce changes anywhere along the line, but the changes are not forced on us by purely logical considerations; rather, we make various pragmatic or practical adjustments in the network of our sentences or beliefs to accommodate the ongoing character of our experiences. Language, on this view, is not atomistic. It does not consist of a set of propositions, each of which can be assessed in isolation. Rather, it consists of a holistic network, and, in this network, 6 CONTEMPORARY PHILOSOPHY IN THE UNITED STATES propositions as groups confront experience; propositions individually are not simply assessed as true or false. (This holism of scientiﬁc discourse was inﬂuenced by the French philosopher of science, Duhem, and the view is frequently referred to as ‘the Duhem–Quine thesis’.) Most philosophers today accept some version or other of Quine’s rejection of the analytic–synthetic distinction. Not everybody agrees with his actual argument (I, for one, do not), but now there is general scepticism about our ability to make a strict dis- tinction between those propositions that are true by deﬁnition and those that are true as a matter of fact. The rejection of the analytic–synthetic distinction has profound consequences for analytic philosophy, as we shall see in more detail later. At this point it is important to state that if there is no well-deﬁned class of analytic propositions, then the philosopher’s propositions cannot themselves be clearly identi- ﬁed as analytic. The results of philosophical analysis cannot be sharply distinguished from the results of scientiﬁc investigation. On the positivist picture, philosophy was not one among other sciences; rather, it stood outside the frame of scientiﬁc discourse and analysed the logical relations between, on the one hand, that discourse and its vocabu- lary and, on the other, experience and reality. Philosophers, so to speak, analysed the relation between language and reality, but only from the side. If we accept Quine’s rejec- tion of the analytic–synthetic distinction, then philosophy is not something that can be clearly demarcated from the special sciences. It is, rather, adjacent to, and overlaps with, other disciplines. Although philosophy is more general than other disciplines, its propositions do not have any special logical status or special logical priority with regard to the other disciplines. 2.2 Austin’s theory of speech acts The British philosopher J. L. Austin was suspicious of both the distinction between ana- lytic and synthetic propositions, and the distinction between evaluative and descriptive utterances. During the 1950s he developed an alternative conception of language (Austin 1962). His ﬁrst observation was that there is a class of utterances that are obvi- ously perfectly meaningful, but which do not even set out to be either true or false. A man who says, for example, ‘I promise to come and see you’ or a qualiﬁed authority who says to a couple, ‘I pronounce you man and wife’ is neither reporting on nor describing a promise or a marriage respectively. Such utterances should be thought	Blackwell
not as cases of describing or stating, but rather as doing, as acting. Austin baptized these utterances ‘performatives’ and contrasted them with ‘constatives’. The distinction between constatives and performatives was supposed to contain three features: con- statives, but not performatives, could be true or false; performatives, on the other hand, though they could not be true or false, could be felicitous or infelicitous, depending on whether or not they were correctly, completely and successfully performed; and ﬁnally, performatives were supposed to be actions, doings or performances, as opposed to mere sayings or statings. But, as Austin himself saw, the distinctions so drawn did not work. Many so-called performatives turned out to be capable of being true or false; for example, warnings could be either true or false. And statements, as well as performa- tives, could be infelicitous. For example, if one made a statement for which one had insufﬁcient evidence, one would have made an infelicitous statement. And ﬁnally, 7 JOHN R. SEARLE stating is as much performing an action as promising or ordering or apologizing. The abandonment of the performative–constative distinction led Austin to a general theory of speech acts. Communicative utterances in general are actions of a type he called ‘illocutionary acts’. One great merit of Austin’s theory of speech acts is that it enabled subsequent philosophers to construe the philosophy of language as a branch of the philosophy of action. Since speech acts are as much actions as any other actions, the philosophical analysis of language is part of the general analysis of human behaviour. And since intentional human behaviour is an expression of mental phenomena, it turns out that the philosophy of language and the philosophy of action are really just different aspects of one larger area, namely, the philosophy of mind. On this view, the philosophy of lan- guage is not ‘ﬁrst philosophy’; it is a branch of the philosophy of mind. Though Austin did not live to carry out the research programme implicit in his initial discoveries, subsequent work, including my own, has carried this research further. By treating speaking as a species of intentional action we can give a new sense to a lot of old questions. For example, the old question, ‘How many kinds of utterances are there?’ is too vague to be answered. But if we ask ‘How many kinds of illocutionary acts are there?’, we can give a precise answer, since the question asks, ‘How many pos- sible ways are there for speakers to relate propositional contents to reality in the per- formance of actions that express illocutionary intentions?’ An analysis of the structure of those intentions reveals ﬁve basic types of illocutionary act: we tell people how things are (Assertives), we try to get them to do things (Directives), we commit ourselves to doing things (Commissives), we express our feelings and attitudes (Expressives) and we bring about changes in the world through our utterances, so that the world is changed to match the propositional content of the utterance (Declarations). (For details see Searle 1979 and 1983.) 2.3 Wittgenstein’s rejection of foundationalism The single most inﬂuential analytic philosopher of the twentieth century, and indeed, the philosopher whom most analytic philosophers would regard as the greatest philoso- pher of the century, is Ludwig Wittgenstein. Wittgenstein published only one short book during his lifetime, which represents his early work, but with the posthumous publication of his Philosophical Investigations in 1953, a series of his later writings began to become available. Now, we have a sizeable corpus of the work he did in the last twenty years of his life. Through painstaking analy- sis of the use of language, particularly through analysis of psychological concepts, Wittgenstein attempted to undermine the idea that philosophy is a foundational enter-	Blackwell
prise. He asserted, on the contrary, that philosophy is a purely descriptive enterprise, that the task of philosophy is neither to reform language nor to try to place the various uses of language on a secure foundation. Rather, philosophical problems are removed by having a correct understanding of how language actually functions. A key notion in Wittgenstein’s conception of language is the notion of a language game. We should think of the words in language as being like the pieces in a game. They are not to be understood by looking for some associated idea in the mind, or by follow- ing some procedure of veriﬁcation, or even by looking at the object for which they stand. 8 CONTEMPORARY PHILOSOPHY IN THE UNITED STATES Rather, we should think of words in terms of their use, and referring to objects in the world is only one of many uses that words have. The meaning of a word is given by its use, and the family of uses that a group of words has constitutes a language game. Examples include the language game we play in describing our own sensations, or the language game we play in identifying the causes of events. This conception of language leads Wittgenstein to the rejection of the conception that the task of philosophical analysis is reductionist or foundationalist. That is, Wittgenstein rejects the idea that lan- guage games either have or need a foundation in something else, and he rejects the idea that certain language games can be reduced to certain other kinds of language games. The effect, Wittgenstein says, of philosophical analysis is not to alter our existing lin- guistic practices or to challenge their validity; it is simply to describe them. Language neither has nor needs a foundation in the traditional sense. I said that Wittgenstein was the single most inﬂuential philosopher in the analytic tradition, but there is a sense in which it seems to me he has still not been properly understood, nor had his lessons been fully assimilated by analytic philosophers. I will have more to say about his inﬂuence later. 2.4 Rawls’s theory of justice The conception of moral philosophy in the positivist and post-positivist phases of ana- lytic philosophy was extremely narrow. Strictly speaking, according to the positivists, moral utterances could not be either true or false, so there was nothing that the philoso- pher could say, qua philosopher, by way of making moral judgements. The task for the moral philosopher was to analyse moral discourse, to analyse the meaning and use of moral terms such as ‘good’, ‘ought’, ‘right’, ‘obligation’, etc. It is important to see that this conception of moral philosophy was a strict logical consequence of the acceptance of the distinction between evaluative and descriptive utterances. For if evaluative utter- ances cannot be either true or false, and if ﬁrst-order moral discourse consists in evalu- ative utterances, and if the task of the philosopher is to state the truth, it follows that the philosopher, qua philosopher, cannot make any ﬁrst-order moral judgements. As a philosopher, all he or she can do is the second-order task of analysing moral concepts. Some philosophers of the positivist and post-positivist periods rejected this narrow conception of moral philosophy, and there were a series of attacks mounted on the distinction between evaluative and descriptive utterances, including some attacks by myself in the mid-1960s (Searle 1964). It remained, however, for John Rawls to reopen the traditional conception of political and moral philosophy with the publication of his book A Theory of Justice in 1971. For the purposes of the present discussion, the im- portant thing about Rawls’s work was not that he refuted the traditional dichotomy of descriptive and evaluative utterances, but that he simply ignored it and proceeded to develop a theory of political institutions of a sort that has a long philosophical tradi- tion and which the positivists thought they had overcome. Rawls, in effect, revived the	Blackwell
social contract theory, which had long been assumed to be completely defunct; but he did it by an ingenious device: he did not attempt, as some traditional theorists had done, to show that there might have been an original social contract, nor did he try to show that the participation of individuals in society involved a tacit contract. Rather, he used the following thought experiment as an analytic tool: think of the sort of society that 9 JOHN R. SEARLE rational beings would agree to if they did not know what sort of position they them- selves would occupy in that society. If we imagine rational beings, hidden behind a veil of ignorance, who are asked to select and agree on forms of social institutions that would be fair for all, then we can develop criteria for appraising social institutions on purely rational grounds. The importance of Rawls for our present discussion is not whether he succeeded in developing new foundations for political theory, but the fact that his work gave rise to a renewed interest in political philosophy, which was soon accompanied by a renewed interest in the traditional questions of moral philosophy. Moral and political philoso- phy had been conﬁned to a very small realm by the positivist philosophers, and for that reason seemed sterile and uninteresting. Very little work was done in that area, but since the 1970s it has grown enormously, and is now a ﬂourishing branch of analytic philosophy. 2.5 Post-positivist philosophy of science Throughout the positivist period the model of empirical knowledge was provided by the physical sciences, and the general conception was that the empirical sciences proceeded by the gradual but cumulative growth of empirical knowledge through the systematic application of scientiﬁc method. There were different versions of scientiﬁc method, according to the philosophers of that period, but they all shared the idea that scientiﬁc, empirical propositions are essentially ‘testable’. Initially a proposition was thought testable if it could be conﬁrmed, but the most inﬂuential version of this idea is Popper’s claim that empirical propositions are testable if they are falsiﬁable in principle. That is, in order for a proposition to tell us how the world is as opposed to how it might be or might have been, there must be some conceivable state of affairs that would render that proposition false. Propositions of science are, strictly speaking, never veriﬁable – they simply survive repeated attempts at falsiﬁcation. Science is in this sense fallible, but it is at the same time rational and cumulative. This picture of the history of science was very dramatically challenged in Thomas (1962). According to Kuhn, the Kuhn’s book The Structure of Scientiﬁc Revolutions history of science shows not a gradual and steady accumulation of knowledge but periodic revolutionary overthrows of previous conceptions of reality. The shift from Aristotelian physics to Newtonian physics, and the shift from Newtonian physics to relativistic physics are both illustrations of how one ‘paradigm’ is replaced by another. When the burden of puzzling cases within one paradigm becomes unbearable, a new paradigm emerges, which provides not just a new set of truths but a whole new way of looking at the subject matter. ‘Normal sciences’ always proceed by puzzle-solving within a paradigm, but revolutionary breakthroughs, rather than puzzle-solving within a paradigm, are matters of overthrowing one paradigm and replacing it with another. Just as Kuhn challenged the picture of science as essentially a matter of a steady accumulation of knowledge, so Paul Feyerabend challenged the conception of there being a unitary rational ‘scientiﬁc method’ (Feyerabend 1975). Feyerabend tried to show that the history of science reveals not a single rational method but rather a series of opportunistic, chaotic, desperate (and sometimes even dishonest) attempts to cope 10	Blackwell
CONTEMPORARY PHILOSOPHY IN THE UNITED STATES with immediate problems. The lesson that Feyerabend draws from this is that we should abandon the constraining idea of there being such a thing as a single, rational method that applies everywhere in science; rather, we should adopt an ‘anarchistic’ view, according to which ‘anything goes’. Reactions to Kuhn and Feyerabend, not surpris- ingly, differ enormously among analytic philosophers. Kuhn sometimes seems to be arguing that there is not any such thing as the real world existing independently of our scientiﬁc theories, which it is the aim of our scientiﬁc theories to represent. Kuhn, in short, seems to be denying realism. Most philosophers do not take this denial of realism at all seriously. Even if Kuhn were right about the structure of scientiﬁc revolutions, this in no way shows that there is no independent reality that science is investigating. Again, most philosophers would accept Feyerabend’s recognition of a variety of methods used in the history of science, but very few people take seriously the idea that there are no rational constraints on investigation whatever. Nonetheless, the effect of these authors has been important in at least the following respect. The positivists’ con- ception of science as a steady accumulation of factual knowledge, and of the task of the philosopher as the conceptual analysis of scientiﬁc method, has given way to an attitude to science that is at once more sceptical and more activist. It is more sceptical in the sense that few philosophers are looking for the one single method that pervades every enterprise called ‘science’, but it is more activist in the sense that philosophy of science interacts more directly with scientiﬁc results. For example, recent philosophi- cal discussions about quantum mechanics, or about the signiﬁcance of Bell’s theorem within quantum mechanics, reveal that it is now impossible to say exactly where the problem in physics ends and the problem in philosophy begins. There is a steady inter- action and collaboration between philosophy and science on such philosophically puzzling questions. 3 Some Recent Developments The results of the changes that I have just outlined are to make analytic philosophy on the one hand a more interesting discipline, but on the other hand a much less well- deﬁned research project. In the way that the veriﬁcation principle formed the core ideology of the logical positivists and in the way that the conceptual analysis formed the core research project of the post-positivistic analytic philosopher, there is now no ideological point of reference that is commonly agreed upon; nor is there a universally accepted research programme. For example, conceptual analysis thirty years ago was taken to be the heart of analytic philosophy, but now many philosophers would deny that it is the central element in the philosophical enterprise. Some philosophers, indeed, would say that the traditional enterprise of attempting to ﬁnd logically necessary and sufﬁcient conditions for the applicability of a concept is misconceived in principle. They think the possibility of such an enterprise has been refuted by Quine’s refutation of the analytic–synthetic distinction, as well as Wittgenstein’s observation that many philo- sophically puzzling concepts have not a central core or essence of meaning, but a variety of different uses united only by a ‘family resemblance’. Many other philosophers would say that conceptual analysis is still an essential part of the philosophical enter- prise, as indeed it has been since the time of Plato’s dialogues, but it is no longer seen 11 JOHN R. SEARLE to be the whole of the enterprise. Philosophy is now, I believe, a much more interesting subject than it was a generation ago because it is no longer seen as something separate from, and sealed off from, other disciplines. In particular, philosophy is now seen by	Blackwell
most analytic philosophers as being adjacent to and overlapping with the sciences. My own view, which I feel is fairly widely shared, is that words like ‘philosophy’ and ‘science’ are in many respects misleading, if they are taken to imply the existence of mutually exclusive forms of knowledge. Rather, it seems to me that there is just knowl- edge and truth, and that in intellectual enterprises we are primarily aiming at knowl- edge and truth. These may come in a variety of forms, whether in history, mathematics, physics, psychology, literary criticism or philosophy. Philosophy tends to be more general than other subjects, more synoptic in its vision, more conceptually or logically oriented than other disciplines, but it is not a discipline that is hermetically sealed off from other subjects. The result is that many areas of investigation which were largely ignored by analytic philosophers a generation ago have now become thriving branches of philosophy, including cognitive science, the philosophy of biology and the philoso- phy of economics. In what follows, I will conﬁne my discussion to ﬁve major areas of philosophical research: cognitive science, the causal theory of reference, intentionalis- tic theories of meaning, truth-conditional theories of meaning, and Wittgenstein’s conception of language and mind and his response to scepticism. 3.1 Philosophy and cognitive science Nowhere is the new period of collaboration between philosophy and other disciplines more evident than in the new subject of cognitive science. Cognitive science from its very beginnings has been ‘interdisciplinary’ in character, and is in effect the joint prop- erty of psychology, linguistics, philosophy, computer science and anthropology. There is, therefore, a great variety of different research projects within cognitive science, but the central area of cognitive science, its hardcore ideology, rests on the assumption that the mind is best viewed as analogous to a digital computer. The basic idea behind cog- nitive science is that recent developments in computer science and artiﬁcial intelligence have enormous importance for our conception of human beings. The basic inspiration for cognitive science went something like this: human beings do information process- ing. Computers are designed precisely to do information processing. Therefore one way to study human cognition – perhaps the best way to study it – is to study it as a matter of computational information processing. Some cognitive scientists think that the com- puter is just a metaphor for the human mind; others think that the human mind is literally a computer program. But it is fair to say that without the computational model there would not have been a cognitive science as we now understand it. This conception of human cognition was ideally suited to the twentieth-century analytic tradition in philosophy of mind because of the analytic tradition’s resolute materialism. It was anti-mentalistic and anti-dualistic. The failure of logical behav- iourism led not to a revival of dualism but to more sophisticated versions of material- ism. I will now brieﬂy summarize some of the recent developments in materialistic philosophies of mind that led to the computational theory of the mind. The logical behaviourists’ thesis was subject to many objections, the most important being the objection that it ignores internal mental phenomena. In science and common 12 CONTEMPORARY PHILOSOPHY IN THE UNITED STATES sense it seems more natural to think of human behaviour as being caused by internal mental states rather than to think of the mental states as simply consisting of the behaviour. This weakness in behaviourism was corrected by the materialist identity thesis, sometimes called ‘physicalism’. According to the physicalist identity theory, mental states are identical with states of the brain. We do not know in detail what these	Blackwell
identities are, but the progress of the neurosciences makes it seem overwhelmingly probable that every mental state will be discovered to be identical with some brain state. In the early version of the identity thesis it was supposed that every type of mental state would be discovered to be identical with some type of physical state, but after some debate this began to seem more and more implausible. There is no reason to suppose that only systems with neurons like ours can have mental states; indeed, there is no reason to suppose that two human beings who have the same belief must therefore be in the same neurophysiological state. So, ‘type–type identity theory’ naturally gave way to ‘token–token identity theory’. The token identity theorists claimed that every par- ticular mental state is identical with some particular neurophysiological state, even if there is no type correlation between types of mental states and types of physical states. But that only leaves open the question, ‘What is it that two different neurophysiologi- cal states have in common if they are both the same mental state?’ To many analytic philosophers it seemed obvious that the answer to our question must be that two neurophysiological states are the same type of mental state if they serve the same function in the overall ecology of the organism. Mental states on this view can be deﬁned in terms of their causal relations to input stimuli, to other mental states, and to external behaviour. This view is called ‘functionalism’ and it is a natural development from token–token identity theory. However, the functionalist has to answer a further obvious question: ‘What is it about the states that gives them the causal relations that they do have?’ If mental states are deﬁned in terms of their causal relations, then what is it about the structure of dif- ferent neurophysiological conﬁgurations that can give them the same causal relations? It is at precisely this point that the tradition of materialism in analytic philosophy con- verges with the tradition of artiﬁcial intelligence. The computer provides an obvious answer to the question that I have just posed. The distinction between the software and the hardware, the program and the physical system that implements the program, pro- vides a model for how functionally equivalent elements at a higher level can be realized in or implemented by different physical systems at a lower level. Just as one and the same program can be implemented by quite different physical hardware systems, so one and the same set of mental processes can be implemented in different neurophysiologi- cal or other forms of hardware implementations. Indeed, on the most extreme version of this view, the mind is to the brain as the program is to the hardware. This sort of functionalism came to be called ‘computer functionalism’ or ‘Turing machine func- tionalism’, and it coincides with the strong version of ‘artiﬁcial intelligence’ (Strong AI), the version that says having a mind just is having a certain sort of program. I have refuted Strong AI in a series of articles (Searle 1980a, 1980b). The basic idea of that refutation can be stated quite simply. Minds cannot be equivalent to programs because programs are deﬁned purely formally or syntactically and minds have mental contents. The easiest way to see the force of the refutation is to see that a system, say oneself, could learn to manipulate the formal symbols for understanding a natural 13 JOHN R. SEARLE language without actually understanding that language. I might have a program that enables me to answer questions in Chinese simply by matching incoming symbols with the appropriate processing and output symbols, but nonetheless I still would not thereby understand Chinese. However, though the project of computer functionalism is almost certainly a failure, the results of the enterprise are in many respects quite	Blackwell
useful. Important things can be learned about the mind by pursuing the computer metaphor, and the research effort has not necessarily been wasted. The most exciting recent development has been to think of mental processes not on the model of the con- ventional serial digital computer, but rather to think of brain processes on the model of parallel distributed processing computers. The most exciting recent development, in my view, in cognitive science has been the development of such ‘neural net models’ for human cognition. In concluding this section, I want to point out that in my view the chief weakness of analytical philosophy of mind, a weakness it shares with the past 300 years in the philosophy of mind, has been its assumption that there is somehow an inconsistency between mentalism and materialism. Analytic philosophers, along with the rest of the Cartesian tradition, have characteristically assumed that ‘mental’ implies ‘non- material’ or ‘immaterial’ and that ‘material’ or ‘physical’ implies ‘non-mental’. But if one reﬂects on how the brain works, it seems that both of these assumptions are obviously false. What that shows is that our whole vocabulary, our whole terminology of the mental and physical, needs wholesale revision. 3.2 The causal theory of reference A central question in analytic philosophy of language, since Frege (and indeed in phi- losophy since the time of Plato), has been: How does language relate to the world? How do words hook on to things? In answering this question, the analytic tradition had char- acteristically found a connection between the notion of reference and the notion of truth. An expression, such as a proper name, refers to or stands for or designates an object because associated with that name is some descriptive content, some concept of the object in question, and the object in question satisﬁes or ﬁts that descriptive content. The expression refers to the object only because the description is true of the object. This is the standard reading of Frege’s famous distinction between sense and reference, between Sinn and Bedeutung. Expressions refer to objects in virtue of their sense and the sense provides a description, a ‘mode of presentation’, of the object in question. Something analogously applies with general terms: general terms are true of an object because each general term has associated with it a cluster of features, and the term will be true of the object if the object in question has those features. In the 1970s this conception of the relation between language and reality was attacked by a number of philosophers, most prominently Donnellan (1970), Kripke (1972) and Putnam (1975). A variety of arguments were mounted against the tradi- tional conception of meaning and reference, but the common thread running through these arguments was that the descriptive content associated with a word provided neither necessary nor sufﬁcient conditions for its application. A speaker might refer to an object even though the associated description that he or she had was not true of that object; a speaker might have a description that was satisﬁed by an object even though 14 CONTEMPORARY PHILOSOPHY IN THE UNITED STATES that was not the object to which he or she was referring. The most famous version of this argument was Putnam’s ‘twin earth’ example. Imagine a planet in a distant galaxy exactly like ours in every respect except that on this planet what they call ‘water’ has a different chemical composition. It is not composed of H2O but has an extremely com- plicated formula that we will abbreviate as ‘XYZ’. Prior to 1750, prior to the time that anyone knew the chemical composition of water, the people on twin earth had in their minds exactly the same concept of water as the people on earth. Nonetheless our word ‘water’ does not refer to the stuff on twin earth. Our word ‘water’, whether or not we knew it in 1750, refers to H2O; and this is a matter of objective causal relations in the	Blackwell
world which are independent of the ideas that people have in their heads. Meanings on this view are not concepts in people’s heads, but objective relations in the world. Well, if associated ideas are not sufﬁcient for meaning, what is? The answer given by the three authors I have mentioned is that there must be some sort of causal connection between the use of the word and the object or type of entity in the world that it applies to. Thus, if I use the word ‘Socrates’, it refers to a certain Greek philosopher only because there is a causal chain connecting that philosopher and my current use of the word. The word ‘water’ is not deﬁned by any checklist of features; rather, ‘water’ refers to what- ever stuff in the world was causally related to certain original uses of the word ‘water’, and these uses subsequently came to be accepted in the community and were then passed down through a causal chain of communication. There is a very natural way of connecting the computer functionalist conception of the mind with the causal theory of reference. If the mind were a computer program, and if meaning were a matter of causal connections to the world, then the way the mind acquires meanings is for the system that implements the computer program to be involved in causal interactions with the world. 3.3 Intentionalistic theories of meaning Much of the best work in speech act theory done after the publication of Austin’s How to Do Things with Words in 1962, and my Speech Acts in 1969, attempted to combine the insights of Paul Grice’s account of meaning with the framework provided by the theory of speech acts. In a series of articles beginning in the late 1950s (Grice 1957, 1968), Grice had argued that there is a close connection between the speaker’s inten- tions in the performance of an utterance and the meaning of that utterance. In his original formulation of this view, Grice analysed the speaker’s meaning in terms of the intention to produce an effect on the hearer by means of getting the hearer to recog- nize the intention to produce that very effect. Thus, for example, according to Grice, if a speaker intends to tell a hearer that it is raining, then in the speaker’s utterance of the sentence, ‘It is raining’, the speaker’s meaning will consist of his or her intention to produce in the hearer the belief that it is raining by means of getting the hearer to recognize his or her intention to produce that very belief. Subsequent work by Grice altered the details of this account, but the general principle remained the same: meaning is a matter of a self-referential intention to produce an effect on a hearer by getting the hearer to recognize the intention to produce that effect. Grice combined this analysis of meaning with an analysis of certain principles of conversational co- operation. In conversation, people accept certain tacit principles, which Grice calls 15 JOHN R. SEARLE ‘Maxims of Conversation’ – they accept the principles that the speaker’s remarks will be truthful and sincere (the maxim of quality), that they will be relevant to the conversational purposes at hand (the maxim of relation), that the speaker will be clear (the maxim of manner) and that the speaker will say neither more nor less than is necessary for the purposes of the conversation (the maxim of quantity). There has been a great deal of controversy about the details of Grice’s analysis of meaning, but the basic idea that there is a close connection between meaning and intention has been accepted and has proved immensely useful in analysing the struc- ture of certain typical speech act phenomena. My own view is that Grice confuses that part of meaning which has to do with representing certain states of affairs and certain illocutionary modes, and that part of meaning that has to do with communicating those representations to a hearer. Grice, in short, confuses communication with	Blackwell
representation. However, the combination of an intentionalistic account of meaning, together with rational principles of co-operation, is immensely fruitful in analysing such problems as those of ‘indirect speech acts’ and ﬁgurative uses of language such as metaphors. So, for example, in an indirect speech act, a speaker will characteristi- cally mean something more than what he or she actually says. To take a simple example, in a dinner table situation a speaker who says ‘Can you pass the salt?’ would usually not just be asking a question about the salt-passing abilities of the hearer; he or she would be requesting the hearer to pass the salt. Now the puzzle is this: how is it that speakers and hearers communicate so effortlessly when there is a big gulf between what the speaker means and what he or she actually says? In the case of metaphor, a similar question arises: how does the speaker communicate so effortlessly his or her metaphorical meaning when the literal meaning of the sentence uttered does not encode that metaphorical meaning? A great deal of progress has been made on these and other problems using the apparatus that Grice contributed to the theory of speech acts. One of the marks of progress in philosophy is that the results of philosophical analy- sis tend to be appropriated by other disciplines, and this has certainly happened with speech act theory. Speech act theory is now a thriving branch of the discipline of lin- guistics, and the works of Austin and Grice, as well as my own, are as well known among linguists as they are among philosophers. 3.4 Truth-conditional theories of meaning Philosophers such as Quine and his former student, Donald Davidson, have always felt that intentionalistic theories of meaning of the sort proposed by Grice and Searle were philosophically inadequate, because the intentionalistic notions seemed as puzzling as the notion of meaning itself and because they could necessarily involve linguistic meaning in their ultimate analyses. So Quine and Davidson attempted to give accounts of meaning that did not employ the usual apparatus of intentionality. The most inﬂu- ential version of this attempt is Davidson’s project of analysing meaning in terms of truth conditions. The basic idea is that one knows the meaning of a sentence if one knows under what conditions it is true or false. Thus, one knows the meaning of the German sentence ‘Schnee ist weiss’ if one knows that it is true if and only if snow is white. Now since a theory of meaning for a language should be able to state the 16 CONTEMPORARY PHILOSOPHY IN THE UNITED STATES meaning of each of the sentences of the language, and since the meanings of the sen- tences of the language are given by truth conditions, and since truth conditions can be speciﬁed independently of the intentionalistic apparatus, it seems to Davidson that a theory of truth (that is, a theory of the truth conditions of the sentences) of a language would provide a theory of meaning for that language. In order to carry out the project of explaining meaning in terms of truth, Davidson employs the apparatus of Tarski’s semantic deﬁnition of truth, a deﬁnition that Tarski had worked out in the 1930s. Tarski points out that it is a condition of adequacy on any account of truth that for any sentence s and any language L, the account must have the consequence that s is true in L if and only if p, where for s can be substituted the structural description of any sentence whatever, for L, the name of the language of which s is a part, and for p, the sentence itself or a trans- lation of it. Thus, for example, in English, the sentence ‘Snow is white’ is true if and only if snow is white. This condition is usually called ‘convention T’ and the corre- sponding sentences are called ‘T-sentences’. Now Davidson notes that convention T employs the fact that the sentence named by s has the same meaning as the sentence expressed by p, and thus Tarski is using the	Blackwell
notion of meaning in order to deﬁne the notion of truth. Davidson proposes to turn this procedure around by taking the notion of truth for granted, by taking it as a primitive, and using it to explain meaning. Here is how it works. Davidson hopes to get a theory of meaning for a speaker of a language that would be sufﬁcient to interpret any of the speaker’s utterances by getting a theory that would provide a set of axioms which would entail all true T-sentences for that speaker’s language. Thus, if the speaker speaks German, and we use English as a meta-language in which to state the theory of the speaker’s language, Davidson claims we would have an adequate theory of the speaker’s language if we could get a set of axioms which would entail a true T-sentence stated in English for any sentence that the speaker uttered in German. Thus, for example, our theory of meaning should contain axioms which entail that the speaker’s utterance ‘Schnee ist weiss’ is true in the speaker’s language if and only if snow is white. Davidson further claims that we could make this into an empirical theory of the speaker’s language by proceeding to associ- ate the speaker’s utterances with the circumstances in which we had empirical evidence for supposing that the speaker held those utterances to be true. Thus, if we hear the speaker utter the sentence ‘Es regnet’, we might look around and note that it was raining in the vicinity, and we might then form the hypothesis that the speaker holds true the sentence ‘Es regnet’ when it is raining in his or her immediate vicinity. This would provide the sort of empirical data on which we would begin to construct a theory of truth for the speaker’s language. It is important to note that we are to think of this as a thought experiment and not as an actual procedure that we have to employ when we try to learn German, for example. The idea is to cash out the notion of meaning in terms of truth conditions, and then cash out the notion of truth conditions in terms of a truth theory for a lan- guage, which is a theory that would entail all the true T-sentences of the language. The 17 JOHN R. SEARLE empirical basis on which the whole system rests is that of the evidence we could get concerning the conditions under which a speaker holds a sentence to be true. If the project could in principle be carried out, then we would have given an account of meaning which employed only one intentionalistic notion, the notion of ‘holding true’ a sentence. Over the past twenty years there has been quite an extensive literature on the nature of this project and how it might be applied to several difﬁcult and puzzling sorts of sentences – for example, indexical sentences, sentences about mental states or modal sentences. Enthusiasm for this project seems to have waned somewhat in recent years. In my view, the central weakness of Davidson’s enterprise is as follows: any theory of meaning must explain not only what a speaker represents by his or her utterances, but also how he or she represents them, under what mental aspects the speaker repre- sents truth conditions. For this reason, a theory of meaning cannot just correlate a speaker’s utterance with states of affairs in the world; it must explain what is going on in the speaker’s head which enables the speaker to represent those states of affairs under certain aspects with the utterances that the speaker makes. Thus, for example, suppose that snow is composed of H2O molecules in crystalline form, and suppose the colour white consists of light wave emissions of all wavelengths, then the sentence ‘Schnee ist weiss’ is true if and only if H2O molecules in crystalline form emit light of all wavelengths. Now this second T-sentence is just as empirically substantiated as the earlier example, ‘Schnee ist weiss’ is true if and only if snow is white. Indeed, it is a matter of scientiﬁc necessity that the state of affairs described by the former is identi-	Blackwell
cal with the state of affairs described by the latter. But the former example simply does not give the speaker’s meaning. The speaker might hold true the sentence ‘Schnee ist weiss’ under these and only these conditions and not know the slightest thing about H2O molecules and wavelengths of light. The T-sentence gives the truth conditions, but the speciﬁcation of the truth conditions does not necessarily give the meaning of the sentence, because the speciﬁcation does not yet tell us how the speaker represents those truth conditions. Does he or she represent them under the aspect of snow being white, or what is the same fact in the world, does he or she represent them under the aspect of frozen H2O crystals emitting light of all wavelengths? Any theory that cannot give that information is not a theory of meaning. There are various attempts to meet these sorts of objections, but, in my view, they are not successful. In the end, all truth deﬁnitional accounts of meaning, like the be- haviourist accounts which preceded them, end up with a certain ‘indeterminacy’ of meaning. They cannot account in objective terms for all of the subjective details of meaning, and both Davidson and Quine have acknowledged that their views result in indeterminacy. 3.5 Wittgenstein’s legacy Wittgenstein’s work covers such a vast range of topics, from aesthetics to math- ematics, and covers these topics with so much depth and insight, that it continues to be a source of ideas and inspiration for analytic philosophers and is likely to continue to be so for many years to come. I will mention only three areas. 18 CONTEMPORARY PHILOSOPHY IN THE UNITED STATES 3.5.1 Philosophical psychology One of Wittgenstein’s main areas of research was that of psychological concepts such as belief, hope, fear, desire, want, expect and sensation concepts such as pain and seeing. Perhaps his single most controversial claim in this area is that concerning a private language. He claims that it would be logically impossible for there to be a lan- guage that was private in the sense that its words could only be understood by the speaker because they referred to the speaker’s private inner sensations and had no external deﬁnition. Such a language would be absurd, he said, because for the applica- tion of such words there would be no distinction between what seemed right to the speaker and what really was right. But unless we can make a distinction between what seems right and what really is right, we cannot speak of right or wrong at all, and hence we cannot speak of using a language at all. ‘An inner process’, says Wittgenstein (1953), ‘stands in need of outward criteria’. Wittgenstein is here attacking the entire Cartesian tradition, according to which there is a realm of inner private objects, our inner mental phenomena, and the meanings of the words that stand for these entities are entirely deﬁned by private ostensive deﬁnitions. No other single claim of Wittgenstein’s has aroused as much controversy as the ‘private language argument’. It continues to be a source of fascination to contemporary philosophers, and many volumes have been written about Wittgenstein’s analysis of psychological concepts. 3.5.2 Following a rule Wittgenstein is part of a long tradition that emphasizes the distinction between the modes of explanation of the natural sciences and the modes of explanation of human behaviour and human cultural and psychological phenomena generally. His analysis of this problem chieﬂy deals with the phenomenon of human behaviour which is inﬂu- enced or determined by mental contents, and, most importantly, by the phenomena of human beings following a rule. What is it for a human being to follow a rule? Wittgenstein’s analysis of this stresses the difference between the way that rules guide human behaviour and the way that natural phenomena are results of causes.	Blackwell
Wittgenstein throughout emphasizes the difference between causes and reasons, and he also emphasizes the roles of interpretation and rule following. On the most extreme interpretation of Wittgenstein’s remarks about following a rule, he is the proponent of a certain type of scepticism. According to one view of Wittgenstein, he is arguing that rules do not determine their own application, that anything can be interpreted to accord with a rule, and consequently that anything can be interpreted to conﬂict with a rule. If taken to its extreme, this argument would have the consequence that, logically speaking, rules do not constrain human behaviour at all. And if that is right, then mental contents, such as knowledge of meanings of words or principles of action or even beliefs and desires, do not constrain human behaviour, because they are everywhere subject to an indeﬁnite range of different interpretations. Wittgenstein’s solution to this scepticism is to propose that interpretation comes to an end when we simply accept the cultural practices of the community in which we are imbedded. Interpretation comes to an end, and we just act on a rule. Acting on a rule is a practice, and it is one that we are brought up to perform in our culture. The sceptical implications of Wittgenstein’s account of rule following are resolved by an appeal to a naturalistic solution: we are simply the sort of beings who follow culturally and biologically conditioned practices. 19 JOHN R. SEARLE This interpretation of Wittgenstein is largely due to Saul Kripke (1982) and it has aroused considerable controversy. My own view is that Kripke has misinterpreted Wittgenstein in certain crucial respects, but whether or not his interpretation is correct, it has been a source of continuing discussion in contemporary philosophy. 3.5.3 Philosophical scepticism Important work on philosophical scepticism has been continued by philosophers who are inspired or provoked by Wittgenstein, notably Thomson Clarke and Barry Stroud. These philosophers point out that a really serious analysis of our use of epistemic dis- course shows that the problem of scepticism cannot be simply overcome by the usual analytic philosopher’s methods of pointing out that the sceptic raises the demand for justiﬁcation beyond that which is logically appropriate. Clarke and Stroud claim that the problem of scepticism goes deeper than this solution will allow. Following Wittgenstein in investigating the depth grammar of language, they ﬁnd that any solution to the sceptic’s predicament – that is, any justiﬁcation for our claims to have knowledge about the world – rests on a much deeper understanding of the difference between ordinary or plain discourse and philosophical discourse. Work in this line of research is continuing at present. 4 Overall Assessment I have not attempted to survey all of the main areas of activity in contemporary ana- lytic philosophy. Most importantly, I have left out contemporary work in ethics. Perhaps of comparable importance, I have had nothing to say about purely technical work in logic. There is, furthermore, a thriving branch of analytic philosophy called ‘action theory’, which should be mentioned at least in passing. The general aim of analytic action theory is to analyse the structure of human actions in terms of the causal rela- tions between such mental states as beliefs, desires and intentions, and the bodily move- ments which are in some sense constitutive of the actions. Finally, it is worth calling attention to the fact that among analytic philosophers there has been a great revival of interest in the history of philosophy. Traditional analytic philosophers thought of the history of philosophy as mostly the history of mistakes. Some of the history of the subject could be useful for doing real philosophy; but the overall conception was that the history of philosophy had no more special relevance to philosophy than the history	Blackwell
of mathematics to mathematics, or the history of chemistry to chemistry. This attitude has changed recently, and there is now a feeling of the historical continuity of analytic philosophy with traditional philosophy in a way that contrasts sharply with the origi- nal view of analytic philosophers, who thought that they marked a radical, or indeed, revolutionary break with the philosophical tradition. It is too early to provide an assessment of the contribution that will be made by work done in philosophy at the present time, or even in the past few decades. My own view is that the philosophy of mind and social philosophy will become ever more central to the entire philosophical enterprise. The idea that the study of language could replace the study of mind is itself being transformed into the idea that the study of language is really a branch of the philosophy of mind. Within the philosophy of mind, perhaps 20 CONTEMPORARY PHILOSOPHY IN THE UNITED STATES the key notion requiring analysis is that of intentionality – that property of the mind by which it is directed at or about or of objects and states of affairs in the world inde- pendent of itself. Most of the work done by analytic philosophers in the philosophy of mind has tended to cluster around the traditional mind–body problem. My own view is that we need to overthrow this problem: in its traditional version, it was based on the assumption that mental properties and physical properties were somehow different from each other, and that therefore, there was some special problem not like other prob- lems in biology as to how they could both be characteristics of the human person. Once we see that so-called mental properties really are just higher-level physical properties of certain biological systems, I believe this problem can be dissolved. Once it is dissolved, however, we are still left with the task of analysing what is the central problem in the philosophy of language and in cognitive science, as well as the philosophy of mind, namely, the way that human representational capacities relate the human organism to the world. What are called ‘language’, ‘mind’, ‘thinking’, ‘speaking’ and ‘depicting’ are just different aspects of this mode of relating to reality. I believe that the causal theory of reference will be seen to be a failure once it is recognized that all representations must occur under some aspect or other, and that the extensionality of causal relations is inadequate to capture the aspectual character of reference. The only kind of causation that could be adequate to the task of reference is intentional causation or mental causation, but the causal theory of reference cannot concede that ultimately reference is achieved by some mental device, since the whole approach behind the causal theory was to try to eliminate the traditional mentalism of theories of reference and meaning in favour of objective causal relations in the world. My prediction is that the causal theory of reference, though it is at present by far the most inﬂuential theory of reference, will prove to be a failure for these reasons. Perhaps the single most disquieting feature of analytic philosophy in the ﬁfty-year period that I have been discussing is that it has passed from being a revolutionary minority point of view held in the face of traditionalist objections to becoming itself the conventional, establishment point of view. Analytic philosophy has become not only dominant but intellectually respectable, and, like all successful revolutionary move- ments, it has lost some of its vitality in virtue of its very success. Given its constant demand for rationality, intelligence, clarity, rigour and self-criticism, it is unlikely that it can succeed indeﬁnitely, simply because these demands are too great a cost for many people to pay. The urge to treat philosophy as a discipline that satisﬁes emotional rather than intellectual needs is always a threat to the insistence on rationality and intelli-	Blackwell
gence. However, in the history of philosophy, I do not believe we have seen anything to equal the history of analytic philosophy for its rigour, clarity, intelligence and, above all, its intellectual content. There is a sense in which it seems to me that we have been living through one of the great eras in philosophy. References Austin, J. L. 1962: How to do Things with Words. Oxford: Clarendon Press. Donnellan, K. 1970: Proper Names and Identifying Descriptions. Synthèse , 21, 335–58. Feyerabend, P. 1975: Against Method. London: Humanities Press. Grice, H. P. 1957: Meaning. Philosophical Review, 66. 21 JOHN R. SEARLE —— 1968: Utterer’s Meaning, Sentence-Meaning, and Word-Meaning. Foundations of Language, 4, 1–18. Kripke, S. 1972: Naming and Necessity. In G. Harman and D. Davidson (eds) Semantics of Natural Language, Dordrecht: Reidel. —— 1982: Wittgenstein on Rules and Private Language. Cambridge, MA: Harvard University Press. Kuhn, T. 1962: The Structure of Scientiﬁc Revolutions . Chicago: University of Chicago Press. Putnam, H. 1975: The Meaning of ‘Meaning’. In his Philosophical Papers, Vol. 2: Mind, Language and Reality, Cambridge: Cambridge University Press. Quine, W. V. O. 1953: Two Dogmas of Empiricism. In his From a Logical Point of View, Cambridge, MA: Harvard University Press. Rawls, J. 1971: A Theory of Justice. Cambridge, MA: Harvard University Press. Searle, J. R. 1964: How to Derive ‘Ought’ from ‘Is’. Philosophical Review, 73. —— 1969: Speech Acts: An Essay in the Philosophy of Language. Cambridge: Cambridge University Press. —— 1979: Expression and Meaning. Cambridge: Cambridge University Press. —— 1980a: Minds, Brains and Programs. Behavioral and Brain Sciences, 3, 417–24. —— 1980b: Intrinsic Intentionality. Behavioral and Brain Sciences, 3, 450–6. —— 1983: Intentionality: An Essay in the Philosophy of Mind. Cambridge: Cambridge University Press. Wittgenstein, L. 1953: Philosophical Investigations (translated by G. E. M. Anscombe). Oxford: Blackwell. 22 Contemporary Philosophy: A Second Look B E R NA R D W I L L I A M S 1 The Identity of Analytical Philosophy Given the title of John Searle’s essay, this second introduction might have been expected to complement the ﬁrst geographically, by dealing with present philosophical develop- ments in places other than the United States, but this is not in fact what it will try to do. Philosophy in the United States, in other English-speaking parts of the world, and in many other countries as well, is now very largely the same. In these places, there is one philosophical culture, and inasmuch as it contains different approaches, and some of the philosophy that is done within that culture is distinct from ‘analytical’ philoso- phy, that itself is not a matter of geographical region. It is true that ‘analytical’ philosophy, the style of philosophy described in Searle’s essay and overwhelmingly represented in this volume, is often professionally distin- guished (in job advertisements, for instance) from ‘continental’ philosophy, and this does represent, in a clumsy way, something which until recently was true: that the ways in which philosophy was done in France, Germany and other countries of continental Europe were typically different from the ‘analytical’ style. To a much more limited extent, that remains so. (Chapters 40–2 describe the situation in continental Europe.) However, it is absurd to mark philosophical differences with these two labels. Apart from involving a strange cross-classiﬁcation – rather as though one divided cars into front- wheel drive and Japanese – the labels are seriously misleading, in helping one to forget lay in continental Europe (notably that the origins of analytical philosophy itself so, when its founding father is taken to be Frege and its greatest representative Wittgenstein), and that the interests of ‘continental’ philosophy are not conﬁned to the European continent. Moreover, it is not simply a matter of labelling. It is not that the distinction in itself	Blackwell
is unproblematical, and only needs more aptly chosen titles to represent it. The distinctions involved are obscure, and the titles serve to conceal this fact. The term ‘continental’ serves to discourage thought about the possible contrasts to analytical philosophy, and so about the identity of analytical philosophy itself. At the same time, the vague geographical resonance of the term does carry a message, that analytical philosophy is familiar as opposed to exotic, and perhaps – if some older stereotypes are in play – that it is responsible as opposed to frivolous. This is indeed what many BERNARD WILLIAMS analytical philosophers believe about it, and they believe it not so much in contrast to activities going on in some remote continent, but, in many cases, as opposed to work done in their local departments of literature. It is not true that work in other styles does not exist in the heartlands of analytical philosophy; it merely does not exist in depart- ments of philosophy. The distinctions involved are not geographical but professional, and what is at issue is the identity of philosophy as a discipline. In particular, what is at issue is the identity of philosophy as a subject that can sustain ongoing, cumulative, research. If it can do this, it can make a claim which the human- ities do not always ﬁnd it easy to make, except to the extent that they are branches of history: that there is something to be found out within their disciplines, that they can add to knowledge. It has been part of the attraction of analytical philosophy that, without the procedures of the experimental or theoretical sciences and with a more human subject matter, it can claim to achieve results which command, if not agree- ment, at least objective discussion, and which represent intellectual progress. It has achievements that are not arbitrarily personal, and they compare favourably to those of the social sciences (at least if one leaves aside the quite peculiar case of economics). I do not think that these claims are empty. I think that the achievements of ana- lytical philosophy are remarkable, and I agree with Searle that the subject is in various ways more interesting than it was forty years ago. Its virtues are indeed virtues. I think it is hard to be in good faith a teacher of philosophy unless you believe that there is something worth doing that you, in virtue of your experience, can do and which you can help other people to do. I think that the virtues of good philosophy are to a con- siderable extent workmanlike. Quite certainly, no philosophy which is to be worthwhile should lose the sense that there is something to be got right, that it is answerable to argument and that it is in the business of telling the truth. These things, I believe, are represented by the best of what is called analytical phi- losophy, and to that extent I am committed to it. My own work has largely been in its style. Yet, having now worked in it for a long time, and having, like Searle, seen it change, I am a great deal more puzzled about it than I once was; in particular, I am puzzled about the ways in which it must understand itself if it is to have those virtues, and also about the costs of sustaining those virtues. There is one understanding of these virtues which is certainly widespread among analytical philosophers, and which directly serves the promise of ongoing research: that these are indeed the virtues of a science. Some philosophers who are impressed by this conception of what they are doing ritualize it into the forms of presentation familiar from the sciences. Sometimes this is mere scientism, but in other cases it signals the fact that their branch of phi- losophy is near neighbour to some science, such as quantum mechanics or cognitive science. But the virtues of workmanlike truthfulness which analytical philosophy typically	Blackwell
cultivates are much more important than any attempt to make philosophy look like a science. With many other branches of philosophy there is no plausible version of sharing a party wall with science, and yet these virtues are still regarded as virtues. In fact, even in the case of the more scientiﬁc areas of philosophy, it is obvious enough that these virtues are not recommended only because they are possessed by its scien- tiﬁc neighbours: they are taken to be intellectual virtues, good in the same way for philosophy and for science. 24 CONTEMPORARY PHILOSOPHY: A SECOND LOOK But how far can philosophy cultivate just these virtues and remain true to other aspects of the legacy of the subject, to aims that it has pursued in the past? The sci- ences aim to make claims that can and should be conveyed in ways that are minimally expressive; they are not meant to convey feeling, or to display much literary imagina- tion, or to speak (at least overtly) in a persuasive mode. But if we think in particular of moral or political philosophy, is this ambition actually true to the traditions of the subject, even as those are embodied in the historical canon of analytical philosophy itself? Is it true to a tradition that contains Plato, Hobbes, Hume, Kant and (come to that) John Stuart Mill? Is it in fact true to any great ﬁgure of that tradition, except perhaps Aristotle? And if we are to take him as our model, we are left with many ques- tions to consider – whether, for instance, the affectless treatises that we possess do rep- resent his voice; if so, whether the tone does not represent a quite special view of ethical life; and whether we should not weigh rather soberly the fact that the closest previous imitation of Aristotle was to be found in a movement called scholasticism. Particularly in moral and political philosophy, but not only there, there is a question of what the procedures typical of analytical philosophy mean. There are many virtuous and valuable things that they make possible, and at the same time there are resources of philosophy in the past that they seemingly exclude, and it is important not to assume that this balance is simply given to us, above all by an unquestionable and transparent interpretation of the ideals of intellectual responsibility. I do not want to suggest that the adoption of the analytical style is a mere abdication, a cowardly refusal to adopt a more imaginative and committed manner which (critics sometimes suggest) is obviously to hand. Still less is it simply a matter of scientistic camouﬂage. It is a feature of our time that the resources of philosophical writing typically available to analytical philosophy should present themselves so strongly as the responsible way of going on, the most convincing expression of a philosopher’s claim on people’s atten- tion. But that is an historical fact, and we should try to understand it as such. I do not think that I adequately understand it, and, for that reason, I would not like to predict what other possibilities there may prove to be for a philosophy that preserves the merits of analytical philosophy. In the rest of this essay, I shall try to give an outline of some principal concerns of analytical moral and political philosophy. This will supplement the account that John Searle has given of the state of the art in other areas, but I hope also that in describing some of what analytical philosophy has recently done for these subjects, it may en- courage readers to ask what new things it might be able to do. 2 Meta-ethics Philosophical studies have often been understood, in the analytical tradition but not only there, as being higher-order, in the sense that natural science, for instance, will study natural phenomena, while the philosophy of science will study, from some par- ticular points of view, the operations of science. Some of moral philosophy (or, as I shall	Blackwell
also call it, ethics) is certainly a higher-order study. It discusses such things as the nature of moral judgements, and asks whether they express genuine beliefs, whether they can be objectively true, and so forth. Such higher-order questions are the concern 25 BERNARD WILLIAMS of meta-ethics. At one time (thirty to ﬁfty years ago) it was widely thought in analyti- cal philosophy that ethics consisted only of meta-ethics. A powerful source of this con- ception was the belief in a ﬁrm distinction between fact and value, to which Searle has referred. However, the idea of ethics as simply meta-ethics does not follow directly from the distinction between fact and value, and those who used the distinction to support that idea needed two further assumptions. One was that philosophy itself should be in the business of ‘fact’ (which, for the pur- poses of the distinction, included theory) and not of value. This was connected with a certain conception of philosophy, important to the identity of analytical philosophy, in which it is taken to derive its authority from its theoretical stance as an abstract intel- lectual enquiry. Some earlier philosophers, such as G. E. Moore, had indeed believed in the distinction between fact and value, but had supposed that philosophers, in one way or another, could have quite a lot to say substantively about values. The journey from the fact–value distinction to a view of ethics as only meta-ethics involved the assump- tion that this was impossible or inappropriate. The second assumption involved in the journey was that meta-ethics itself could be value-neutral, that the study of the nature of ethical thought did not commit one to any substantive moral conclusions. A yet further assumption, which was not necessary to the journey but did often accompany it, was that meta-ethics should be linguistic in style, and its subject should be ‘the language of morals’. This latter idea has now almost entirely disappeared, as the purely linguistic conception of philosophical study has more generally retreated. Beyond that, however, there are now more doubts about the extent to which meta-ethics can be value-neutral, and, in addition, philosophers simply feel freer in making their own ethical commitments clear. Meta-ethics remains a part of ethics, but most writings in philosophical ethics now will declare substantive moral positions, either in close association with some meta-ethical outlook or in a more free-standing manner. Recent meta-ethical discussions have carried on the traditional interest in the objec- tivity of ethics. In this connection, ‘moral judgements’ are often grouped together and compared with everyday statements of fact or with scientiﬁc statements. Some theories claim that moral judgements lack some desirable property that factual statements can attain, such as objectivity or truth, while other theories claim that they can have this property. These debates, particularly those conducted under the inﬂuence of positivism, have tended to assimilate two different issues. One concerns the prospects of rational agreement in ethical matters. The other concerns the semantic status of moral judge- ments: whether they are, typically, statements at all (as opposed, for instance, to prescriptions), and whether they aim at truth. Objectivity is best understood in terms of the ﬁrst kind of question. There clearly are substantive and systematic disagreements about ethical questions, both between dif- ferent societies or cultures, and within one society (particularly when, as now, the culture of one society may be highly pluralist). Some of these disagreements may turn out to be due to misunderstanding or bad interpretation and dissolve when local prac- tices are better understood, but this is not true of all of them. Since ancient times it has been suggested that these disagreements have a status different from disagreements about facts or about the explanation of natural phenomena. With the latter, if the	Blackwell
parties understand the question at issue, they see how after further enquiry they may 26 CONTEMPORARY PHILOSOPHY: A SECOND LOOK end up in one of several positions: they may come to rational agreement on one answer or another, they may recognize that such evidence as they can obtain underdeter- mines the answer and leaves them with intelligible room for continued disagreement, or they may advance in understanding to a point at which they see that the question in its original form cannot be answered, for instance because it was based on a false presupposition. By contrast, it is suggested that we can understand an ethical dispute perfectly well, and yet it be clear that it need not come out in any of these ways. Disagreeing about an ethical matter, the parties may radically disagree about the kinds of considerations that would settle the question, and the suggestion is that at the end of the line there may be no rational way of arriving at agreement. This is the suggestion that ethical claims lack objectivity. Some theories have associated this position with a view about the semantic status of moral utterances. Emotivism, a theory closely associated with positivism, held that moral utterances were merely expressions of emotion, not far removed from expletives, and it took this, reasonably enough, not to be an objectivist theory. In this case, the semantic account and the denial of objectivity went closely together. However, it is a mistake to think that the two issues are in general simply related to one another. A clear illustration of this is Kant’s theory. Kant supposed that moral statements, or at any rate the most basic of them, were actually prescriptions, and he understood the fundamental principle of morality to be an imperative. However, when the issue is expressed in terms of rational agreement or disagreement, Kant is quite certainly an objectivist: the Categorical Imperative, together in some cases with empirical informa- tion, determines for any rational agent what morality requires, and all rational agents are in a position to agree on it. Another example of objectivity which is at least non- committal about the semantics involved comes from virtue theory. Aristotle believed that experienced and discriminating agents who had been properly brought up would reach rational agreement in action, feeling, judgement and interpretation. He believed, moreover, that this possibility was grounded in the best development and expression of human nature, and that views about what counted as the best human development could themselves command rational agreement. This certainly offers a kind of objec- tivity, but it does not particularly emphasize agreement in belief; no doubt some agree- ment in belief will matter, but so equally will agreement in feeling and in practical decision. However, even if objectivity need not imply rational agreement in belief, it may be argued that the converse holds: that a theory which represents moral judgements as basically expressing beliefs must be committed to objectivity. Beliefs, this argument goes, are true or false. If moral judgements express beliefs, then some of them are true and there is such a thing as truth in morality. So if people disagree about what to believe, someone must be wrong. This certainly sounds as though there must be objec- tivity. The difﬁculty with this argument is that it seems to be too easy to agree that moral judgements admit of truth or falsehood. They are certainly called ‘true’ and ‘false’, as even the emotivists had to concede, and the claim that nevertheless they are not really true or false needs some deciphering. Emotivism itself offered a semantic analysis in terms of which such judgements turn out not really to be statements, which certainly gives some content to the claim that they are not really true or false. However, such 27 BERNARD WILLIAMS	Blackwell
analyses run into difﬁculty precisely because the air of being a statement that sur- rounds moral judgements is not merely superﬁcial – they behave syntactically just as other kinds of statements do. An alternative is to argue that moral judgements can indeed be true or false, but that nothing interesting follows from this. On some theories of truth, sometimes called ‘redundancy’ theories, to claim that ‘P’ is true is to do no more than to assert that P. Any theory of truth must accept the equivalence ‘‘‘P’’ is true if and only if P’; the pecu- liarity of redundancy theories is to claim that this is all there is to the nature of truth. If this is correct, then the truth or falsehood of moral judgements will follow simply from their taking a statemental form, which allows them to be asserted or denied. Objec- tivity will then either be understood as something that necessarily goes along with truth and falsehood, in which case, on the redundancy theory, it will be no more interesting or substantive than truth; or it will be more interesting and substantive – implying for instance the possibility of rational agreement – in which case, on the redundancy view, it will not follow just from the fact that moral judgements can be true or false. It is widely, though not universally, agreed that an adequate theory of truth needs to go beyond the redundancy view, but it is disputed how far it needs to go. Some argue that if one takes seriously the claim that a given proposition is true, then this does imply the idea that there could be convergence in belief on the proposition under favourable circumstances. This approach brings the idea of truth itself nearer to that of objectiv- ity as that has been introduced here. Others hold that a properly ‘minimalist’ theory of truth need not bring in such a strong condition. If objectivism and the mere truth of moral statements have often been assimilated to one another, realism, equally, is often assimilated to one or both of them. Yet we should expect realism, if it is an issue at all, to be a further issue. Elsewhere in phi- losophy, for instance in the philosophy of mathematics or the philosophy of science, it can be agreed that statements of a certain kind (about numbers, or about subatomic particles) are capable of truth, and also that they can command rational agreement, and yet it is thought by many philosophers that this does not answer the question whether those statements should be interpreted realistically, where this means (very roughly indeed) that the statements are taken to refer to objects that exist independently of our thoughts about them. Even if it is not easy to give a determinate sense to such questions, at any rate one would not expect realism to follow trivially from the claim that moral statements can be true or false. Some philosophers, inﬂuenced by the late John Mackie and in the line of Hume, deny realism by claiming that the moral properties of people, actions and so forth, are not ‘in the world’ but are ‘projected’ on to it from our feelings and reactions. According to the most familiar version of this view, secondary qualities such as colours are also pro- jected on to the world, and this raises the question whether the metaphor has not mis- located the most signiﬁcant issues about moral properties. The theory implies that ethical outlooks are ‘perspectival’ or related to human experience in ways in which physical theory (at least) is not, but this does not take us very far: it will not tell us any- thing very distinctive about ethical realism to know only that ethical concepts are per- spectival in a sense in which colour concepts, or perhaps psychological concepts, are also perspectival. An anti-realism that gives moral properties much the same status as 28 CONTEMPORARY PHILOSOPHY: A SECOND LOOK colours will probably satisfy many moral realists. We need to ask how far the moral concepts and outlooks of various human groups can intelligibly differ while the rest of	Blackwell
their ways of describing the world, in particular their psychological concepts for describing people’s behaviour, remain the same. Again, how far can their psychologi- cal concepts themselves intelligibly vary, and how should we understand those variations? In considering such questions, it is helpful to abandon a very limiting assumption which has been made up to this point in the discussion, namely, that all ‘moral judge- ments’ are essentially of the same kind and stand in the same relation to such matters as truth and objectivity. In considering moral disagreement, philosophers have con- centrated on cases in which the parties express themselves in terms of the ‘thin’ ethical concepts such as ‘good’, ‘right’ and ‘ought’. The parties share the same moral and other concepts, and disagree about whether a given judgement should be asserted or denied: they disagree, for instance, about whether capital punishment is wrong. To represent disagreement in this way may seem to isolate in a helpful way its moral focus. But a lot of moral discussion – to differing degrees in different societies – is conducted in terms of ‘thick’ concepts, such as ‘brutality’ or ‘betrayal’ or ‘sentimentality’, and it is a mistake to suppose that such concepts are merely convenient devices for associating a bunch of empirical considerations with a thin ethical concept. It has been increasingly accepted in recent discussions that the application of such concepts is guided by their evaluative point, and that one cannot understand them without grasping that point. (This does not mean that anyone who understands such a concept must have adopted it as his or her own, but it does mean that he or she needs to have imaginatively iden- tiﬁed, as an ethnographer does, with those who use it.) At the same time, however, such concepts apply to some empirical states of affairs and not to others, and there is room for truth, objectivity and knowledge to be displayed in their application. If this is correct, then it may be more helpful to consider ethical disagreement, not at the ultimate cutting edge of the practical judgements about what ought or ought not to happen, but further back, in the network of more substantive and thicker concepts that back up such judgements. Such concepts will typically serve more purposes than expressing bare ‘moral judgements’. They may play a role, for instance, in a scheme of psychological explanation. The question will then become, rather, why and to what extent different cultures differ in their ethical concepts, and, more broadly, in the frame- works of understanding that go with such concepts. Seen in this light, meta-ethical questions move further away from being questions in the philosophy of language or the theory of justiﬁcation or epistemology, and become more like questions in the theory of cultural understanding. Indeed they may become directly questions of cultural understanding. The most basic question about objectivity may turn out to be the ques- tion of the extent to which different human societies share an underlying determinate framework of ethical concepts. By turning in such a direction, philosophical discussion becomes more empirical and historical, more richly related to other disciplines, and more illuminating. At the same time, it means that philosophers have to know about more things, or people in other disciplines have to take on issues in philosophy. To that extent, philosophy tends to lose a distinctive subject matter and its identity becomes blurred. 29 BERNARD WILLIAMS 3 Ethical Theory I have already said that analytical philosophers are happier than they once were to recognize that what they say in moral philosophy is likely to have a substantive ethical content, and that even meta-ethics is likely to have some such consequences. There is a problem, however, about how this is related to the authority of philosophy. If philoso- phers are going to offer moral opinions – within their subject, that is to say, and not	Blackwell
simply as anyone offers moral opinions – they need to have some professional claim to attention. They are not, as philosophers, necessarily gifted with unusual insight or imagination, and they may not have a signiﬁcantly wide experience or knowledge of the world. Their claim to attention rests on their capacity for drawing distinctions, seeing the consequences of assumptions, and so forth: in general, on their ability to develop and control a theoretical structure. If the authority of philosophy lies in its status as a theoretical subject, the philosopher’s special contribution to substantive ethical issues is likely to be found in a theoretical approach to them. One of the most common enterprises in moral philosophy at present is the development of various ethical theories. The aim of ethical theory is to cast the content of an ethical outlook into a theoreti- cal form. An ethical theory must contain some meta-ethics, since it takes one view rather than another of what the structure and the central concepts of ethical thought must be, though it need not have an opinion on every meta-ethical issue. It is commit- ted to putting forward in a theoretical form a substantive ethical outlook. In doing this, ethical theories are to different degrees revisionary. Some start with a supposedly unde- niable basis for ethics, and reject everyday moral conclusions that conﬂict with it. (It is a good question why the basis should be regarded as undeniable if it has such con- sequences.) Others, less dogmatically, consider the moral conclusions that would be delivered by conﬂicting outlooks, and decide which outlook makes the most coherent systematic sense of those conclusions that we (that is to say, the author and those readers who agree with him or her) ﬁnd most convincing. Unsystematized but carefully considered judgements about what we would think it right to do in a certain situation, or would be prepared to say in approval or criticism of people and their actions, are often called in the context of such a method ‘moral intuitions’. (The term ‘intuition’ has a purely methodological force: it means only that these judgements seem to us, after consideration, pre-theoretically convincing, not that they are derived from a faculty of intuiting moral truths.) A preferred method is to seek what John Rawls has called a ‘reﬂective equilibrium’ between theory and intuitions, modifying the theory to accom- modate robust intuitions, and discarding some intuitions which clash with the theory, particularly if one can see how they might be the product of prejudice or confusion. Moral theories are standardly presented as falling into three basic types, centring respectively on consequences, rights and virtues. The ﬁrst are unsurprisingly called ‘consequentialist’, and the last ‘virtue theories’. The second are often called ‘deonto- logical’, which means that they are centred on duty or obligation, but this is a cross- classiﬁcation, since consequentialist theories also give prominence to an obligation, that of bringing about the best consequences. (In the case of the most familiar and the consequences is popular consequentialist theory, utilitarianism, the value of 30 CONTEMPORARY PHILOSOPHY: A SECOND LOOK expressed in terms of welfare or preference-satisfaction.) In terms of obligations, the difference is rather that pure consequentialist theories present only one basic obliga- tion, while the second type of theory has many. A more distinctive mark of difference is to be found in the idea of a right: the second type of theory grounds many of an agent’s obligations in the rights of others to expect certain behaviour from that agent, a kind of consideration that utilitarians and other consequentialists regard as being at best derivative, and at worst totally spurious. Another way of understanding the division into three is in terms of what each	Blackwell
theory sees as most basically bearing ethical value. For the ﬁrst type of theory, it is good states of affairs, and right action is understood as action tending to bring about good states of affairs. For the second type, it is right action; sometimes what makes an action right is a fact about its consequences, but often it is not – its rightness is determined rather by respect for others’ rights, or by other obligations that the agent may have. Virtue theory, ﬁnally, puts most emphasis on the idea of a good person, someone who could be described also as an ethically admirable person. This is an important empha- sis, and the notion of a virtue is important in ethics. However, once the types of theory are distinguished in this way, it is hard to see them as all in the same line of business. Consequentialist and rights theories aim to systematize our principles or rules of action in ways that will, supposedly, help us to see what to do or recommend in particular cases. A theory of the virtues can hardly do that: the theory itself, after all, is going to say that what you basically need in order to do and recommend the right things are virtues, not a theory about virtues. Moreover, virtuous people do not think always, or usually, about the virtues. They think about such things as good consequences or people’s rights, and this makes it clear that ‘virtue theory’ cannot be on the same level as the other two types of theory. 4 Morality, Politics and Analytical Philosophy Among moral concepts, that of rights is closest to law and also to politics, and phi- losophical discussions of them often cross those boundaries. Given these relations, it is not surprising that the kind of theory most often constructed to articulate the idea of moral rights is contractualist, invoking the idea of an agreement that might be ratio- nally arrived at by parties in some hypothetical situation in which they were required to make rules by which they could live together. The inspiration of contractualist theories goes back, in particular, to Kant. Kant’s own construction relies on some ideas that are not shared by many modern theorists, in particular that a commitment to the basic principle of morality (the so-called ‘Categorical Imperative’) is presupposed by the very activity of a rational agent. It also involves a very obscure doctrine of freedom. The modern theories inspired by Kantian ideas are less committed than Kant was to showing that morality is ultimate rationality, and they allow also more empirical material into the construction than Kant did. The leading example of such a theory is that of John Rawls. His model of a set of contracting parties reaching an agreement behind ‘a veil of ignorance’ has had an immense inﬂuence on thinking about morality. It was designed for a purpose in politi- cal philosophy, of constructing a theory of social justice. In Rawls’s theory, the veil of 31 BERNARD WILLIAMS ignorance is introduced to disguise from each contractor his own particular advantages and disadvantages and his own eventual position in the society that is being designed. The political theory that uses this thought experiment is liberal, giving a high priority to liberty and at the same time emphasizing redistribution of resources in the interests of the disadvantaged. It is signiﬁcant that when Rawls ﬁrst produced his theory he saw it in universalist terms, as offering a construction of social justice which would apply to any society that met the conditions (very roughly) of being able to think about social ideals and having the resources to implement them. Now, however, he has moved in the direction of seeing the theory of justice as one that expresses the aspirations of a particular social formation, the modern pluralist state. Much recent political philosophy has centred on this liberal project, of deﬁning terms of just coexistence for people living in a pluralist society. One interpretation of that aim	Blackwell
is to look for terms of coexistence that will not presuppose a common conception of the good. On such an account, citizens can understand themselves as sharing a social exis- tence although they have as individuals, or as members of communities less extensive than the state, varying conceptions of a good life. Rawls has interpreted his own purpose in those terms; it is expressed, for instance, in the fact that the parties in the original position were supposed to make their decision on (broadly) self-interested grounds, and not in the light of any antecedent conception of what a good society would be. The values that they were taken to have were expressed only in the list of ‘primary goods’ in terms of which they made their choice, a list which made it clear already that they set a high value on liberty, for instance, and did not assess everything in terms of utility. In fact, Rawls’s aim of making his theory as independent as possible from substan- tial claims about the good does put it under some strain. At least in the ﬁrst version of the theory, the basic conception of justice included a large-scale commitment to eco- nomic redistribution, and while this made it very welcome to many liberals (particu- larly in the American sense of that word), it laid it open to the criticism from others more inclined to libertarianism, such as Robert Nozick, that the theory incorporated not only rightful terms of coexistence but a substantial and distinctive conception of social justice as economic equality. In the later versions of Rawls’s theory, this concep- tion is less prominent, but even more weight than before is laid on the idea that coex- istence in a liberal pluralist society is not ‘a mere modus vivendi’ but a condition that calls on important moral powers of toleration and respect for autonomy. This empha- sis does seem to express a distinctive conception of the good, of a Kantian kind, to a point at which it looks as though the condition of pluralism is not simply a contingency of modern life, but an important vehicle of ethical self-expression. Others, such as Ronald Dworkin, have pursued liberal theory, and in pluralist terms, while accepting a commitment to a distinctive conception of the good. Others, again, have claimed to reject liberal pluralism altogether and have turned in what is some- times called a ‘communitarian’ direction. It is hard to tell in some cases where writers of this tendency stand in relation to the politics of liberalism. While a few, notably Alasdair MacIntyre, despairing of the whole enterprise, try only to diagnose our condition and store some ethical goods for a better time, others seem to share in the liberal undertaking but prefer, in opposition to Rawls’s Kantianism, a more Hegelian type of discourse to express it. 32 CONTEMPORARY PHILOSOPHY: A SECOND LOOK It is a rather odd feature of communitarian theories, at least if they take a tradi- tionalist turn, that they recommend a politics that does not sit very easily with the exis- tence of such theories themselves, except perhaps as a kind of interim measure. They seek a politics in which people’s relations are formed by shared understandings which to a considerable extent must be unspoken and taken for granted, and the exchange of abstract political theories plays no obvious role in this conception of social life. (Hegel, many of whose concerns are re-enacted in these debates, thought he had an answer to this, in his conception of a society that could ultimately reconcile abstract under- standing and concrete practice, but few current disputants are happy to pay the price of admission to the Hegelian system.) The liberals, on the other hand, have a conception of modern political life which at least coherently embraces the existence of their own theories, since they understand the modern state as a formation in which authority is peculiarly vested in discursive argument, rather than in traditional or charismatic leadership. It is true that many	Blackwell
political philosophers in the analytical tradition (unlike Rawls himself, and also unlike Habermas, who comes from a sociological tradition) do not see the role of their theo- ries in these terms, but rather as advancing trans-historical views about the demands of a just political order. But even if liberals do not always recognize the point them- selves, the role of theory in liberal political philosophy can be given a special justiﬁca- tion in terms of current political reality. Liberals coherently believe that the project of political theory makes sense, because they are committed to thinking that in our circumstances it makes sense to engage in a political activity of explaining the basic principles of democratic government in such terms. In this respect, moral philosophy is in an altogether worse situation. It typically lacks an account of why the project of articulating moral theories makes any sense at all. As many writers have pointed out, it bears little relation to the psychology of people’s ethical lives, and inasmuch as it claims that turning morality into a theory makes it more rational, there is a pressing question of what concept of rationality is being invoked. To a limited extent, there may be an answer to that question, inasmuch as some ethical questions, such as those raised by medical ethics, are public questions, closely tied to politics and the law. In those cases, we need a public discourse to legiti- mate some answers, since it is a public issue what should be permitted, and mere appeals to ethical or professional authority will no longer do. But it would be a mistake to suppose that in such cases we are presented with a pure concept of moral rational- ity which we then apply to our historical circumstances. Rather, what counts as a ratio- nal way of discussing such questions is inﬂuenced by the historical circumstances, and above all by the need to give a discursive justiﬁcation, in something like a legal style, for procedures which increasingly are adopted in a public domain and can be challenged in it. Moreover, many important ethical issues are not of this kind at all. Morality has always been connected not only with law and politics, but also with the meaning of an individual’s life and his or her relations to other people. In these connections, the authority of theory over the moral life remains quite opaque. Certainly, it will not do to rely on the inference: philosophy must have something to say about the moral life; the most responsible form of philosophy is analytical philosophy; what analytical phi- losophy is best at is theory; so philosophy’s contribution to the moral life is theory. 33 BERNARD WILLIAMS In rejecting this uninviting argument, some will want to attack the second premise. However, it may be more constructive, and offer more of a challenge to thinking about what one wants of philosophy, if one reconsiders the third premise, that what analyti- cal philosophy does best is theory. Analytical philosophy’s own virtues, such as its unfanatical truthfulness, could encourage it in ethics to remind us of detail rather than bludgeon us with theory. Truthfulness in personal life, and even in politics, is not necessarily opposed to the exercise of the imagination. It is relevant here that an imaginative truthfulness is a virtue in the arts. Writers in moral philosophy sometimes urge us to extend our ethical understanding by turning to imaginative literature, in particular the novel. To the extent that this is good advice, it is not because novels are convenient sources of psy- chological information, still less because some of them are morally edifying. It is because imaginative writing can powerfully evoke the strength of ethical considera- tions in giving sense to someone’s life or to a passage of it, and, equally, present the pos-	Blackwell
sible complexity, ambivalence and ultimate insecurity of those considerations. Good literature stands against the isolation of moral considerations from the psychological and social forces that both empower and threaten them. But this isolation of moral considerations from the rest of experience is an illusion very much fostered by moral philosophy itself; indeed, without that illusion some forms of moral philosophy could not exist at all. So there are lessons here not just for philosophy’s use of other writing, but for philosophical writing itself. The truthfulness that it properly seeks involves imaginative honesty and not just argumentative accuracy. Analytical philosophy, or some recognizable descendant of it, should be able to make a richer contribution to ethics than has often been the case up to now. If it is to do so, it will need to hold on to two truths which it tends to forget (not only in ethics, but most damagingly there): that philosophy cannot be too pure, and must merge with other kinds of understanding; and that being soberly truthful does not exclude, but may actually demand, the imagination. 34 PART I AREAS OF PHILOSOPHY 1 Epistemology A. C. G R AY L I N G For most of the modern period of philosophy, from Descartes to the present, epistemol- ogy has been the central philosophical discipline. It raises questions about the scope and limits of knowledge, its sources and justiﬁcation, and it deals with sceptical arguments concerning our claims to knowledge and justiﬁed belief. This chapter ﬁrstly considers dif- ﬁculties facing attempts to deﬁne knowledge and, secondly, explores inﬂuential responses to the challenge of scepticism. Epistemology is closely related to METAPHYSICS (chapter 2), which is the philosophical account of what kinds of entities there are. Epistemologi- cal questions are also crucial to most of the other areas of philosophy examined in this volume, from ETHICS (chapter 6) to PHILOSOPHY OF SCIENCE (chapter 9) and PHILOSO- PHY OF MATHEMATICS (chapter 11) to PHILOSOPHY OF HISTORY (chapter 14). Chapters on individuals or groups of philosophers from DESCARTES (see chapter 26) to KANT (chapter 32) discuss classical epistemology, while several chapters about more recent philosophers also follow epistemological themes. Introduction Epistemology, which is also called the theory of knowledge, is the branch of philosophy concerned with enquiry into the nature, sources and validity of knowledge. Among the chief questions it attempts to answer are: What is knowledge? How do we get it? Can our means of getting it be defended against sceptical challenge? These questions are implicitly as old as philosophy, although their ﬁrst explicit treatment is to be found in PLATO (c.427–347 BC) (see chapter 23), in particular in his Theaetetus. But it is primar- ily in the modern era, from the seventeenth century onwards – as a result of the work of DESCARTES (1596–1650) (chapter 26) and LOCKE (1632–1704) (chapter 29) in asso- ciation with the rise of modern science – that epistemology has occupied centre-stage in philosophy. One obvious step towards answering epistemology’s ﬁrst question is to attempt a deﬁnition. The standard preliminary deﬁnition has it that knowledge is justiﬁed true belief. This deﬁnition looks plausible because, at the very least, it seems that to know something one must believe it, that the belief must be true, and that one’s reason for A. C. GRAYLING believing it must be satisfactory in the light of some criterion – for one could not be said to know something if one’s reasons for believing it were arbitrary or haphazard. So each of the three parts of the deﬁnition appears to express a necessary condition for knowl- edge, and the claim is that, taken together, they are sufﬁcient. There are, however, serious difﬁculties with this idea, particularly about the nature of the justiﬁcation required for true belief to amount to knowledge. Competing pro-	Blackwell
posals have been offered to meet the difﬁculties, either by adding further conditions or by ﬁnding a better statement of the deﬁnition as it stands. The ﬁrst part of the follow- ing discussion considers these proposals. In parallel with the debate about how to deﬁne knowledge is another about how knowledge is acquired. In the history of epistemology there have been two chief schools of thought about what constitutes the chief means to knowledge. One is the ‘rational- ist’ school (see chapters 26 and 27), which holds that reason plays this role. The other is the ‘empiricist’ (see chapters 29, 30 and 31), which holds that it is experience, prin- cipally the use of the senses aided when necessary by scientiﬁc instruments, which does so. The paradigm of knowledge for rationalists is mathematics and logic, where neces- sary truths are arrived at by intuition and rational inference. Questions about the nature of reason, the justiﬁcation of inference, and the nature of truth, especially nec- essary truth, accordingly press to be answered. The empiricists’ paradigm is natural science, where observation and experiment are crucial to enquiry. The history of science in the modern era lends support to empiri- cism’s case; but precisely for that reason philosophical questions about perception, observation, evidence and experiment have acquired great importance. But for both traditions in epistemology the central concern is whether we can trust the routes to knowledge they respectively nominate. Sceptical arguments suggest that we cannot simply assume them to be trustworthy; indeed, they suggest that work is required to show that they are. The effort to respond to scepticism therefore provides a sharp way of understanding what is crucial in epistemology. Section 2 below is accord- ingly concerned with an analysis of scepticism and some responses to it. There are other debates in epistemology about, among other things, memory, judge- ment, introspection, reasoning, the ‘a priori–a posteriori’ distinction, scientiﬁc method and the methodological differences, if any, between the natural and the social sciences; however, the questions considered here are basic to them all. 1 Knowledge 1.1 Deﬁning knowledge There are different ways in which one might be said to have knowledge. One can know people or places, in the sense of being acquainted with them. That is what is meant when one says, ‘My father knew Lloyd George’. One can know how to do something, in the sense of having an ability or skill. That is what is meant when one says, ‘I know how to play chess’. And one can know that something is the case, as when one says, ‘I know that Everest is the highest mountain’. This last is sometimes called ‘propositional knowledge’, and it is the kind epistemologists most wish to understand. 38 EPISTEMOLOGY The deﬁnition of knowledge already mentioned – knowledge as justiﬁed true belief – is intended to be an analysis of knowledge in the propositional sense. The deﬁnition is arrived at by asking what conditions have to be satisﬁed if we are correctly to describe someone as knowing something. In giving the deﬁnition we state what we hope are the necessary and sufﬁcient conditions for the truth of the claim ‘S knows that p’, where ‘S’ is the epistemic subject – the putative knower – and ‘p’ a proposition. The deﬁnition carries an air of plausibility, at least as applied to empirical knowl- edge, because it seems to meet the minimum we can be expected to need from so con- sequential a concept. It seems right to expect that if S knows that p, then p must at least be true. It seems right to expect that S must not merely wonder whether or hope that p is the case, but must have a positive epistemic attitude to it: S must believe that it is true. And if S believes some true proposition while having no grounds, or incorrect grounds, or merely arbitrary or fanciful grounds, for doing so, we would not say that S knows p; which means that S must have grounds for believing p which in some sense	Blackwell
properly justify doing so. Of these proposed conditions for knowledge, it is the third that gives most trouble. The reason is simply illustrated by counter-examples. These take the form of cases in which S believes a true proposition for what are in fact the wrong reasons, although they are from his or her own point of view persuasive. For instance, suppose S has two friends, T and U. The latter is travelling abroad, but S has no idea where. As for T, S saw him buying and thereafter driving about in a Rolls Royce, and therefore believes that he owns one. Now, from any proposition p one can validly infer the disjunction ‘p or q’. So S has grounds for believing ‘T owns a Rolls or U is in Paris’, even though, ex hypothesi, he has no idea of U’s location. But suppose T in fact does not own the Rolls – he bought it for someone else, on whose behalf he also drives it. And further suppose that U is indeed, by chance, in Paris. Then S believes, with justiﬁcation, a true proposition: but we should not want to call his belief knowledge. Examples like this are strained, but they do their work; they show that more needs to be said about justiﬁcation before we can claim to have an adequate account of knowledge. 1.2 Justiﬁcation A preliminary question concerns whether having justiﬁcation for believing some p entails p’s truth, for, if so, counter-examples of the kind just mentioned get no purchase and we need not seek ways of blocking them. There is indeed a view, called ‘infallibil- ism’, which offers just such a resource. It states that if it is true that S knows p, then S cannot be mistaken in believing p, and therefore his justiﬁcation for believing p guar- antees its truth. The claim is, in short, that one cannot be justiﬁed in believing a false proposition. This view is rejected by ‘fallibilists’, who claim that one can indeed have justiﬁcation for believing some p although it is false. Their counter to infallibilism turns on identi- fying a mistake in its supporting argument. The mistake is that whereas the truth of ‘S knows that p’ indeed rules out the possibility that S is in error, this is far from saying that S is so placed that he cannot possibly be wrong about p. It is right to say: (1) ‘it is impossible for S to be wrong about p if he knows p’, but it is not invariably right to say 39 A. C. GRAYLING (2) ‘if S knows p, then it is impossible for him to be wrong about p’. The mistake turns on thinking that the correct wide scope reading (1) of ‘it is impossible’ licenses the narrow scope reading (2) which constitutes infallibilism. An infallibilist account makes the deﬁnition of knowledge look simple: S knows p if his belief in it is infallibly justiﬁed. But this deﬁnition renders the notion of knowledge too restrictive, for it says that S can justiﬁably believe p only when the possibility of p’s falsity is excluded. Yet it appears to be a commonplace of epistemic experience that one can have the very best evidence for believing something and yet be wrong (as the account of scepticism given below is at pains to show), which is to say that fallibilism seems the only account of justiﬁcation adequate to the facts of epistemic life. We need justiﬁcation can give us an adequate therefore to see whether fallibilist theories of account of knowledge. The problem for fallibilist accounts is precisely the one illustrated by the Rolls Royce example above, and others similar to it (so-called ‘Gettier examples’, introduced in Gettier 1963), namely, that one’s justiﬁcation for believing p does not connect with the truth of p in the right way, and perhaps not at all. What is required is an account that will suitably connect S’s justiﬁcation both with his belief that p and with p’s truth. What is needed is a clear picture of ‘justiﬁed belief ’. If one can identify what justi- ﬁes a belief, one has gone all or most of the way to saying what justiﬁcation is; and en route one will have displayed the right connection between justiﬁcation, on the one	Blackwell
hand, and belief and truth on the other. In this connection there are several standard species of theory. Foundationalism One class of theories of justiﬁcation employs the metaphor of an ediﬁce. Most of our ordi- nary beliefs require support from others; we justify a given belief by appealing to another or others on which it rests. But if the chain of justifying beliefs were to regress without terminating in a belief that is in some way independently secure, thereby providing a foundation for the others, we would seem to lack justiﬁcation for any belief in the chain. It appears necessary therefore that there should be beliefs which do not need justiﬁca- tion, or which are in some way self-justifying, to serve as an epistemic underpinning. On this view a justiﬁed belief is one which either is, or is supported by, a founda- tional belief. The next steps therefore are to make clear the notion of a ‘foundation’ and to explain how foundational beliefs ‘support’ non-foundational ones. Some way of understanding foundationalism without reliance on constructional metaphors is needed. It is not enough barely to state that a foundational belief is a belief that requires no justiﬁcation, for there must be a reason why this is the case. What makes a belief inde- pendent or self-standing in the required way? It is standardly claimed that such beliefs justify themselves, or are self-evident, or are indefeasible or incorrigible. These are not the same things. A belief might be self-justifying without being self-evident (it might take hard work to see that it justiﬁes itself). Indefeasibility means that no further evidence or other, competing, beliefs, can render a given belief insecure. Yet this is a property that the belief might have independently of whether or not it is self-justifying. And so on. But what these characterizations are intended to convey is the idea that a certain immunity from doubt, error or revision attaches to the beliefs in question. 40 EPISTEMOLOGY It might even be unnecessary or mistaken to think that it is belief that provides the foundations for the ediﬁce of knowledge: some other state might do so. Perceptual states have been offered as candidates, because they appear to be suitably incorrigible – if one seems to see a red patch, say, then one cannot be wrong that one seems to see a red patch. And it appears plausible to say that one’s belief that p needs no further justiﬁcation or foundation than that things appear to one as p describes them to be. These suggestions bristle with difﬁculties. Examples of self-evident or self-justifying beliefs tend to be drawn from logic and mathematics – they are of the ‘x is x’ or ‘one plus one equals two’ variety, which critics are quick to point out give little help in grounding contingent beliefs. Perceptual states likewise turn out to be unlikely candidates for foun- dations, on the grounds that perception involves the application of beliefs which them- selves stand in need of justiﬁcation – among them beliefs about the nature of things and the laws they obey. What is most robustly contested is the ‘myth of the given’, the idea that there are ﬁrm, primitive and original data which experience supplies to our minds, antecedent to and untainted by judgement, furnishing the wherewithal to secure the rest of our beliefs. There is a difﬁculty also about how justiﬁcation is transmitted from foundational beliefs to dependent beliefs. It is too strong a claim to say that the latter are deducible from them. Most if not all contingent beliefs are not entailed by the beliefs that support them; the evidence I have that I am now sitting at my desk is about as strong as empirical evidence can be, yet given the standard sceptical considerations (such as, for example, the possibility that I am now dreaming) it does not entail that I am sitting here. If the relation is not a deductive one, what is it? Other candidate relations – inductive	Blackwell
or criterial – are by their nature defeasible, and therefore, unless somehow supplemented, insufﬁcient to the task of transmitting justiﬁcation from the foundations to other beliefs. The supplementation would have to consist of guarantees that the circumstances that defeat non-deductive justiﬁcation do not in fact obtain. But if such guarantees – under- stood, to avoid circularity, as not being part of the putative foundations themselves – were available to protect non-deductive grounds, then appeal to a notion of foundations looks simply otiose. 1.3 Coherence Dissatisfaction with foundationalism has led some epistemologists to prefer saying that a belief is justiﬁed if it coheres with those in an already accepted set. The immediate task is to specify what coherence is, and to ﬁnd a way of dealing in a non-circular way with the problem of how the already accepted beliefs came to be so. Hard on the heels of this task comes a number of questions. Is coherence a negative criterion (that is, a belief lacks justiﬁcation if it fails to cohere with the set) or a posi- tive one (that is, a belief is justiﬁed when it coheres with the set)? And is it to be under- stood strongly (by which coherence is sufﬁcient for justiﬁcation) or weakly (by which coherence is one among other justifying features)? The concept of coherence has its theoretical basis in the notion of a system, under- stood as a set whose elements stand in mutual relations of both consistency and (some kind of) interdependence. Consistency is of course a minimum requirement, and goes without saying. Dependence is more difﬁcult to specify suitably. It would be far too 41 A. C. GRAYLING strong – for it would give rise to assertive redundancy – to require that dependence means mutual entailment among beliefs (this is what some have required, citing geometry as the closest example). A more diffuse notion has it that a set of beliefs is coherent if any one of them follows from all the rest, and if no subset of them is logically independent of the remainder. But this is vague, and anyway seems to require that the set be known to be complete before one can judge whether a given belief coheres with it. A remedy might be to say that a belief coheres with an antecedent set if it can be inferred from it, or from some signiﬁcant subset within it, as being the best explanation in the case. To this someone might object that not all justiﬁcations take the form of explanations. An alternative might be to say that a belief is justiﬁed if it survives com- parison with competitors for acceptance among the antecedent set. But here an objec- tor might ask how this can be sufﬁcient, since by itself this does not show why the belief merits acceptance over equally cohering rivals. Indeed, any theory of justiﬁcation has to ensure as much for candidate beliefs, so there is nothing about the proposal that distinctively supports the coherence theory. And these thoughts leave unexamined the question of the ‘antecedent set’ and its justiﬁcation, which cannot be a matter of coherence, for with what is it to cohere in its turn? 1.4 Internalism and externalism Both the foundationalist and coherence theories are sometimes described as ‘internal- ist’ because they describe justiﬁcation as consisting in internal relations among beliefs, either – as in the former case – from a vertical relation of support between supposedly basic beliefs and others dependent upon them, or – as in the latter – from the mutual support of beliefs in an appropriately understood system. Generally characterized, internalist theories assert or assume that a belief cannot be justiﬁed for an epistemic subject S unless S has access to what provides the justiﬁcation, either in fact or in principle. These theories generally involve the stronger ‘in fact’ requirement because S’s being justiﬁed in believing p is standardly cashed in terms of his having reasons for taking p to be true, where having reasons is to be understood in	Blackwell
an occurrent sense. Here an objection immediately suggests itself. Any S has only ﬁnite access to what might justify or undermine his beliefs, and that access is conﬁned to his particular view- point. It seems that full justiﬁcation for his beliefs would rarely be available, because his experience would be restricted to what is nearby in space and time, and he would be entitled to hold only those beliefs which his limited experience licensed. A related objection is that internalism seems inconsistent with the fact that many people appear to have knowledge despite not being sophisticated enough to recognize that thus-and-so is a reason for believing p – that is the case, for example, with children. A more general objection still is that relations between beliefs, whether of the foun- dationalist or coherence type, might obtain without the beliefs in question being true of anything beyond themselves. One could imagine a coherent fairy tale, say, which in no point corresponds to some external reality, but in which beliefs are justiﬁed never- theless by their mutual relations. 42 This uneasy reﬂection prompts the thought that there should be a constraint on theories of justiﬁcation, in the form of a demand that there should be some suitable connection between belief possession and external factors – that is, something other than the beliefs and their mutual relations – which determines their epistemic value. This accordingly prompts the idea of an alternative: externalism. EPISTEMOLOGY 1.5 Reliability, causality and truth-tracking Externalism is the view that what makes S justiﬁed in believing p might not be anything to which S has cognitive access. It might be that the facts in the world are as S believes them to be, and that indeed they caused S to believe them to be so by stimulating his or her sensory receptors in the right kind of way. S need not be aware that this is how his or her belief was formed. So S could be justiﬁed in believing p without it. One main kind of externalist theory is reliabilism, the thesis – or cluster of theses – having it that a belief is justiﬁed if it is reliably connected with the truth. According to one inﬂuential variant, the connection in question is supplied by reliable belief-forming processes, ones which have a high success rate in producing true beliefs. An example of a reliable process might be normal perception in normal conditions. Much apparent plausibility attaches to theories based on the notion of external linkage, especially of causal linkage, between a belief and what it is about. An example of such a theory is Alvin Goldman’s (1986) account of knowledge as ‘appropriately caused true belief ’, where ‘appropriate causation’ takes a number of forms, sharing the property that they are processes which are both ‘globally’ and ‘locally’ reliable – the former meaning that the process has a high success rate in producing true beliefs, the latter that the process would not have produced the belief in question in some ‘rel- evant counterfactual situation’ where the belief is false. Goldman’s view is accordingly a paradigm of a reliabilist theory. An elegant second-cousin of this view is offered by Robert Nozick (1981). To the conditions (1) p is true and (2) S believes p Nozick adds (3) if p were not true, S would not believe p and (4) if p were true, S would believe it. Conditions (3) and (4) are intended to block Gettier-type counter-examples to the justiﬁed true belief analysis by annexing S’s belief that p ﬁrmly to p’s truth. S’s belief 43 A. C. GRAYLING that p is connected to the world (to the situation described by p) by a relation Nozick calls ‘tracking’: S’s belief tracks the truth that p. He adds reﬁnements in an attempt to deﬂect the counter-examples that philosophers are always ingenious and fertile at devising. If these theories seem plausible it is because they accord with our pre-theoretical views. But as one can readily see, there are plenty of things to object to in them, and a	Blackwell
copious literature does so. Their most serious ﬂaw, however, is that they are question- begging. They do not address the question of how S is to be conﬁdent that a given belief is justiﬁed; instead they help themselves to two weighty realist assumptions, one about the domain over which belief ranges and the other about how the domain and S are connected, so that they can assert that S is justiﬁed in believing a given p even if what justiﬁes him lies outside his own epistemic competence. Whatever else one thinks of these suggestions, they do not enlighten S, and therefore do not engage the same problem that internalist theories address. But worst of all – so an austere critic might say – the large assumptions to which these theories help themselves are precisely those that epistemology should be exam- ining. Externalist and causal theories, in whatever guise and combination, are better done by empirical psychology where the standard assumptions about the external world and S’s connections with it are premised. Philosophy, surely, is where these premises themselves come in for scrutiny. 1.6 Knowledge, belief and justiﬁcation again Consider this argument: ‘If anyone knows some p, then he or she can be certain that p. But no one can be certain of anything. Therefore no one knows anything.’ This argu- ment (advanced in this form by Unger 1975) is instructive. It repeats Descartes’s mistake of thinking that the psychological state of feeling certain – which someone can be in with respect to falsehoods, such as the fact that I can feel certain that Arkle will win the Derby next week, and be wrong – is what we are seeking in epistemology. But it also exempliﬁes the tendency in discussions of knowledge as such to make the deﬁ- nition of knowledge so highly restrictive that little or nothing passes muster. Should one care if a suggested deﬁnition of knowledge is such that, as the argument just quoted tells us, no one can know anything? Just so long as one has many well-justiﬁed beliefs which work well in practice, can one not be quite content to know nothing? For my part, I think one can. This suggests that in so far as the points sketched in preceding paragraphs have inter- est, it is in connection with the justiﬁcation of beliefs and not the deﬁnition of knowledge that they do so. Justiﬁcation is an important matter, not least because in the areas of application in epistemology where the really serious interest should lie – in questions about the PHILOSOPHY OF SCIENCE (chapter 9), the PHILOSOPHY OF HISTORY (chapter 14) or the concepts of evidence and proof in LAW (see chapter 13) – justiﬁcation is the crucial problem. That is where epistemologists should be getting down to work. By comparison, efforts to deﬁne ‘knowledge’ are trivial and occupy too much effort in epistemology. The disagreeable propensity of the debate generated by Gettier’s counter-examples – anticipated beautifully in Russell’s review of James (Russell 1910: 95) – to proceed on a chessboard of ‘-isms’, as exempliﬁed above, is a symptom. 44 The general problem with justiﬁcation is that the procedures we adopt, across all walks of epistemic life, appear highly permeable to difﬁculties posed by scepticism. The problem of justiﬁcation is therefore in large part the problem of scepticism; which is precisely why discussion of scepticism is central to epistemology. EPISTEMOLOGY 2 Scepticism Introduction The study and employment of sceptical arguments might in one sense be said to deﬁne epistemology. A chief epistemological aim is to determine how we can be sure that our means to knowledge (here ‘knowledge’ does duty for ‘justiﬁed belief ’) are satisfactory. A sharp way to show what is required is to look carefully at sceptical challenges to our epistemic efforts, challenges which suggest ways in which they can go awry. If we are able not just to identify but to meet these challenges, a primary epistemological aim will	Blackwell
have been realized. Scepticism is often described as the thesis that nothing is – or, more strongly, can be – known. But this is a bad characterization, because if we know nothing, then we do not know that we know nothing, and so the claim is trivially self-defeating. It is more telling to characterize scepticism in the way just suggested. It is a challenge directed against knowledge claims, with the form and nature of the challenge varying according to the ﬁeld of epistemic activity in question. In general, scepticism takes the form of a request for the justiﬁcation of those knowledge claims, together with a statement of the reasons motivating that request. Standardly, the reasons are that certain considerations suggest that the proposed justiﬁcation might be insufﬁcient. To conceive of scepticism like this is to see it as being more philosophically troubling and important than if it is described as a positive thesis asserting our ignorance or incapacity for knowledge. 2.1 Early scepticism Some among the thinkers of antiquity – Pyrrho of Elis (c.360–c.270 BC) and his school, and Plato’s successors in his Academy – expressed disappointment at the fact that cen- turies of enquiry by their philosophical predecessors seemed to have borne little fruit either in cosmology or ethics (this latter was broadly construed to include politics). Their disappointment prompted them to sceptical views. The Pyrrhonians argued that because enquiry is arduous and interminable, one should give up trying to judge what is true and false or right and wrong; for only thus will we achieve peace of mind. A less radical form of scepticism overtook Plato’s successors in the Academy. They agreed with Pyrrho that certainty must elude us, but they tempered their view by accepting that the practical demands of life must be met. They did not think it a work- able option to ‘suspend judgement’ as Pyrrho recommended, and therefore argued that we should accept those propositions or theories which are more PROBABLE (pp. 308–11) than their competitors. The views of these thinkers, known as Academic sceptics, are recorded in the work of Sextus Empiricus (c.150–c.225). 45 A. C. GRAYLING In the later Renaissance – or, which is the same thing, in early modern times – with religious certainties under attack and new ideas abroad, some of the sceptical argu- ments of the Academics and Pyrrhonians acquired a special signiﬁcance, notably as a result of the use to which René Descartes put them in showing that they are powerful tools for investigating the nature and sources of knowledge. In Descartes’s day the same person could be both astronomer and astrologer, chemist and alchemist, or physician and magician. It was hard to disentangle knowledge from nonsense; it was even harder to disentangle those methods of enquiry which might yield genuine knowledge from those that could only deepen ignorance. So there was an urgent need for some sharp, clean epistemological theorizing. In his Meditations (1986) Descartes accordingly identiﬁed epistemology as an essential preliminary to physics and mathematics, and attempted to establish the grounds of certainty as a propaedeutic to science. Descartes’s ﬁrst step in that task was to adapt and apply some of the traditional arguments of scepticism. (I shall comment on his use of scepticism again later.) The Anatomy of Scepticism Sceptical arguments exploit certain contingent facts about our ways of acquiring, testing, remembering and reasoning about our beliefs. Any problem that infects the acquisition and employment of beliefs about a given subject matter, and in particular any problem that infects our conﬁdence that we hold those beliefs justiﬁably, threatens our hold on that subject matter. The contingent facts in question relate to the nature of perception, the normal human vulnerability to error and the existence of states of mind – for example, dreaming and delusion – which can be subjectively indistinguishable from those that we	Blackwell
normally take to be appropriate for acquiring justiﬁed beliefs. By appealing to these considerations the sceptic aims to show that there are signiﬁcant questions to be answered about the degree of conﬁdence that we are entitled to repose in our standard epistemic practices. Sceptical considerations pose problems for epistemologists of both the rationalist and the empiricist camps. This division into competing schools of thought about knowledge is rough but useful, giving a shorthand way of marking the difference between those who hold that reason is the chief means to knowledge, and those who accord that role to expe- rience. Rationalists emphasize reason because in their view the objects of knowledge are propositions that are eternally, immutably and necessarily true – the examples they offer are the propositions of mathematics and logic – and these, they say, can only be acquired by reasoning. Empiricists hold that substantive and genuine knowledge of the world can only be learned through experience, by means of the senses and their extension via such instruments as telescopes and microscopes. The rationalist need not deny that empirical awareness is an important, even an ineliminable, aid to reason, nor need the empiricist deny that reason is an important, even an ineliminable, aid to experience; but both will insist that the chief means to knowledge is respectively one or the other. The reﬁnements of debate about these matters merit detailed examination for which this is not the place. For present purposes, the point to note is that scepticism is a problem for both schools of thought. For both, possibilities of error and delusion pose a challenge. For the empiricist in particular, to these must be added distinctive problems about perception. 46 EPISTEMOLOGY 2.2 Error, delusion and dreams One characteristic pattern of sceptical argument is drawn from a set of considerations about error, delusions and dreams. Consider the error argument ﬁrst. We are fallible creatures; we sometimes make mistakes. If, however, we are ever to be able to claim to know (that is, at least to be justiﬁed in believing) some proposition p, we must be able to exclude the possibility that at the time of claiming to know p we are in error. But since we typically, or at least frequently, are not aware of our errors as we make them, and might therefore unwittingly be in error as we claim to know p, we are not justiﬁed in making that claim. The same applies when a person is the subject of a delusion, illusion or hallucina- tion. Sometimes people undergoing one or other of these states do not know that they are doing so, and take themselves to be having veridical experiences. Clearly, although they think they are in a state which lends itself to their being justiﬁed in claiming to know p, they are not in such a state. Therefore they are not justiﬁed in claiming to know p. So in order for anyone to claim knowledge of some p, they must be able to exclude the possibility that they are the subject of such states. This pattern of argument is at its most familiar in the argument from dreaming employed by Descartes. One way of setting it out is as follows. When I sleep I sometimes dream, and when I dream I sometimes – indeed, often – do not know that I am dream- ing. So I can have experiences that appear to be veridical waking experiences on the basis of which I take myself to be justiﬁed in claiming to know such and such. But because I am dreaming, I do not in fact know such and such; I merely dream that I do. Might I not be dreaming now? If I cannot exclude the possibility that I am now, at this moment, dreaming, I am unable to claim knowledge of the things I at this moment take myself to know. For example, it seems to me that I am sitting at a desk next to a window admitting a view of trees and lawns. But because I might be dreaming that this is so, I cannot claim to know it. In these arguments the possibility of error, delusion or dreaming acts as what might	Blackwell
be called a ‘defeater’ to knowledge claims. The pattern is: if one knows p, then nothing is acting to subvert one’s justiﬁcation to claim knowledge of p. But one can seem to oneself fully entitled to claim to know some p, and in fact lack that entitlement, as the foregoing considerations show. So our claims to knowledge are in need of better grounds than we standardly take ourselves to have. We must ﬁnd a way of defeating the defeaters. 2.3 Perception Both rationalist and empiricist views about the sources of knowledge are threatened by the arguments just sketched. Arguments that pose particular problems for empiricism are suggested by the nature and limitations of perception, the best current account of which tells us something like the following story. Light reﬂects from the surfaces of objects in the physical environment and passes into the eyes, where it irritates the cells of the retinas in such a way as to trigger impulses in the optic nerves. The optic nerves convey these impulses to the region of the cerebral cortex that processes visual data, where they stimulate certain sorts 47 A. C. GRAYLING of activity. As a result, in ways still mysterious to science and philosophy, coloured ‘motion pictures’ arise in the subject’s consciousness, representing the world outside his or her head. This remarkable transaction is repeated mutatis mutandis in the other sensory modalities of hearing, smell, taste and touch, giving rise to perceptions of har- monies and melodies, perfumes and piquancies, smoothness, softness, warmth – and so forth. This model can be used to furnish another sceptical application of the defeater argu- ment. The complex causal story thus told is one which – so the sceptic can point out – might be interrupted in problematic ways at any point along its length. The experiences which we say result from the interaction of our senses and the world might occur in us for other reasons. They might occur when, as noted above, we dream, hallucinate or suffer delusions; or, to be fanciful, they might be produced in us by a god, or by a scientist who has connected our brains to a computer. From the point of view of the experiencing subject, there might be no way of telling the difference. So, says the sceptic, unless we ﬁnd means of excluding these possibilities, we are not entitled to claim knowledge of what we standardly take ourselves to know. 2.4 Perceptual relativities These same considerations about perception can prompt sceptical challenge by a dif- ferent route. A little reﬂection of the kind taught us by Locke, BERKELEY (1685–1753) (chapter 30) and other earlier contributors to the debate shows that some of these prop- erties we seem to perceive in objects are not ‘in the objects themselves’ but are in fact creatures of the perceptual relation. The qualities of objects – their colour, taste, smell, sound and texture – vary according to the condition of the perceiver or the conditions under which they are perceived. The standard examples are legion: grass is green in daylight, black at night; tepid water feels warm to a cold hand, cool to a hot hand; objects look large from close by, small from far away; and so on. These perceptual relativities are cited by the sceptic to raise questions not just about whether perception is a trustworthy source of information about the world, but whether the world can be said to exist independently of perception at all. For what if the properties by whose means we detect the presence of objects cannot be described apart from their being objects of perception? Consider the old conundrum whether a sound is made by the tree that falls in the forest when no sentient being is present to hear it do so. The answer, on a standard theory of perception current in contemporary science, is that the tree falls in complete silence. For if there is no ear to hear, there is no sound; there are only at best the conditions – vibrating airwaves – which would	Blackwell
cause sound to be heard if there were normally functioning eardrums, aural nerves and the rest to be stimulated by them. These considerations suggest a sceptical picture in which perceivers are in some- thing like the following predicament. Imagine a man wearing a visorless helmet which so encloses his head that he cannot see, hear, taste or smell anything outside it. Imagine that a camera, a microphone and other sensors are afﬁxed to the top of the helmet, transmitting pictures and other information to its interior. And suppose ﬁnally that it is impossible for the wearer to remove the helmet to compare this information with whatever is outside, so that he cannot check whether it faithfully represents the 48 EPISTEMOLOGY exterior world. Somehow the wearer has to rely on the intrinsic character of the information available inside the helmet to judge its reliability. He knows that the information sometimes comes from sources other than the exterior world, as in dreams and delusions; he has deduced that the equipment afﬁxed to the helmet works upon the incoming data and changes it, for example adding colours, scents and sounds to its picture of what intrinsically has none of these properties (at very least, in those forms); he knows that his beliefs about what lies outside the helmet rest on the inferences he draws from the information available inside it, and that his inferences are only as good as his fallible, error-prone capacities allow them to be. Given all this, asks the sceptic, have we not a job of work to do to justify our claims to knowledge? 2.5 Methodological and problematic scepticism Before considering these arguments and canvassing some ways of responding to them, it is important to note two things. One is that sceptical arguments are not best dealt with by attempts at piecemeal – that is, one-by-one – refutation. The second is that there is a vitally important distinction to be drawn between two ways in which scepticism can be employed in epistemology. It is important to note these matters because other- wise the prima facie implausibility of most sceptical arguments will mislead us into underestimating their signiﬁcance. I take each point in turn. Attempted refutation of sceptical arguments piecemeal is, arguably, futile for two good reasons. As suggested at the outset, sceptical arguments are at their strongest not when they seek to prove that we are ignorant about some subject matter but when they ask us to justify our knowledge claims. A challenge to justify is not a claim or a theory, and cannot be refuted; it can only be accepted or ignored. Since the sceptic offers reasons why justiﬁcation is needed, the response might be to inspect those reasons to see whether the challenge needs to be met. This indeed is one good response to scepti- cism. Where the reasons are cogent, the next good response is to try to meet the chal- lenge thus posed. The second reason is that sceptical arguments taken together have the joint effect of showing that there is work to be done if we are to get a satisfactory account of knowl- edge – and scepticism indicates what is needed. If one could refute, or show to be ungrounded, one or another individual sceptical argument, others would be left in place still demanding that such an account be sought. These points can be illustrated by considering Gilbert Ryle’s (1900–76) attempt to refute the argument from error by using a ‘polar concept’ argument. There cannot be counterfeit coins, Ryle observed, unless there are genuine ones, nor crooked paths unless there are straight paths, nor tall men unless there are short men. Many concepts come in such polarities, a feature of which is that one cannot grasp either pole unless one grasps its opposite at the same time. Now, ‘error’ and ‘getting it right’ are concep- tual polarities. If one understands the concept of error, one understands the concept	Blackwell
of getting it right. But to understand this latter concept is to be able to apply it. So our very grasp of the concept of error implies that we sometimes get things right. Ryle obviously assumed that the error sceptic is claiming that, for all we know, we might always be in error. Accordingly his argument – that if we understand the concept 49 A. C. GRAYLING of error, we must sometimes get things right – is aimed at refuting the intelligibility of claiming that we might always be wrong. But of course the error sceptic is not claim- ing this. He or she is simply asking how, given that we sometimes make mistakes, we can rule out the possibility of being in error on any given occasion of judgement – say, at this present moment. But the sceptic need not concede the more general claims that Ryle makes, namely, that for any conceptual polarity, both poles must be understood, and – further and even more tendentiously – to understand a concept is to know how to apply it, and for it to be applicable is for it actually to be applied (or to have been applied). This last move is question-begging enough, but so is the claim about conceptual polarities itself. For the sceptic can readily cite cases of conceptual polarities – ‘perfect–imperfect’, ‘mortal–immortal’, ‘FINITE–INFINITE’ (chapter 11) – where it is by no means clear that the more exotic poles apply to anything, or even that we really understand them. After all, taking a term and attaching a negative preﬁx to it does not guarantee that we have thereby grasped an intelligible concept. These comments suggest that sceptical arguments, even if singly they appear implausible, jointly invite a serious response; which is what, in large measure, episte- mology seeks to offer. But there is still the matter of the distinction between method- ological and problematic scepticism to be explained, and here a brief recapitulation of Descartes’s use of sceptical arguments will be helpful. Descartes’s Method of Doubt Descartes’s aim was to ﬁnd a basis for knowledge, which he did by looking for a starting point about which he could be certain. To ﬁnd certainty he needed to rule out anything that could be doubted, however absurd that doubt, for only in this way would we be left with what is truly indubitable. In the ﬁrst Meditation he embarks on this task by borrowing some sceptical arguments from the ancients. First he cites the fact that we can be misled in perception. But this is not a thoroughgoing enough scepticism, for even if we misperceive there is still much that we can know. So he next considers the possibility that on any occasion of claiming to know something, one might be dreaming. This sceptical thought catches more in its net, but is still insufﬁcient, for even in dreams we can know such things as, for example, mathematical truths. So, to get as sweeping a consideration as possible, Descartes introduces the ‘evil demon’ idea. Here the supposi- tion is that with respect to everything about which one could possibly be misled, an evil demon is indeed misleading one. Famously, what such a being cannot mislead one about is cogito ergo sum – when one thinks ‘I exist’, this proposition is true. It is essential to note that Descartes’s use of these arguments is purely methodological. The rest of the Meditations is devoted to showing that we know a great deal, because the fact (as Descartes unsuccessfully tries to prove) that there is a good DEITY (see chapter 15) guarantees that, just so long as we use our faculties responsibly, whatever is perceived with clearness and distinctness to be true will indeed be true. This is because a good deity, unlike an evil one, would not wish us to embrace ignorance. Descartes was by no means a sceptic, nor did he think that sceptical arguments, least of all the one employed as a device to set aside as many beliefs as possible, were persuasive. The ‘method of doubt’ is merely a tool. 50 EPISTEMOLOGY Descartes’ successors, however, were far more impressed by the sceptical arguments he	Blackwell
employed than his answer to them. For the tradition of epistemological thinking after his time, these and allied sceptical arguments were not mere methodological devices, but serious problems requiring solution. Hence the distinction I draw here between methodological and problematic scepticism. It is clear that there are sceptical considerations that have merely methodological utility, and are not genuinely problematic, because they do not represent a stable and cogent challenge to our ordinary epistemological standards. Descartes’s ‘evil demon’ is a case in point. Since the hypothesis that there is such a thing is as arbitrary and groundless as a hypothesis can get, it does not merit being taken seriously otherwise than as a ploy to make a point. But sceptical considerations about perception, error, delusion and dreams raise more interesting and troubling general issues, and accord- ingly merit examination. Among the many things worth noting about Descartes’s discussion are the follow- ing two. Firstly, as hinted earlier, his quest for certainty is arguably misconceived. Certainty is a psychological state one can be in independently of whether or not one believes truly. The falsity of a belief is no bar to one’s feeling certain that it is otherwise. Descartes sought to specify ways of recognizing which of our beliefs are true, but he led himself into talk of certainty because – and this is the second point – he assumed that epistemology’s task is to provide one with a way of knowing, from one’s own sub- jective viewpoint, when one possesses knowledge. Accordingly, he starts with the private data of a single consciousness and attempts to move outside it, seeking guar- antees for the process en route. Nearly all of Descartes’s successors in epistemology, up to and including RUSSELL (1872–1970) (chapter 37) and Ayer (1910–89), accepted this perspective on their task. In this respect at least they are all therefore Cartesians. It is largely for this reason, as we shall see hinted below, that they found it hard to meet scepticism’s challenge. 2.6 Some responses to scepticism The sceptical challenge tells us that we suffer an epistemic plight, namely, that we can have the best possible evidence for believing some p, and yet be wrong. Stated succinctly and formally, scepticism is the observation that there is nothing contradictory in the conjunction of statements s embodying our best grounds for a given belief p with the falsity of p. An informative representation of scepticism thus summarized is as follows. Scepti- cal arguments open a gap between, on the one hand, the grounds a putative knower has for some knowledge claim, and, on the other hand, the claim itself. Responses to scepticism generally take the form of attempts either to bridge this gap or to close it. The standard perceptual model, in which beliefs are formed by sensory interaction with the world, postulates a causal bridge across the gap; but that bridge is vulnerable to sceptical sabotage, so the causal story at least needs support. Descartes, as noted, iden- tiﬁed the epistemological task as the need to specify a guarantee – call it X – which, added to our subjective grounds for belief, protects them against scepticism and thus elevates belief into knowledge. His candidate for X was the goodness of a deity; reject- ing this candidate (while continuing to accept his view of the epistemological task) 51 A. C. GRAYLING obliges us to ﬁnd an alternative. If an X cannot be found to support a bridge across the sceptical gap, the option is to try closing it – or more accurately, to show that there is no gap at all. Both the quest for X and the closing of the gap have constituted major epistemological endeavours against scepticism in modern philosophy. Some of these endeavours, in brief, are as follows. Descartes’s immediate successors were, as mentioned, unpersuaded by his attempt to bridge the gap by invoking a good divinity to serve as X. LOCKE (chapter 29), without	Blackwell
much fanfare, employed a weaker version of the Cartesian expedient by saying that we can ignore sceptical threats to the causal story because ‘the light that is set up in us shines bright enough for all our purposes’. From Locke’s point of view it does not matter whether the inner light is set up by God or nature; the point is that there is something – X, the inner light which could be, perhaps, reason, empirical intuition or native trust in the reliability of the senses – that gives us grounds for accepting our ordinary knowledge-acquiring means as adequate. Others, not content with such unsatisfactory moves, look for X elsewhere, and claim to ﬁnd it in some version of foundationalism, the thought – sketched above – that our epistemic system has a basis in special beliefs that are in some way self-justifying or self- evident and which, in conjunction with the evidence we ordinarily employ in making knowledge claims, secures them against scepticism. As we saw earlier, a chief ground for rejecting such theories is alleged to be that none of them identiﬁes satisfactory can- didates for ‘foundations’. But one stimulating way of making something like a founda- tionalist case is offered by Kant, whose attempt prompted others. 2.7 Transcendental arguments KANT (1724–1804) (chapter 32) regarded failure to refute scepticism as a ‘scandal’ to philosophy, and offered his Critique of Pure Reason (1929) as a solution. His thesis is that our minds are so constituted that they impose a framework of interpretative concepts upon our sensory input, among them those of the objectivity and causal intercon- nectedness of what we perceive. Application of these concepts transforms mere passive receipt of sensory data into EXPERIENCE (pp. 726–33) properly so called. Our faculties are such that when raw data comes under the interpreting activity of our concepts, they have already had spatial and temporal form conferred on them by the nature of our sensory capacities; all our experience, considered as relating to what is outside us, is experience of a spatially structured world, and all our experience, considered as relat- ing to its received character in our minds, is of a temporally structured world. Upon the spatio-temporal data thus brought before our minds we impose the categories, that is, the concepts that make experience possible by giving it its determinate character. And here is Kant’s point: if the sceptic asks us to justify our claims to knowledge, we do so by setting out these facts about how experience is constituted. Kant claimed HUME (1711–76) (chapter 31) as his inspiration for these ideas, because Hume had argued that although we cannot refute scepticism – reason was not, he claimed, up to the task – we should not be troubled, for human nature is so consti- tuted that we simply cannot help having the beliefs that scepticism challenges us to justify. Those beliefs include, for example, that there is an external world, that causal relations hold between events in the world, that inductive reasoning is reliable, and so 52 EPISTEMOLOGY forth. From this hint Kant elaborated his theory that the concepts that the sceptic asks us to justify are constitutive features of our capacity to have any experience at all. The strategy, if not the details, of Kant’s attack upon scepticism has prompted inter- est in more recent philosophy. The argument he employs is a transcendental argument, brieﬂy characterizable as one which says that because A is a necessary condition for B, and, because B is the case, A must be the case also. An example of such an argument in action against scepticism is as follows. A typical sceptical challenge concerns belief in the unperceived continued existence of objects. What justiﬁes our holding this belief and premising so much upon it? The transcendental arguer answers that because we take ourselves to occupy a single uniﬁed world of spatio-temporal objects, and because on this view spatio-	Blackwell
temporal objects have to exist unperceived in order to constitute the realm as single and uniﬁed, a belief in their unperceived continued existence is a condition of our thinking both about the world and our experience of it in this way. Since we do indeed think that the sceptic asks us to justify is thereby justiﬁed. A this way, the belief contemporary thinker who makes notable use of this style of argument is P. F. Strawson (b. 1919). 2.8 Idealism and phenomenalism There is, in parallel to these Kantian ways of responding to the sceptical challenge, another approach, which denies the existence of a scepticism-generating gap. The chief ﬁgures in this camp are Berkeley and, more recently, the phenomenalists, who – allowing for differences among them, and for the fact that the two latter held these views only for part of their careers – include MILL (1806–73) (chapter 35), Russell and Ayer. In Berkeley’s view, scepticism arises from thinking that behind or beyond our sensory experiences there lies a material world. The word ‘material’ means ‘made of matter’, and ‘matter’ is a technical philosophical term supposed to denote an empirically unde- tectable substance believed by Berkeley’s philosophical predecessors to underpin the sensorily detectable properties of things, such as their colours, shapes and textures. Berkeley rejected the concept of matter thus understood – it is a common misreading of him to take it that he thereby denied the existence of physical objects; he did no such thing – arguing that because physical objects are collections of sensible qualities, and because sensible qualities are ideas, and because ideas can only exist if perceived, the existence of objects therefore consists in their being perceived; if not by ﬁnite minds such as our own, then everywhere and at all times by an inﬁnite mind. (We may note that Berkeley thought that his refutation of scepticism was at the same time a power- ful new argument for the existence of God.) Berkeley’s habit of saying that things exist ‘in the mind’ has led uncritical readers to suppose he means that objects exist only in one’s head, which is what a subjective idealist or solipsist might try to hold. Berkeley’s idealism, whether or not it is otherwise defensible, is at least not quite so unstable a view. His ‘in the mind’ should be read as meaning ‘with essential reference to experience or thought’. For present purposes, the point is that Berkeley sought to rebut scepticism by denying the existence of a gap between experience and reality, on the grounds that experience 53 A. C. GRAYLING and reality are the same thing. (He had a theory of how, despite this, we can never- theless imagine, dream and make mistakes.) The phenomenalists, with one very impor- tant difference, argued likewise. Their view, brieﬂy stated, is that all our beliefs about the world are derived from what appears to us in experience. When we analyse appear- ances – the ‘phenomena’ – we see that they are built out of the basic data of sense: the smallest visible colour patches in our visual ﬁelds, the least sounds in our auditory ﬁelds. Out of these sense-data we ‘logically construct’ the chairs and tables, rocks and mountains constituting the familiar furniture of the everyday world. An alternative but equivalent way of putting this point, the phenomenalists claim, is to say that statements about physical objects are merely convenient shorthand for longer and more complicated statements about how things seem to us in the usual employment of our sensory capacities. And to say that objects continue to exist unper- ceived is to say – in Mill’s phrase – that they are ‘permanent possibilities of sensation’, meaning that one would experience them if certain conditions were fulﬁlled. Berkeley holds that things remain in existence when not perceived by ﬁnite minds because they are perceived by a deity. The phenomenalists argue that what it means to	Blackwell
say that things exist unperceived is to say that certain counterfactual conditionals are true, namely, those asserting that the things in question would be perceived if some per- ceiver were suitably placed with respect to them. These conditionals are notoriously problematic, because it is not clear how to understand them. What, in particular, makes them true when they are (or seem quite obviously to be) true? The usual answers, in terms of possible worlds, laws, ideal regularities and similar exotica, do little to help. It is not clear that much of an advance is made over Berkeley’s ubiquitous deity by sub- stituting barely true counterfactuals in its place. Berkeley’s view has the modest attrac- tion that everything in the world is actual – anything that exists is perceived – whereas in the phenomenalist’s universe most of what exists does so as a possibility rather than an actuality, namely, as a possibility of perception. One thing is clear, at least: that one does not get phenomenalism simply by sub- tracting the theology from Berkeley’s theory. One has to do that and then, in the result- ing metaphysical gap, substitute a commitment to the existence of barely true counterfactuals, with an accompanying commitment to the existence of possibilia. Both Berkeley’s theory and phenomenalism thus exact high prices for closing the sceptical gap. 2.9 Sceptical epistemology versus anti-Cartesianism Some epistemologists do not attempt to refute scepticism for the good reason that they think it true or at least irrefutable. Their views might be summarized as stating that scepticism is the inevitable result of epistemological reﬂection, so we should accept either that we are only ever going to have imperfectly justiﬁed beliefs, always subject to revision in the light of experience, or that we have to recognize that scepticism, despite being irrefutable, is not a practical option, and therefore we have to live as most people anyway do, namely, by simply ignoring it. Some commentators on Hume interpret him as taking this latter view of the matter, and accordingly call it the ‘Humean’ response to scepticism. In Stroud (1984) and Strawson (1985), something like the Humean view is taken. 54 EPISTEMOLOGY Others in the recent debate are more combative, among them DEWEY (1859–1952) (see chapter 36) and WITTGENSTEIN (1889–1951) (chapter 39). Despite substantial differences in other respects, these two thinkers hold an interesting view in common, which is that scepticism results from accepting the Cartesian starting-point among the private data of individual consciousness. If instead, they say, we begin with the public world – with considerations relating to facts about the essentially public character of human thought and language – a different picture emerges. Dewey argued that the Cartesian model makes the epistemic subject a merely passive recipient of experiences, like someone sitting in the dark of a cinema watching the screen; but, he pointed out, ours is in fact a participant perspective – we are actors in the world, and our acquisition of knowledge is the result of our doings there. Wittgenstein contested the very coherence of the Cartesian approach by arguing that PRIVATE LANGUAGE (pp. 817–20) is impossible. A private language in Wittgenstein’s sense is one that is logically available only to one speaker, which is what a Cartesian subject would need in order to begin discoursing about his private inner experience. His argument is this: language is a rule-governed activity, and one only succeeds in speak- ing a language if one follows the rules for the use of its expressions. But a solitary would-be language-user would not be able to tell the difference between actually fol- lowing the rules and merely believing that he is doing so; so the language he speaks cannot be logically private to himself; it must be shareable with others. Indeed, Wittgen- stein argues that language can only be acquired in a public setting (he likens language-	Blackwell
learning to the training of animals; to learn a language is to imitate the linguistic behaviour of one’s teachers), which similarly weighs against the idea that the Cartesian project is even in principle possible. The anti-sceptical possibilities of the private language argument seem not to have been wholly apparent to Wittgenstein himself. In draft notes on scepticism and knowl- edge written in the last months of his life – later published under the title On Certainty (1969) – he offers a response to scepticism, which marks a return to a more traditional approach, not unlike that offered by Hume and Kant. It is that there are some things we have to accept in order to get on with our ordinary ways of thinking and speaking. Such propositions as that there is an external world, or that the world came into exis- tence a long time ago, are simply not open to doubt; it is not an option for us to ques- tion them. Nor therefore, says Wittgenstein, can we say that we know them, because knowledge and doubt are intimately related, in that there can only be knowledge where there can be doubt, and vice versa. The propositions we cannot doubt constitute the ‘scaffolding’ of our ordinary thought and talk, or – Wittgenstein varies his metaphors – they are like the bed and banks of a river, down which the stream of ordinary discourse ﬂows. In this sense the beliefs that scepticism attempts to challenge are not open to negotiation; which, says Wittgenstein, disposes of scepticism. These thoughts are as suggestive as they are in the philosophies of Hume and Kant; but one of the problems with Wittgenstein’s way of putting them is that he uses foun- dationalist concepts in describing the relation of ‘grammatical’ propositions to ordinary ones, but repudiates foundationalism as such, and seems to allow a version of rela- tivism by doing so – the river’s bed and banks, he says, might in time be worn away. But relativism is just scepticism in disguise – it is, indeed, arguably the most powerful and 55 A. C. GRAYLING troubling form of scepticism, for it is the view that knowledge and truth are relative to a point of view, a time, a place, a cultural or cognitive setting: and knowledge and truth thus understood are not knowledge and truth. Concluding Remarks There is much one would like to insist upon in trying correctly to describe the work that needs to be done in epistemology, for that is the necessary preliminary to making what progress we can. Here I shall simply underline a couple of remarks already made above. Firstly, debates over the deﬁnition of ‘knowledge’ seem to me to be a side-show. The justiﬁcation of claims in the natural sciences, the social sciences (not least history) and law is where the real work cries out to be done in epistemology. And this comment applies only to the empirical case: what of the epistemological questions that press in ETHICS (chapter 6) and the PHILOSOPHY OF MATHEMATICS (chapter 11)? There can be no guarantee – and indeed it is unlikely – that high generalities about justiﬁcation and knowledge will apply univocally across all these ﬁelds. ‘Justiﬁcation’ is a dummy concept that needs to be cashed out in terms particular to particular ﬁelds; so much should be obvious from the fact that unrestrictedly general accounts of justiﬁcation prove hopelessly vulnerable to counter-example. Secondly, little in current literature about scepticism makes one conﬁdent that its nature is properly understood. Scepticism deﬁnes one of the central problems in epistemology, namely, the need to show how justiﬁcation of belief is possible. This is done by meeting the challenge to show that sceptical considerations do not after all defeat our best epistemic endeavours in this or that speciﬁed ﬁeld. Implicit in this characterization are two important claims: ﬁrstly, that scepticism is best understood as a challenge, not as a claim that we do or can know nothing; and secondly, that the best	Blackwell
way to respond to scepticism is not by attempting to refute it on an argument-by- argument basis, but by showing how we come by justiﬁcation for what we believe. Somehow these two points, which were obvious to our predecessors, seem to have been lost to sight. Further Reading General Some useful texts are R. Nozick (1981); J. Dancy and E. Sosa (1992); B. Williams (1978); K. Lehrer (1974); L. BonJour (1985); P. F. Strawson (1985); and G. Pappas and M. Swain (1978). The classic texts in epistemology include Plato’s Meno and Theaetetus, Descartes’s Meditations, Locke’s Essay Concerning Human Understanding, Berkeley’s Principles of Human Knowledge and Three Dialogues Between Hylas and Philonous, Hume’s Treatise of Human Nature and Enquiry Concerning Human Understanding, and Kant’s Critique of Pure Reason. One of the best short elementary books remains Russell’s The Problems of Philosophy (1912, much reprinted). 56 EPISTEMOLOGY Knowledge The debate about knowledge and justiﬁcation commands a large literature, of which the follow- ing are good examples: W. Alston (1983); L. BonJour (1985); A. Brueckner (1988); R. Chisholm (1977); J. Dancy (1985); F. Dretske (1971); R. Feldman (1985); E. Gettier (1963); A. Goldman (1979, 1980, 1986); G. Harman (1973, 1984); K. Lehrer (1974); P. Moser (1985); R. Nozick (1981); J. Pollock (1979, 1984, 1986); R. Shope (1983); E. Sosa (1981). Scepticism The best general introduction to sceptical arguments remains Bertrand Russell’s The Problems of Philosophy, but it is essential to see the arguments in a classic setting, and for this one must read René Descartes’s Meditations on First Philosophy (translated by J. Cottingham, 1986), especially the First Meditation. Useful discussions of sceptical arguments and the sense-datum theory are to be found in A. J. Ayer (1956) and J. L. Austin (1961), although one should also look at Ayer’s reply (1967). For discussion of perception see J. Dancy (1988); T. Crane (1992); R. Swartz (1965); F. Jackson (1977); and M. Perkins (1983). For an attempt at being sceptical see P. Unger (1975). For responses to scepticism inﬂuenced by Kant see P. F. Strawson (1959, 1985) and A. C. Grayling (1985). Allied lines of thought occur in G. E. Moore (1959) and L. Wittgenstein (1969). More recent discussions are B. Stroud (1984) and M. Williams (1991). Scepticism, foundationalism and coherence theories of knowledge are discussed in K. Lehrer (1974) and in useful papers collected by G. S. Pappas and M. Swain (1978) and M. Clay and K. Lehrer (1989). For a discussion of the views variously taken by Dewey and Wittgenstein see R. Rorty (1979). A textbook which surveys the ﬁeld and provides a useful bib- liography is J. Dancy (1985). For the history of scepticism see M. Burnyeat (1983) and R. Popkin (1979). References Classical texts Berkeley, G. 1995 [1710]: The Principles of Human Knowledge (edited by H. Robinson). Published with Three Dialogues. Oxford: Oxford University Press. —— 1995 [1713]: Three Dialogues Between Hylas and Philonous (edited by H. Robinson). Published with The Principles of Human Knowledge. Oxford: Oxford University Press. Descartes, R. 1986 [1641]: Meditations on First Philosophy (translated by J. Cottingham). Cambridge: Cambridge University Press. Hume, D. 1978 [1739–40]: A Treatise of Human Nature (edited by P. H. Nidditch). Oxford: Clarendon Press. —— 1975 [1748]: Enquiry Concerning Human Understanding (edited by P. H. Nidditch). Oxford: Clarendon Press. Kant, I. 1929 [1781 and 1787]: Critique of Pure Reason (1st edn 1781, 2nd edn 1787) (trans- lated by N. K. Smith). London: Macmillan. Locke, J. 1975 [1690]: An Essay Concerning Human Understanding (edited by P. H. Nidditch). Oxford: Clarendon Press. Plato 1961: Meno and Theaetetus. In The Collected Dialogues of Plato (edited by E. Hamilton and H. Cairns). Princeton, NJ: Princeton University Press. Other writings Alston, W. 1983: What is Wrong with Immediate Knowledge? Synthèse, 55. 57	Blackwell
A. C. GRAYLING Austin, J. L. 1961: Sense and Sensibilia. Oxford: Oxford University Press. Ayer, A. J. 1956: The Problem of Knowledge. London: Macmillan. —— 1967: Has Austin Refuted the Sense-Datum Theory? Synthèse, 17. BonJour, L. 1985: The Structure of Empirical Knowledge. Cambridge, MA: Harvard University Press. Brueckner, A. 1988: Problems with Internalist Coherentism. Philosophical Studies, 54. Burnyeat, M. (ed.) 1983: The Sceptical Tradition. Berkeley: University of California Press. Chisholm, R. 1977: Theory of Knowledge. Englewood Cliffs, NJ: Prentice-Hall. Clay, M. and Lehrer, K. (eds) 1989: Knowledge and Scepticism. Boulder, CO: Westview Press. Crane, T. (ed.) 1992: The Contents of Experience. Cambridge: Cambridge University Press. Dancy, J. 1985: Introduction to Contemporary Epistemology. Oxford: Blackwell. —— (ed.) 1988: Perceptual Knowledge. Oxford: Oxford University Press. Dancy, J. and Sosa, E. (eds) 1992: A Companion to Epistemology. Oxford: Blackwell. Dretske, F. 1971: Conclusive Reasons. Australasian Journal of Philosophy, 49. Feldman, R. 1985: Reliability and Justiﬁcation. The Monist, 68. Gettier, E. 1963: Is Justiﬁed True Belief Knowledge? Analysis, 23. Goldman, A. 1979: What is Justiﬁed Belief? In G. Pappas (ed.) Justiﬁcation and Knowledge. Dordrecht: Reidel. —— 1980: The Internalist Conception of Justiﬁcation. In Midwest Studies in Philosophy 5, ed. P. A. French et al. Minneapolis: University of Minnesota Press. —— 1986: Epistemology and Cognition. Cambridge, MA: Harvard University Press. Grayling, A. C. 1985: The Refutation of Scepticism. London: Duckworth. Harman, G. 1973: Thought. Princeton, NJ: Princeton University Press. —— 1984: Positive versus Negative Undermining in Belief Revision. Noûs, 18. Jackson, F. 1977: Perception. Cambridge: Cambridge University Press. Lehrer, K. 1974: Knowledge. Oxford: Clarendon Press. Moore, G. E. 1959: Philosophical Papers. London: Allen and Unwin. Moser, P. 1985: Empirical Justiﬁcation. Dordrecht: Reidel. Nozick, R. 1981: Philosophical Explanations. Cambridge, MA: Harvard University Press. Pappas, G. and Swain, M. (eds) 1978: Knowledge and Justiﬁcation. Ithaca, NY: Cornell University Press. Perkins, M. 1983: Sensing the World. Indianapolis: Hackett Publishing. Pollock, J. 1979: A Plethora of Epistemological Theories. In G. Pappas (ed.) Justiﬁcation and Knowl- edge. Dordrecht: Reidel. —— 1984: Reliability and Justiﬁed Belief. Canadian Journal of Philosophy, 14. —— 1986: Contemporary Theories of Knowledge. Totowa, NJ: Rowman and Littleﬁeld. Popkin, R. 1979: The History of Scepticism. Berkeley: University of California Press. Rorty, R. 1979: Philosophy and the Mirror of Nature. Princeton, NJ: Princeton University Press. Russell, B. 1910: Philosophical Essays. New York: Longman, Green. —— 1912: The Problems of Philosophy. London: Oxford University Press. Shope, R. 1983: The Analysis of Knowing. Princeton, NJ: Princeton University Press. Sosa, E. 1981: The Raft and the Pyramid: Coherence versus Foundations in the Theory of Knowl- edge. In Midwest Studies in Philosophy 5, ed. P. A. French et al. Minneapolis: University of Minnesota Press. Strawson, P. F. 1959: Individuals. London: Methuen. —— 1985: Scepticism and Naturalism: Some Varieties. London: Methuen. Stroud, B. 1984: The Signiﬁcance of Philosophical Scepticism. Oxford: Clarendon Press. Swartz, R. (ed.) 1965: Perceiving, Sensing and Knowing. Berkeley: University of California Press. 58 EPISTEMOLOGY Unger, P. 1975: Ignorance. Oxford: Clarendon Press. Williams, B. 1978: Descartes: The Project of Pure Enquiry. Harmondsworth: Penguin Books. Williams, M. 1991: Unnatural Doubts. Oxford: Blackwell. Wittgenstein. L. 1969: On Certainty. Oxford: Blackwell. Discussion Questions 1 How important is it to have a deﬁnition of knowledge? 2 If a proposition is false, can one be justiﬁed in believing it? 3 Why do ‘Gettier examples’ raise difﬁculties for a fallibilist account of knowledge? 4 Could states other than beliefs provide the foundations of knowledge?	Blackwell
5 If there are foundational beliefs, how are they related to dependent beliefs? 6 Is a belief justiﬁed if it coheres with an already accepted set of beliefs? Do we have an adequate account of the notion of coherence? 7 Does justiﬁcation consist in internal relations among beliefs? 8 Can a belief be justiﬁed for someone who does not have cognitive access to what justiﬁes the belief? 9 Is knowledge ‘appropriately caused true belief ’? 10 Does the notion of ‘tracking’ help to explain how beliefs are justiﬁed? 11 If there is no certainty, can there be knowledge? 12 Should we seek a single account of justiﬁed belief or different accounts tailored to the different areas in which epistemological questions may be asked? 13 What role does scepticism play in philosophy? 14 How can we disentangle those methods of enquiry that might yield genuine knowledge from those that can only deepen our ignorance? 15 Our knowledge claims are sometimes in error without our knowing it. Does this undermine justiﬁcation for any such claims? 16 Could you be dreaming now? 17 Could the experiences I take to be of the world have some other origin, without my being able to tell that this is so? 18 Can the properties by which we detect the presence of objects be described apart from their being objects of perception? 19 Are we like the man in a visorless helmet who cannot check the information trans- mitted to him to see whether it faithfully represents the external world? 20 Can sceptical arguments that are singly implausible jointly require a serious response? 21 What is the importance of distinguishing methodological and problematic scepticism? 22 Can epistemology provide a way of knowing from a ﬁrst-person subjective viewpoint? Is there any other viewpoint available? 23 If scepticism opens a gap between the grounds for a knowledge claim and the claim itself, is it better to bridge the gap or to close it? 24 Could the possibility of our having experience be unintelligible to us unless we held a certain belief, and yet that belief be false? 25 Are experience and reality the same thing? 26 Can we accept the role of counterfactuals in a phenomenalist account of physi- cal objects? 59 A. C. GRAYLING 27 What follows from beginning our account of knowledge with the public world rather than with the private data of individual consciousness? 28 Does the ‘private language argument’ show that the Cartesian project is impossible? 29 If there are propositions that are simply not open to doubt, how can we identify them? How are they related to propositions that we can doubt? 60 2 Metaphysics S I M O N B L AC K B U R N With a section on Time by Robin Le Poidevin Metaphysics is the exploration of the most general features of the world. We conceive of the world about us in various highly general ways. It is orderly, and structured in space and time; it contains matter and minds, things and properties of things, necessity, events, causation, creation, change, values, facts and states of affairs. Metaphysics seeks to understand these features of the world better. It aims at a large-scale investigation of the way things hang together. Within this broad description there are two conceptions of the subject. Metaphysicians may think of themselves as investigating the facts, or discovering the broad structures of reality. Or, they may see the enterprise as more self-reﬂective, gaining an understanding of how we represent the facts to ourselves: how our ‘conceptual scheme’ or perhaps any possible conceptual scheme, structures our own thought about reality. Once this description is completed, it may be that everything pos- sible has been done, for we have no alternative but to continue to think from within the conceptual scheme whose features we have mapped. 1 Metaphysics and its Doubters An inﬂuential distinction in the self-image of metaphysics is due to Strawson (1959). Metaphysics may be a purely descriptive enterprise. Or, it may be that there is reason	Blackwell
for revision; the ways we think about things do not hang together, and some categories are more trustworthy than others. Revisionary metaphysics then seeks to change our ways of thought in directions it ﬁnds necessary. The distinction between revisionary and descriptive metaphysics is not sharp, for it is out of the descriptions that the need for revision allegedly arises, and in fact the metaphysicians Strawson cites as revi- sionary – PLATO (c.427–347) (chapter 23), DESCARTES (1596–1650) (chapter 26), BERKELEY (1685–1753) (chapter 30) – believed themselves to be discovering no more than things that are implications of our necessary ways of thinking. Revisionary meta- physics is frequently associated with ambitious, and sometimes wild, philosophical speculation; descriptive metaphysics is intellectually more conservative. But metaphysics SIMON BLACKBURN characteristically has a practical dimension; since Plato it has been concerned to promote views about the nature of human beings, their relations with nature, or GOD (chapter 15), or with the larger SOCIAL WHOLES (p. 383) of which they are a part. Here it may be that descriptive metaphysics is ethically and politically more radical, for a pre- ferred description of human life may have far-reaching implications, and indeed it is not possible for any serious ETHICS (chapter 6) or POLITICAL PHILOSOPHY (chapter 8) to be entirely silent about metaphysics. And radical views may have conservative impli- cations, as when postmodernist views of the relative and contingent nature of all our judgements lead to failures of conviction, and the retreat of the intellectual into social and political inertia (Rorty 1989). Sometimes these themes have been presented as general topics of Being, and some authors give the impression that metaphysics is the science of Being. This is, however, misleading, for there may be nothing or little to be said about Being as such, even if there is a lot to say about the kinds of things that exist, and the categories under which they fall. And in fact the study of Being rapidly turns into the study of things in these categories, and of the relationship with them that characterizes particular thinkers at particular times. But what is right in the idea that metaphysics is the science of Being is that the most abstract study in this abstract discipline concerns the broad nature of reality, and the possibility of its objective representation. What motivates the study of metaphysics? Centrally, problems arise when we cannot see how things hang together. We suffer from disquiet with the plurality of different kinds of things that exist, or with the mixture of elements that we want to keep together. Paradigmatic metaphysical problems arise when broad areas of our commitments clash, as for instance when we think of ourselves as complex natural organisms on the one hand, or as conscious purposive or even free agents on the other, or when we think of TIME (pp. 82–5) as ﬂowing on the one hand, but recognize that it makes no sense to ask how fast it ﬂows, on the other. Consider the list of minds, physical things, abstract objects, values, events, processes, dispositions, necessities, states of affairs, properties, facts and other basic categories that we seem to recognize. We need a story about how they relate to each other: are some of them equivalent to others? Are some made up of others? Are some of them redundant, or others suspicious in other ways? And just as notably, we need a story about how we relate to them. How do we know about them, and why are they important to us? Such vague disquiet is typically focused by the dis- covery of paradox and inconsistency, apparently showing that our conceptions are inadequate. Among the earliest examples of philosophical argument in the Western tradition are the paradoxes of motion of Zeno of Elea (ﬁfth century BC), purporting to show that everyday thought about the ﬂow of time and motion in space involves	Blackwell
contradiction. What we all naturally believe cannot possibly be true. The moral Zeno wanted us to draw from this discovery is not quite clear, but the paradox served until the nineteenth century as a spur to philosophers and mathematicians to ﬁnd a better way of conceptualizing change in time. It seems that such topics should permit of some disciplined, intelligent, investigation. But it is controversial how such an investigation should be conducted. If metaphysics purports to tell us what things exist, or even just how to think about what things exist, then it seems to be trespassing on the domain of the PHYSICAL SCIENCES (chapter 9), and it is unclear how a philosophical study can have anything signiﬁcant to add to their 62 METAPHYSICS results. The question is whether there is anything left for philosophy to do, once fun- damental science has told us what it enables us to know about these topics. The view that there is something further, and that by some process of rational thought we can obtain reliable views about the nature of space, time, mind, causation and the rest is often described as belief in a ‘ﬁrst philosophy’, or philosophy as a discipline with its own methods and results, and the very possibility of such a ﬁrst philosophy has been a con- stant object of doubt. Metaphysics is thus preoccupied with its own possibility. It is not unique among philosophical topics in this. But whereas we believe in our hearts, as it were, that there must be such a subject as ethics, or that there must be interesting things to be said within the THEORY OF MEANING (see chapter 3) or even mind, there is no such presumption that there could be a disciplined method for achieving results when the topics are the abstract categories I listed. Consider as an example the problem of relating MIND AND MATTER (see chapter 5). Among the ﬁrst philosophical thoughts many people have are ones about their own conscious lives, and the gap there seems to be between that consciousness and the similar lives of others, and the mystery of that consciousness arising at all in the world as we otherwise understand it to be. The fact of consciousness seems undeniable, but it also seems mysterious how it can exist in a physical world of the kind to which we know ourselves to belong. We can think of this as a problem of relating a puzzling or exotic fact or kind of entity to a relatively domestic or familiar class of facts or entities. By this I mean that we may begin by being happy with some kinds of fact, made to obtain by some kinds of thing, such as the facts of physical nature, or facts about the patterns into which events fall. We feel we understand facts of this sort. We understand what makes them obtain, and how we know them to be true. This class is that of the familiar (homely) aspects of the world: ones that do not puzzle us unduly. But then we realize that we have other beliefs as well; ones which on the face of it relate to different kinds of fact, and posit different kinds of thing. In this case minds are posited, but we shall shortly meet other examples. We may not be able to understand how mind is even possible in a physical universe (how minds can relate to bodies; how they can make physical things happen). Then the main motive to metaphysical inquiry is to put the exotic class into some sort of intelligible relation to the familiar class. An intelligible relation will mean that we understand how both classes of facts exist, and how the world is constituted so that each of them obtains, how our everyday thoughts about their relationship can be true. If we think of it this way, we can see three main points at which choice is possible in metaphysics. First, there is the question of the authority of the familiar class. At a par- ticular period of time some starting points may seem natural. We may be at home, for example, with the results and concepts of the physical sciences (and in fact the convic-	Blackwell
tion that the world is fundamentally nothing but a physical world is certainly the most common view among contemporary Western philosophers). If so, these results and concepts give us our class of familiar facts, and the task is that of relating more exotic subjects to them. But to other philosophers there may be something wrong-headed or arbitrary or prejudiced about choosing just one privileged familiar class. Perhaps, for example, when we understand how complex and strange are the facts and states of affairs posited by contemporary physics, we will lose any conﬁdence that others are more exotic by comparison. Even deciding what facts and states of affairs belong to 63 SIMON BLACKBURN physics occupies a great deal of PHILOSOPHY OF SCIENCE (chapter 9): does physical theory, for example, postulate real PROBABILITIES AND CHANCES (pp. 308–11), and if so how are we to think of their existence? What justiﬁes our tendency to take science as an objec- tive description of independent reality, as opposed to seeing it as an INSTRUMENT (pp. 294–5) for prediction and control of empirical fact (and how is this distinction to be drawn)? But even when a privileged class has been identiﬁed and doubts about its nature and the source of its privilege satisfactorily settled, there will be a second question of what counts as a satisfactory ﬁt between the exotic and the familiar. What kinds of rela- tionship should put worries about mind or values or abstract objects to rest, if we wish to see our world as fundamentally the world as described by physics? What accommo- dations should we be looking for? Thirdly, and ﬁnally, there will be different attitudes to the exotic class. If it seems sufﬁciently exotic, so that its objects and facts seem only dimly related to those of the familiar class, a radical response will be eliminativism, or the sug- gestion that it is not possible that there should exist things in the exotic categories, so that the entire exotic area is best forgotten. The task is not to relate its commitments to the familiar class, but to teach people to do without them, rather as atheists do not seek to understand talk about God in some preferred terms, but want instead to abandon it altogether. In this spirit one might want to eliminate reference to abstract objects such as properties or facts, or to values or even to minds. Put like this the enterprise of metaphysics sounds eminently reasonable, and one can imagine well-conducted dispute in all three areas: motives for selecting a familiar class, differing views about the relationship it bears to the exotics, and different attitudes to the exotics. But the problem of a method for conducting such enquiry still remains, and we should recognize that philosophy has contained nearly as many thinkers who count themselves as radically opposed to metaphysics as ones who accept the title. In the modern (post-seventeenth century) Western tradition the ﬁrst philosopher resolutely hostile to metaphysics was David HUME (1711–76) (chapter 31) (Hume 1978). Fol- lowing him there were always philosophers, especially in Britain, of a more cautious empiricist bent, but few were tough-minded enough to believe with Hume that books of metaphysics should be consigned to the ﬂames, as containing nothing but sophistry and illusion. Whether or not it has a method, metaphysics is swayed by larger winds that blow in the prevailing culture. It is widely realized that before the Enlightenment the prevail- ing tone of philosophy was THEOLOGICAL (chapters 15 and 24). It was supposed that the world was the production of an intelligent, rational, caring and perfect being, who pos- sessed a complete understanding of its nature. We imperfect beings could do something to approximate to this understanding, but the insight we obtained in such areas as mathematics and logic afforded us only glimpses of the perfect rational understanding of the whole order of nature that God would possess (Craig 1987). The familiar class	Blackwell
of facts with which people were relatively content included God’s nature, purposes, values and relationship to creation. The abandonment of this picture of the world by Hume and other Enlightenment ﬁgures did not, however, mark the end of metaphysics. On the contrary, the scientiﬁc revolution of the seventeenth century replaced one kind of familiar fact with others: notably, those of physical science and those describing the course of perceptual experience. But this ‘Galilean worldview’, named after Galileo (1564–1642), contained the seeds of its own metaphysical thickets, namely those sur- 64 METAPHYSICS rounding the issue of idealism. When mind is sharply distinguished from matter, as it became in this worldview, the uncomfortable and apparently impassable gulf between them is most naturally closed either by making the mind material, or by making the material mental. Idealism is the umbrella title for philosophies that take the latter course. For the idealist, facts about mental life form the familiar class, and ones about the physical world need some kind of certiﬁcation from them. An idealist, therefore, would be receptive to the kind of instrumental interpretation of physical science men- tioned above, in which scientiﬁc descriptions are taken as instruments for prediction and control of the course of perceptual experience. Whichever priority is assumed, the problem of providing this certiﬁcation dominated philosophy from the time of Descartes until the present. Indeed, perhaps the high-water mark of metaphysical speculation was reached with the attempts to marry idealism with a religious worldview in the nineteenth century. According to the accepted story, under the inﬂuence of KANT (1724–1804) (chapter 32), but forgetting the tightly critical boundaries that Kant himself put upon metaphysics, it was believed that transcendent results – which go beyond the limits of experience and concern God, freedom, immortality and above all the ultimate spiritual nature of the universe – could be established by a variety of a priori arguments. Kant’s mistrust of transcendent reasoning was brushed aside by Fichte (1762–1814) and most inﬂuentially by HEGEL (1770–1831) (chapter 33), and the metaphysician became the specialist in the nature of THE ABSOLUTE (p. 743), or underlying ground of the cosmic order, which was identiﬁed variously with God, pure Freedom, or ﬁnal Self-consciousness, and which provided some sort of goal to the cumulative historical process. An almost religious belief in progress, coupled with a genuinely religious emphasis on the nature of the SPIRIT (p. 742), and above all its elevated moral tone, gave absolute idealism its hold on the minds of Europe. The dominant contemporary spirit reverses this direction, privileging facts about the physical and seeking to understand statements about mind and consciousness in its terms. This is known as physicalism, or less often materialism (the word physicalism is preferred because physics itself asserts that not everything that exists is material; the world includes such items as forces and ﬁelds). Physicalism and idealism share the goal of relating mind and matter in some intelligible way, but differ over what is familiar and what exotic. The need for additional tasks of the same kind may be less obvious in advance of critical reﬂection. The point to remember is that whatever a philosopher might put forward by way of doubts about the possibility of metaphysics, the philo- sophical need for a theory relating central kinds of fact may be driven by wider aspects of the world-picture characteristic of a given time. Hume was an opponent of metaphysics even though he bequeathed to his succes- sors the paradigm metaphysical problems of modern philosophy. His own attitude was simply that such problems were forever insoluble. The difﬁculty is that it is hard to	Blackwell
believe that metaphysical questions are unanswerable until we try to answer them. Perhaps the problems seem formidable because we are prejudiced in our choice of a familiar class (Hume himself favoured facts about the successions of our own ideas or ‘perceptions’ as especially basic or familiar, a choice that frequently amazes contempo- rary thinkers); perhaps we are blind to subtle and insightful ways of relating the exotics to it; perhaps we do not realize that we ought to get rid of exotics that refuse to be accommodated. It is difﬁcult to believe that the entire ﬁeld can be seen, in advance and 65 SIMON BLACKBURN from the armchair, to be a no-go area. But there has been one school this century which believed exactly that. Perhaps the one thing that is best known about the school known as the Vienna Circle, or the logical positivists, was their resolute hostility to metaphysics. It is instruc- tive to see how this hostility ﬂourished, but then eventually crumbled. The positivists were inﬂuenced by the empiricism of Hume, and also by the Tractatus Logico– Philosophicus of WITTGENSTEIN (1889–1951) (chapter 39) (Wittgenstein 1922), a work which purports to set bounds to the limits of meaningful language. Metaphysics lies outside the bounds, representing language that has gone on holiday. The positivists held that all true thought was empirical; anything straying beyond the use of scientiﬁc method also strayed beyond the boundaries of meaning. But metaphysical theses are not properly empirical; they seemed typically to be argued about from the armchair and to bear no visible relationship to scientiﬁc thought or experiment. Hence they do not admit of veriﬁcation or falsiﬁcation, and are not only unscientiﬁc, but also strictly meaningless (they may be allowed some kind of emotional effect, but nothing that permits of assessment as true or false). Instead of metaphysics, all that was left to phi- losophy was describing the correct methods and structures of empirical science. Positivism failed in its crusade against metaphysics, largely because of its own insta- bility. For while in one breath it was proclaiming the subject dead, in another it was itself making remarkably conﬁdent choices in all three of the areas in which meta- physics characteristically makes choices. Positivism needed a familiar class of empiri- cally sound and basic judgements. It needed views about what counts as a satisfactory accommodation with that class, notably so that the theories of physical science were allowed to be respectable, whereas others were not. And, ﬁnally, it had severe views about the areas that were not, by these lights, respectable. Metaphysics soon revenged itself, for all three areas gave trouble, and in all three the movement found itself pulled in different directions, making different choices and indulging in classically metaphysi- cal dispute. Aware of this, later positivist writings grudgingly allowed a place for what looked like metaphysical assertions as recommendations about which LANGUAGE GAMES (p. 9) to play. Thus, the doctrine that there exists mind as well as matter would be con- strued not as an important description of a fundamental fact about reality, but as a rec- ommendation to speak in terms of minds as well as in terms of matter (to play the mental language game). The recommendation might be useful or not, but could not be regarded as true or false. A similar attitude is characteristic of the later work of Wittgenstein (1953), which is shot through with warning that statements that might seem to be certain kinds of description of reality in fact function in different ways. Wittgenstein’s exact intentions are endlessly disputed. But, at least as applied to minds and matter, any pragmatism in the approach is itself acutely uncomfortable. Consider, for instance, the simple proposition that other people besides myself are conscious in	Blackwell
the way that I am. This seems to be a metaphysical thesis, although classifying it as one may itself represent a philosophical choice. It is certainly one that we all believe. But it is extremely strange to suppose that the question whether other people are conscious could be construed as the question of whether it was useful for me to enter the lan- guage game of reacting to them as having minds (this gloss on the question had in fact been put by William JAMES (1842–1910) (chapter 36) much earlier, especially in Prag- matism, and has not received a generally warm welcome). We should also notice that 66 even if we allow the idea that what sound like descriptions of fundamental reality are construed as recommendations about language, there is still argument about what to select as familiar, what is exotic, what should be eliminated and what the relationships are between all these classes. So even the pragmatist gloss may make less difference than might appear at ﬁrst sight. METAPHYSICS Absolute Presuppositions Their own metaphysical prejudices were invisible to the positivists. The thought that this is an inevitable feature of the philosophical condition is another, different, reason for despair about the possibility of metaphysics. This reason was voiced most notably by R. G. COLLINGWOOD (1889–1943) (pp. 436–9) (especially in An Essay in Metaphysics, 1940), and marked a point at which his thought comes surprisingly close to that of both the earlier and later philosophy of Wittgenstein. In this conception, there are commitments that we must have that are not capable of being assessed as true or false, but that never- theless have the greatest intellectual importance. In Wittgenstein’s metaphors, they func- tion as the riverbed within which thought ﬂows, or as the hinges on which ordinary judgement and discourse turn; Collingwood called them the absolute presuppositions of the thought of a time. Because they are presupposed in every activity of thought, they cannot themselves be assessed for truth: in another metaphor, they can be shown but not said. For Collingwood, indeed, they could only be shown historically. That is, at the end of an epoch it would be possible for successors to look back and to ﬁnd that a particular meta- physics had structured its thought. But at the time the basic structures would themselves be invisible, because they were themselves involved in all seeing (Hegel himself may have thought something similar; the famous remark from the preface to his Philosophy of Right (Hegel 1991) that ‘when philosophy paints its grey on grey, then has a form of life grown old. The owl of Minerva takes wing only with the coming of the dusk’ can be seen as the claim that philosophical reﬂection can only exist with hindsight). If this is right, the trap that caught the positivists is therefore universal; however much we aspire to care and objectivity in choosing our familiar facts, and a way of relating exotics to them, we will be working within an historically contingent framework whose main structures will be beyond our own vision, and incapable of our own assessment. Collingwood was an histo- rian, and other episodes in the history of philosophy support his view, for it is easier to see absolute presuppositions at work in the thought of particular writers (and their contem- poraries) when signiﬁcant time has passed, and easy to fear that our successors will marvel at similar blindness in our own appreciation of the way we think. An example I have mentioned already is Hume’s choice of the sequence of perceptions as the basic familiar fact: a choice that seems extraordinarily wrong-headed to nearly all contempo- rary philosophers. Once more, however, many philosophers are reluctant to admit there should exist any aspects of our own thought that cannot themselves be identiﬁed and even critically assessed. While such episodes in the history of philosophy may nourish our modesty, there seems to be no alternative to continuing to ‘work from within’, or in other	Blackwell
words to do the best we can from within our own best understandings. The thought that one day these may come to look local and parochial is like the similar thought, in moral philosophy, that there might one day be a people to whom our own best judgements seemed inadequate. Perhaps there might, but while we cannot see our way to this improvement, we can, and must, simply soldier on as best we can. 67 SIMON BLACKBURN There surfaces here a debate that has preoccupied many recent writers. This is the dispute between realism and its opponents. For the realist it is important that there is no residual reference to us (our language, our sensibilities, our conceptual scheme) when we consider the world. The realist wants real objectivity – a world of facts that are frequently entirely independent of us and that would be as they are whatever our powers of detecting and exploring them. The realist is therefore fundamentally opposed to the view that what we understand and investigate is as much a function of our constitution and our ways of thought as it is of anything independent of us. In a common metaphor, the realist believes that a good conceptual scheme ‘carves reality at the joints’; nature is conceived as possessing its own structure and articulation, and good theory only reﬂects this. This carnivorous metaphor outrages idealists of many varieties, and even philosophers who would scarcely count themselves as idealists. The central problem is that belief that we succeed in carving nature in some especially appropriate way seems to require a way of comparing our own classiﬁcations and the kinds of concept we use, with a pre-existent, naturally ordained structure of properties. But no such comparison can be made, nor is it plain what, apart from ordinary scientiﬁc utility, could justify us in supposing that some properties are intrinsically natural whereas others are not. This problem received forceful expression in Goodman (1955); for an example of industrial-strength realism, see Lewis (1983); for mistrust of the metaphor see Taylor (1993). The choice between realism and its opponents has echoes in many areas. For example, the realist is particularly apt to privilege some familiar class, usually that of physical theory, as being especially well-adapted to nature’s joints. By comparison other commitments may seem to have less to do with the real way that things are. For instance, the sizes and shapes of things around us may seem to be more objective, more independent of us, than their colours, which seem to be largely a function of the nature of our visual systems. Idealists, impressed by the mind’s contribution to any scheme of thought, including that of physics, are typically less partisan. At their most tolerant, they may, as pluralists, become happy to countenance almost all language games or conceptual schemes as alike reﬂecting the particular perspective of some user; no scheme is privileged by having a unique and special relationship with the way things are, and all are justiﬁed in so far as they embody a form of life or way of reacting to the world and coping with it. Pluralism here makes contact with PRAGMATISM (chapter 36), or the view that what ultimately justiﬁes any mode of description of the world is its utility in enabling us to cope with our problems. For the pluralist there may be no urgent task of relating anything familiar to anything exotic, for both alike are conceived of as being no more than one mode of description among others. Pluralism, therefore, serves as yet another avenue from which metaphysics comes to look to be a chimera – and dislike of old-fashioned metaphysics on this ground is one of	Blackwell
the postmodernist mistrust of traditional philosophy. But, once more, it is not so obvious that one can be satisﬁed with pluralism without already having done enough metaphysics to gain the necessarily relaxed attitude to the different categories of being and their relationships. the characteristics of 68 2 Analysis and Logic METAPHYSICS Metaphysics needs a method, but what method can there be? It is generally said that the most inﬂuential and respected Anglo-American thought this century marks a com- plete change of direction from the ambitious and speculative metaphysical system- building characteristic of the previous period. For the ﬁrst half of this century in Anglo-American philosophy, and for some time afterwards in some places, the problem of method for metaphysics was governed by the ideal of analysis. Faced with a meta- physical problem such as the nature of mind, the philosopher would assemble the central terms with which we talk about mind: thought, sensation, will or whatever; careful attention to what is meant by these terms would reveal the way we conceptu- alize the nature of mind. The model for such a procedure would be that of the ana- lytical chemist, discovering the nature of a substance by breaking it into components. The negative part of this aim is clear enough. When meanings are obscure and unclear, the inferences that we are permitted to make are uncertain too, and it is this uncer- tainty that allows for fanciful and monstrous system-building. But if meaning were cor- rectly located, in ways ﬁrst made possible by the new logical tools developed by FREGE (1848–1925) and RUSSELL (1872–1970) (chapter 37), then those inferences would be systematized and established, and correct methods ﬁnally distinguished from impos- tors. Analytic technique was partly important purely as a defence against wild theo- rizing. In this sense the method of analysis was not new, but only a new label for procedures that are as old as philosophy. In many of his dialogues, SOCRATES (chapters 22 and 23) challenges his audience to state exactly what they mean by a disputed term; MEDIEVAL PHILOSOPHY (chapter 24), as much as later empiricism, is dominated by the aim of precise clariﬁcations of meaning. But a number of considerations changed the impact of this work of clariﬁcation on metaphysics. One was the expansion of logic, suggesting a wholly new range for analytic techniques. Far the most inﬂuential example of this expansion in action was the theory of descriptions, revealed by Russell (1905); a ‘paradigm of philosophy’ whereby judgements that we seem to be making about non- existent ‘things’ are revealed to have quite a different LOGICAL FORM (p. 790), and to make no such commitments. Analysis also provided the principal goal for the satisfactory accommodation between the familiar and the exotic. This is the goal of REDUCTION (p. 312), in which the right attention to meaning reveals that what seem to be claims about a puzzling and exotic area are in fact claims of a familiar, homely kind. Speciﬁc problems about exotic facts and things do not arise, for they are shown to be ordinary, familiar facts and things. For example, if claims about behaviour are thought to be relatively intelligible, but claims about minds seem by contrast mysterious, then the solution might be to analyse claims about minds as disguised claims about behaviour, or dispositions to behaviour. If logical truth or logical proof are felt to be intelligible, but claims about numbers mysteriously abstract and dangerously non-empirical, the solution would be to analyse claims about numbers as disguised recipes for purely logical inferences. The programmes this approach gave rise to – the two examples given are BEHAVIOURISM (pp. 174–6) and LOGICISM (pp. 790–1), but there were many more – not only dominated much 69 SIMON BLACKBURN	Blackwell
philosophical thinking in the ﬁrst half of the twentieth century, but go on directing a large proportion of current research effort. If it succeeds, analysis provided an extraordinarily elegant and economical answer to the problem of the relationship of the familiar to the exotic. For if by analysis we lay bare the real ‘logical form’ or structure of our thought, and it turns out that the facts and things we need to refer to are not what they seem to be at ﬁrst glance, then the problem is solved by absorption of the exotic into the familiar – the ﬁnal and simplest kind of solution to the problem of their relationship. We are shown to have no commitments beyond the familiar and the unpuzzling. It is important to notice that analytical philosophy arose not primarily as a crusade against metaphysics, but only against the undisciplined perversions of metaphysics, which seemed to the major early analysts to surround them. It is in fact notable that the pioneering works of MOORE (1873–1958) (chapter 38) and Russell are themselves preoccupied with questions of what there really is. The familiar class, for Frege, Moore and Russell, included some beliefs that others might regard with suspicion, most notably belief in abstract objects. Moore himself at one time held the distinctly meta- physical-sounding thesis that all that really exists are propositions; Frege remained wedded to but puzzled by the distinctively metaphysical belief in the existence of abstract objects. Russell’s empiricism led to ﬂirtations both with doctrines not unlike those of the idealist Berkeley, and with neutral monism (the belief that the same primi- tive ‘stuff ’, ordered in different ways, makes up on the one hand the mental and on the other hand the physical world). What drove analytical philosophy was not originally hostility to metaphysics, but belief that the correct method for pursuing it had ﬁnally been found. One objection to the method used to be made by more speculative or ambitious philosophers, writing under titles like Clarity is not Enough. With concerns echoing those which ﬁrst arose in ANCIENT GREEK PHILOSOPHY (chapters 22 and 23), they worried that the true nature of mind or other metaphysical topics might not be revealed by analysis, because what is analysed is a compendium of common-sense prejudice or folk- lore which is not particularly likely to enshrine the metaphysical truth. This worry has been revised in a different, scientiﬁc, tone of voice more recently. Since common-sense concepts were formed in pre-scientiﬁc days, there is no particular reason to respect them as a source of metaphysical or any other kind of truth. Indeed, some philosophers think the right reaction to such concepts is the response we mentioned above, namely eliminativism, which is the doctrine that everyday thought about some topic is sufﬁ- ciently infected with errors for its categories to be wholly unreliable. Everyday opinions about the nature of mind in principle deserve no more respect than everyday opinions about the nature of mass or the ﬂight of projectiles, and concepts within which every- day thought frames its opinions may be quite worthless, like those of Aristotelian physics. Thought couched in these terms does not need analysis, but elimination (Churchland 1989; Stich 1983). This is certainly possible. But to turn the possibility into a real likelihood, we must at least correctly identify what the common-sense scheme actually requires. What is common sense committed to, and which parts of it may need eliminating? Here there is a need for work at least signiﬁcantly like that of the analytical philosopher: we would need to know exactly what the commitments implicit in the common-sense scheme are, and this means correctly locating what is meant by the salient terms. 70 METAPHYSICS Let me give a simple example. Suppose someone remarks that one doctrine of every- day common sense is that people often behave as they do because of what they think.	Blackwell
Suppose it is then pointed out that science tells us that physics is a closed system: physi- cal events, such as bodies moving, must have physical causes. But, goes the objection, the description of someone as thinking something is not a physical description of them. Hence, it must be an error to believe that people behave as they do because of what they think. This is a simpliﬁed version of an argument that has obvious eliminativist leanings, for once people accept that it is wrong to think of people as behaving as they do because of what they think, then the idea of people as thinkers at all is also fairly immediately threatened. For if thought does not cause behaviour, it does not seem to do anything else either. But the argument is scarcely compelling as it stands. It raises a host of questions, and they are analytic in their nature. Is the ‘because’ in ‘people behave as they do because of what they think’ actually a causal notion? If it is, does this kind of causal explanation compete with other (neurophysiological) causal expla- nations, or is it compatible with them? Might it make sense to suggest that mental events are identical with physical events, and are therefore in good standing as physi- cal causes of things? To assess the argument we must ﬁrst clear up such questions, but clearing up such questions is investigating the very topics that the analytical philoso- pher took as primary. I return below to discussing some of the moves that have attracted attention in connection with these topics. The Flight from Analysis Even if eliminativism cannot entirely do without analysis, in the last forty years, and especially in the last twenty, analysis has ofﬁcially lost much of its lustre as an ideal. Partly the failure of positivism taught philosophers that theory can take us beyond evi- dence – that concepts that may not have met strict empiricist or positivist standards for meaning are nevertheless perfectly proper. To theorize properly on the basis of data is to invent new conceptual structures. But then, the kinds of thing said when we make use of those structures will not be equivalent in meaning to the things said without them. New theoretical concepts have their own meanings, and it is futile conservatism to deny this by trying to analyse their content in old terms. The ﬂight from analysis was also pro- pelled from other directions. One was the failure of many of the reductionist pro- grammes, few of which gave convincing reasons for supposing that the exotic was just the familiar in disguise. Another was a gradual disenchantment with the foundational role attributed to meaning in the analytic paradigm. Quine (1951) convinced many that the equations of meaning demanded by analysis could not be self-sufﬁcient ‘semantic’ facts known merely by knowing the languages concerned, but would themselves equally represent deep theoretical and scientiﬁc choices. For example, if statements about physi- cal objects are analysed into statements about courses of experience (phenomenalism) this will not be the result of a neutral semantic equation, but will represent a theoretical (metaphysical) conviction about what there is and the way it is to be understood. In other words, semantics is driven by science or even metaphysics as much as the other way round. It was also realized that metaphysics ought to be able to proceed even where semantics is silent. For example, in the theory of value it is notorious that value terms resist analysis into terms without evaluative implications (it is, after all, their distinct 71 SIMON BLACKBURN meanings that deﬁne them as value terms), but this failure of analysis surely should not halt the enterprise of trying to understand the speciﬁc nature of valuing as a human activity, not by purely semantic investigation, but by reﬂecting on the psychological, socio- logical, or even biological role of valuing (Gibbard 1990). Values would be domesticated,	Blackwell
but not by analysis, and perhaps similar progress can be made with other areas. Two ﬁnal elements can be discerned in the move from analysis. We have seen that the analytic paradigm was good at presenting a preferred relation between the exotic and the familiar: the one is absorbed in the other. But it is not so good at telling us what to choose as familiar in the ﬁrst place. And in fact just this point has engendered kinds of conﬂict to which the analytic paradigm provides no ready solution. Like Hume, twentieth- century PHENOMENALISTS (pp. 53–4) found the stream of perceptions familiar, and the physical world exotic. Others exactly reverse the priority. Frege and Russell found LOGIC (chapter 4) familiar, and MATHEMATICS (chapter 11) exotic; others think that logical relations are even more in need of theory than the properties of mathematical struc- tures. Some think that categorical properties are familiar, and that we ought to analyse DISPOSITIONS AND POWERS (p. 702) in terms of them; others again reverse the priority, believing that categorical properties are dangerously exotic, whereas science deals famil- iarly with powers and dispositions. In the absence of any method of settling such ques- tions of priority, analysis is rudderless, for we do not even know in which direction to work. Perhaps the most sustained and inﬂuential case on behalf of a fundamentally analytic method in the second half of the twentieth century is that made by Michael Dummett (b. 1925). In many writings, but most visibly in The Logical Basis of Metaphysics (1991), Dummett has argued that the foundation for metaphysics, and indeed its entire legiti- mate domain, lies in the THEORY OF MEANING (chapter 3). Citing the stagnation and dead- lock that ensues when rival metaphysicians attempt direct descriptions of reality, Dummett believes instead that we must take as our subject not so much reality as the way we think about reality. However, there is no investigation of thought of this kind which is not also an investigation of language and logic. This means we must approach metaphysical problems from the ‘bottom-up’, recognizing that the problem is that of ‘the correct model for meaning for statements of the disputed class’ (Dummett 1991: 12). A meaning-theory, as Dummett conceives of it, will give us this model. It must provide a clear view of how our words function, for as things stand we use words with various meanings, but we do not know what it is that we are doing. We have no clear overview of our own practices, and in this sense do not understand ourselves. A meaning-theory will give a representation of what a person learns when learning a language; it will provide a ‘workable account of a practice that agrees with that which we in fact observe’ (ibid.: 14). Such a transparent understanding of our meanings would certify the correct logic to be used, for it is meaning that confers validity or inva- lidity on an inference; hence it would settle disputes such as that between realists and their opponents, anti-realists in disputed areas. One surprising thing about such a meaning-theory is that its construction can proceed, according to Dummett, entirely innocently of metaphysics. It is the neutral basis of metaphysics, and nowhere owes anything to metaphysical doctrines or choices. And another surprising aspect is that it is itself sufﬁciently powerful to settle meta- 72 METAPHYSICS physical controversy: by adjudicating between rival conceptions of truth offered by realists and their opponents, ‘it will resolve these controversies without residue’ (ibid.). Dummett is not quite promoting a purely a priori method for metaphysics, for the data on which a meaning-theory would be constructed would in one sense be empirical: they would concern the inferential practices of competent users of the language. Neverthe- less the data are, in principle, available to any competent language user purely by reﬂec-	Blackwell
tion upon that practice. There is no appeal to general scientiﬁc facts or empirical facts lying outside the sphere of language use. Yet the reﬂection will deliver results that can settle outstanding metaphysical dispute. The pattern of Dummett’s thought is clearly inﬂuenced by his own favourite example of metaphysical controversy, that between CLASSICAL MATHEMATICIANS AND INTUITION- ISTS (pp. 363–7). In this example it is plausible that the dispute is metaphysical in essence, with the classicists thinking of mathematics as objective, real, independent of us, or ‘out there’, and the intuitionists thinking of mathematical truth as extending no further than mathematical proof (constructivism). It is also plausible that the heart of the dispute is the correct logic for mathematical proof and inference. The classical mathematician believes in bivalence (the logical law that every proposition is either true or false), whereas for the intuitionist, since there is no guarantee that every proposition is provable or disprovable, bivalence cannot be assumed. And ﬁnally, Dummett believes that the issue of the correct logic hinges upon questions that would be tackled by a meaning-theory, in the sense that he conceives of it: the intuitionists believe that clas- sical mathematicians deceive themselves into thinking that they can make sense of notions that are in fact senseless. Deciding whether they are right about that would be the function of just the kind of perspicuous representation of meaning that Dummett advocates. Nevertheless there are grave difﬁculties with the programme as it is presented. The most obvious is that it is inconceivable that a meaning-theory with the powers Dummett describes should be constructed in the innocent way that he also requires. Descriptions of what our words mean are historically among the most theory-laden of philosophical claims. They are put forward, naturally, by philosophers who take them- selves to inhabit a world of one kind or another, and who believe that our mental powers take one shape or another. We need only remember empiricists and positivists of all kinds, whose conception of what we could mean by a term disallows understanding of anything not given in perceptual experience. But on top of this philosophical choice, the conception of perceptual experience was in turn not purely empirical, or obtained in some neutral scientiﬁc spirit. Rather, it was the locus of ﬁerce doctrinal dispute, for instance between perceptual atomists such as Hume and Russell, and more holistic idea- lists such as Bradley (1846–1924) or Neurath (1882–1945). An even more central and prolonged dispute over the best way to conceive of understanding is represented by the problem of universals, which surfaced above in the dispute between those who do and those who do not think of science as carving nature at the joints. To some real- ists, meanings reﬂect natural similarities among things which guide and constrain the application of concepts to new cases; to others of an anti-realist persuasion they do not, but the rules that govern application are a free-standing human construction. But neither side conceives themselves as riding roughshod over empirical facts about the nature of language use. 73 SIMON BLACKBURN Dummett is not, of course, denying the obvious fact that persons who think of meta- physics, ontology, perception and epistemology one way will give different descriptions of meaning from those who think of it another way. His claim is that whether or not this is in fact so, it need not be so. The impact of these intruders on a meaning-theory is, somehow, capable of being avoided. But saying that this is possible is far from showing how it is possible, and Dummett’s own explorations in this direction have not proved reassuring. In his famous paper, ‘The Philosophical Basis of Intuitionistic Logic’ (1978), for example, he explained the Wittgensteinian doctrine that meaning is use as	Blackwell
implying that we cannot be understood to mean anything we cannot be observed to mean, thereby betraying a fondness for observation that is not at all epistemologically or metaphysically neutral – on the contrary, it is one of the principal planks of posi- tivism (the less contentious doctrine would be that we cannot be understood to mean anything that we cannot be thought of as meaning, leaving the relationship of thought to observation to be fought over another day). Even more remarkable is the belief that a satisfactory meaning-theory will exhaust the ﬁeld of metaphysics. We can see the magnitude of this claim by a simple example. Suppose we think of the domain of THEOLOGY (chapter 15) as a part of metaphysics. Then is there really a prospect of an innocent, theologically neutral meaning-theory, not only telling us what is meant by the various terms used by religious thinkers, but also telling us the metaphysical truth about such matters? Surely the gap cannot be bridged: anything innocent enough to qualify on the ﬁrst score must leave the truth of theological claims undecided; anything strong enough to decide them (for instance, the claim that a term like ‘God’ derives its meaning from God, who must therefore exist for us to think as we do) is not the innocent, neutral starting point that Dummett requires and promises. The same dilemma is visible in other areas, such as ethics. Careful atten- tion to the nature of evaluative language is certainly a necessary part of any worth- while philosophy of ethics, but without a wider view of the nature of human choice, desire and action it will not itself settle disputes between one and another metaphysics of value. And semantic doctrines that bear on such disputes, such as the view that ethical language is essentially prescriptive in function, or the view that ethical predi- cates work in much the same way as colour predicates, are not the neutral, purely empirical outcome of a meaning-theory, but represent instead wider philosophical thought about the way ethical commitments actually function. For a ﬁnal example, consider the descriptions that are offered of the language in which we talk about neces- sity and possibility. The best-known semantics for these languages are thoroughly ‘real- istic’, conceiving of us as referring to possible worlds, and describing the inferences we make in classical quantiﬁcational terms (if we are talking about possible worlds, neces- sity becomes equivalent to ‘all’ and possibility to ‘some’, and the logic is then under- stood classically). The semantic success of this kind of description is undoubted, but its philosophical signiﬁcance is controversial in the extreme. Those, like Lewis, whose metaphysics is driven entirely by semantics embrace the real existence of different worlds to ours; others reject any such metaphysical implications, and maintain that reference to possible worlds is some kind of useful ﬁction. The example shows clearly how an overview of the logic of an area can leave its metaphysics almost entirely in the dark. In spite of Dummett’s impassioned and weighty advocacy, the prospects, then, for a neutral semantic methodology for metaphysics are not all that bright. 74 METAPHYSICS The strength of Dummett’s position was its insistence that if metaphysics gives us more than vaguely agreeable pictures, then its content should be reﬂected in our practice. That is, if a doctrine such as realism means anything, it must make a difference whether or not we adopt it, and this difference must be manifested some- where in our thought and language. The mathematical example is one in which the classical mathematician allows different inferences from the intuitionist; the difference is therefore one of logic. It is not, however, clear to what extent this case generalizes – certainly, many people who think of themselves as anti-realists about ethics have no	Blackwell
objection to classical logic as a systematization of correct inference in that sphere. Kant thought that it was precisely because mathematics was our own construction that we had the right to pursue it in the certainty that every mathematical proposition is true or false. A metaphysics is an overall structure determining which explanations of our practices we ﬁnd acceptable: it determines what is familiar, what exotic, what counts as an acceptable accommodation and what is intolerable to us. The difference that realism or anti-realism makes is not necessarily one of logic. It comes out pri- marily in what we accept as explanations of our practice, including explanations of the correctness or otherwise of various patterns of inferences, and it is because of this that classical logical practice can happily coexist with an anti-realist explanation of its appropriateness. Dummett’s conception of the subject is almost unique in the contemporary scene in its insistence on a starting-point that is uncontaminated by scientiﬁc and metaphysi- cal doctrines and ideologies. Much more common is the view that such doctrines will themselves be historically conditioned and therefore possibly changeable. It is impor- tant to notice that in detail the nature of such changes is often controversial. For example, it has been claimed by Rorty (1979) that the philosophy of mind and its prob- lems are historically quite local, being mainly the upshot of the rise of Cartesian phi- losophy in the early seventeenth century. To others this is incredible, since it is possible to point to the universal human preoccupation with such possibilities as life after death or the transmigration of souls as a sure sign that the relationship of mind to body has seldom been fully under control. Again, some see the distinction between primary qualities on the one hand – such as extension and mass, which are thought of as real qualities of objective, independent bodies – and secondary qualities on the other hand – such as colour or taste, which are thought of as inherently subjective or mind- dependent – as the local upshot of the science of the same period. But others point to the foundation of the distinction in Greek thought, and its appearance in various forms in Indian or Chinese thought, and argue that the basis is not a particular scien- tiﬁc ideology, but is found in relatively universal or a priori considerations. 3 Naturalism and Identity The authority of logic lay behind the analytic ideal, for logical relationships are those that are laid bare by analysis. When this authority falters, as I have described it doing, the authority of science is the natural substitute, and it is in its shelter that metaphysics is currently mainly conducted. Metaphysics, on this view, is not discontinuous with science. It is theorizing as pursued by the scientist, sometimes on the basis of familiar 75 SIMON BLACKBURN data and sometimes in the light of new scientiﬁc results. This is the naturalistic self- image, most forcibly propounded by the American philosopher W. V. O. Quine (b. 1908), and dominant in the minds of many or most contemporary philosophers. In this view, it is perfectly proper to attempt, for example, a metaphysics of MIND AND BODY (chapter 5). It is to be done in the spirit of science, with an up-to-date understanding of neuro- physiology, computer science, ecology, evolutionary biology, or any other subject that rings the area. The philosopher marshals the results of the sciences, and propounds a concept of the mind–body relationship that best makes the results ﬁt together, just as the scientist marshals the results of empirical enquiry, and propounds a concept of the nature of things that best makes those results ﬁt together. The metaphysicians’ activ- ity differs only in the level of abstraction required, and indeed any difference from the activity of theoretical science is likely to be vague and provisional. Success in philoso-	Blackwell
phizing would most dramatically be shown if the results of philosophical enquiry actu- ally played a role in driving scientiﬁc progress. It is easy to understand why the naturalistic self-image is so popular. First of all, it answers the question of how metaphysics is possible. It is continuous with science, and, since science is possible, so is metaphysics. Secondly, it allows the philosopher some of the prestige and glory of the scientist. It is reassuring to ally philosophical reﬂection with the most secure and intellectually privileged elements of the contemporary culture. Perhaps philosophy is always something of a free-loader when it comes to con- tinuity with the most prestigious activities. When theology ruled the universities, phi- losophy and theology were continuous; in the ﬁrst part of this century, and after the spectacular successes of modern logic, philosophy was deemed continuous with logic; then a little later with linguistics, and now philosophy marches into the future hand- in-hand with science. (The process is of course not as simple as that. Philosophy has had the scientiﬁc self-image on-and-off since Aristotle. But other paradigms have also had their day.) Thirdly, the assimilation of metaphysics to science solves the problem of method. In the abstract, it is difﬁcult to know how to conduct metaphysical enquiry. How is one to go about solving the mind–body problem, or discussing the nature of necessity? In distant times, the answer might have been by reﬂection on what the deity created, by reﬂection on the logic of mental or modal discourse, or by reﬂection on the language of mental ascription or modal embeddings. But the current answer is more reassuring: scientiﬁcally. With so much to motivate contentment with this answer, it may be well to wonder whether an element of wish-fulﬁlment has crept in. Is it possible that the self-image of philosophy as being continuous with science is largely fantasy – that we have been captured by the most superﬁcial resemblances between philosophical and genuinely scientiﬁc activity? Might it be that science-envy has led philosophers to see themselves through a comfortable modern haze? Of course, if approaching a subject scientiﬁcally simply means attempting to follow it through in a disciplined way, taking account of the known facts, building on our predecessor’s labours, using the best ways we have of distinguishing the good from the bad or the true from the false, then philosophy as prop- erly practised, like any other discipline, must be pursued scientiﬁcally. But in that sense a novelist or poet may also practise their craft scientiﬁcally; the obstacle remains that such activities may bear only the most superﬁcial resemblance to the practice of natural sciences. 76 METAPHYSICS The reason for this is clear if we consider what we want from a satisfactory rela- tionship between the exotic and the familiar. Suppose we are happy with scientiﬁc facts: in the case of people, let us suppose these are facts about their behaviour and their brains. The naturalist will believe that in an important sense brains and behaviour is all that there is. What then is there to say about mind? The simplest answer is that mental states or events or properties are states, events, or properties of the preferred natural kind. The equation would be presented as like that between water and H2O, or between lightning and electrical discharge. An enormously important shift took place when these equations emerged as new paradigms of method in metaphysics, largely in the 1970s, and it was due to the work of writers such as Putnam (1975) and Kripke (1980). It came to be believed that we can relate an exotic area to a domestic one in a less hazardous way than by concentrating on equations of meaning. All we may need is an identity claim about the things and properties in question. We do not also need the claim that the very same thing is meant by talking in terms of one and in terms of	Blackwell
the other. It is sufﬁcient, for instance, to understand water in scientiﬁc terms to know that water is H2O; we do not also need the requirement that the two terms mean the same. Similarly, we would bring mind and consciousness down to natural earth if we could identify mental states, processes and events, with physical states, processes and events. We do not also have to claim that it means the same to talk in either way. Iden- tity of things and properties became the ruling relationship to search for, not identity of meaning. However, casual acquaintance with the modern literature shows that philosophical discussion of such an equation is not very much like scientiﬁc discussion of these equa- tions. There, one would ﬁnd various kinds of empirical evidence and theory. In the philosophical case we have a general belief that mind must be thought of as funda- mentally physical, and then we conduct a great deal of armchair theorizing about the nature of the sustaining equations and identities. This theorizing does not proceed with very convincing examples of actual identities to hand, nor, indeed, is it even known whether convincing ‘type–type’ identities exist. Thus, it is always true that lightning is an electrical discharge, or water is H2O, but it may not always be true that a mental event such as a pain coexists with a type of physical event, or whether pain might be ‘variably realized’ in different brains or different psychologies. It is here that philoso- phers discuss whether the equation is contingent or necessary, whether it relates types of mental events to types of physical events or whether it is sufﬁcient that each mental event is identical with some – possibly different – kind of physical event, whether it matters that description of a person’s mentality often invokes relations with the envi- ronment, history and culture, and so on. Without this kind of discussion the equation is of little interest, but it is this theorizing that is insufﬁciently like anything the scien- tist does that casts doubt on the Quinean paradigm. In fact, it may be claimed that the entire discussion of such identities waits upon a piece of analysis that is, unfortunately, not yet completed. We know how to assess ques- tions of identity when the subjects are things; we may know how to assess them in cases like that of water or lightning, where what is at question is the way in which a natural kind of event is constituted. But do we know how to assess cross-category identities when the subjects are facts, events, states or properties? What does it mean, for example, to say that my state when I enjoy a glass of wine is identical with the state of 77 SIMON BLACKBURN some part of my brain, or that the event of my enjoyment is identical with some such event, or that the property of enjoying the wine is identical with some neurophysio- logical property? Such identity claims do not wear their meanings on their sleeve; nor does a metaphysics centred on asserting them. We do not even know whether each of the abstract categories involved – property, state, event or fact – relates to identity in the same way, or whether very different criteria are needed in each case. To give a simple parallel, some philosophers hold that if a padlock opens by having three notches in line, then it makes sense to say that the state or property of being unlocked is, in this padlock (not others), that of having three notches in line; other philosophers doubt if the equa- tion makes sense, or whether, if it does, it is of any metaphysical use (Lewis 1966; Blackburn 1993). It does not follow that these questions will be investigated by exactly the techniques modelled on analytical chemistry, and developed by Russell, Moore, Carnap and their followers. But support for this conception of the subject comes from the following thoughts. Suppose we need to locate a way of thinking accurately, in order to gain an understanding of the categories it uses. Then the primary data are the inferences	Blackwell
that structure the area. These are conveniently thought of as of three kinds. There are things that enable us to make inferences to doctrines in the area (sometimes called the assertibility conditions of such doctrines), there are inferences among such doctrines, and there are things we may infer from them, or their consequences. Staying with the example of mind, the statement that someone is enjoying their glass of wine may be inferred from certain aspects of their behaviour; it enables us to infer that they believe that there is nothing poisonous about the wine (and many other things of the kind), and it enables us to infer (with less conﬁdence) that they are likely to stay put for some while, and unlikely to start doing push-ups or spit on the carpet. The analytic ideal sug- gests that such inferences, and many like them, give the statement its identity. Not to make them would eventually show that someone does not fully understand what is involved in enjoying a glass of wine; it would reveal incompetence with the notions involved. But identifying these inferences is, in effect, the project of analysis. A term is analysed, in the classical tradition, precisely when we understand what licenses its application, and what consequences may be drawn from it. The problem that many authors ﬁnd with this line of thought is that the inferences in question do not seem to come in statement-sized bundles. The mental world is full of surprises and caveats: behaviour that is indicative of enjoying a glass of wine in one person may not be so in another; what else a person wants or believes when they enjoy a glass of wine may be almost indeﬁnitely variable, and for all we know there may be cultures in which doing push-ups or spitting on the carpet is exactly what we would expect by way of showing such enjoyment. The phenomenon, often called the ‘holism of the mental’, means that straightforward identiﬁcation of the patterns of inference associated with understanding a mental ascription is not to be expected. Once more, enjoying a glass of wine is something that can be ‘variably realized’ in different psychologies, evidenced by different behaviour and giving rise to different expression, depending on a whole range of other factors. Identities nevertheless remain popular instruments for relating the familiar to the exotic. But there are cases that raise more obvious difﬁculties than that of mind. Con- sider the project of giving a naturalistic theory of values. Suppose it is pursued on the 78 METAPHYSICS water–H2O model, by ﬁnding a property with which a value such as goodness may plau- sibly be identiﬁed. Let us suppose for simplicity’s sake that a property is selected, such as that of creating happiness. This is thought of as a natural, empirically respectable property, it being a fairly ordinary fact about the world (however hard to assess in some cases) that some things cause happiness and others do not. So the metaphysician natu- ralizes values by making the equation. Admittedly, saying that something is good is not just saying that it creates happiness – Moore (1903) refuted that idea. Nevertheless, the identity-inclined metaphysician insists, the properties are the same. The puzzle with such an idea is that it is quite unclear which of the problems that ethics generates it actually solves. It leaves untouched any investigation of the particular ‘take’ on the cre- ation of happiness that is had by those who think that it is good, as opposed to those who are indifferent to it, or even who think that it is bad. It does not by itself tell us what kind of mistake is made by these other people. Is it an objective mistake, or a more subjective one; is their error an empirical error, or one of logic? What, indeed, does it mean to see the creation of happiness under the ethical heading? While such questions remain, the identity seems itself a poor contribution to the overall project of giving a naturalistic understanding of our engagement with values.	Blackwell
The Limits of Ontology Although doing metaphysics by means of the identity of property, state, event or fact is popular, it introduces a particular attitude to its problems: one in which the primary questions are ones of what exist; ones of ontology. The central question becomes whether we have two things or one, whether the things are events, properties, states, facts or other more homely items. This is also a contrast with the approach of analysts like Dummett. For the analyst, the unit of signiﬁcance is the entire sentence; there is no self-standing investigation into the kinds of thing that parts of sentences such as predicates stand for (however, as Taylor (1993) remarks, realists have often felt motivated to expound the existence and nature of such items as universals and properties). But many problems, such as that of value, suggest that not all metaphysical issues are well approached in these terms. Another example where ontology scarcely seems to be the issue is the problem of natural law or necessity. This problem, brought into prominence by Hume (1978), is that of gaining some conception of any necessity underlying the brute empiri- cal order of events. Events, as we apprehend them, happen one after another in contin- gent patterns that the natural scientist discovers. Descriptions of the way they fall out may be all right; they form the familiar class. What is exotic and hard to think about is a reason why they must fall out as they do, or why the universe must be ordered as it is, or why the order that is discovered must continue to govern its evolutions in the future. The theological worldview could take refuge in the necessity thought to attach to God’s existence and to God’s goodness in continuing to sustain the same natural order; when such an answer no longer appeals to people’s minds there is urgent need for something to ﬁll its place. The problem is essentially metaphysical, for what we are looking for is some conception of the holding of a law of nature: something lying behind or above the actual pattern of events and constraining them to fall out as they do. Now will such a problem be eased by an ontological doctrine? Many suppose it can. They believe that if we can say, sufﬁciently seriously, that natural laws exist, then everything is ﬁne. It would 79 SIMON BLACKBURN not be sufﬁciently serious to say this but only to mean that there are some regularities that are so central to the enterprise of science that they deserve central and privileged places in the ways we systematize our understanding of the patterns of events. The problem is to gain any conception of what it is for these real natural laws to exist, or how things of the nature they need to have are even possible. How can there exist a strait- jacket, whose nature at one time governs how things must fall out at later times or at all other times (Blackburn 1993: chs 3, 5)? More ontology just provides more things, but things and their continued good behaviour are part of the problem, and so do not seem to be any part of the solution. A ﬁnal, even clearer example in which ontology cannot be the central issue is the problem of free will. Realizing that we ourselves, physically, are determined complexes, or at best are permitted only such departures from determination as the random events of quantum mechanics allow, philosophers friendly to free-will may struggle to ﬁnd a further ‘part’ of us, such as a Cartesian, ghostly, governing agency. But the dilemma is not escaped by these means, for it returns to plague whatever extra is added, for it remains to be explained how it is itself neither wholly determined nor wholly random. The additional ontology is no help with the solution. Problems with identity have led many naturalists to back down a little, and stake their faith on a different, and at ﬁrst sight more tractable relation: that of supervenience. 4 Supervenience Whether or not it makes sense to identify mental and physical events, or ethical and	Blackwell
natural properties, at least we ought to think that the mental arises out of the physi- cal, and the ethical out of the natural. This is commonly put by saying that the mental supervenes on the physical, and the ethical on the natural. What does this mean? The fundamental idea is that once the physical aspects of the world are completely ﬁxed, so are the mental. And once the natural facts about a world are completely ﬁxed, so is the question of its ethics – which actions are right and wrong, or who behaved well or badly. Once the familiar is ﬁxed, so is the exotic. If we think in terms of God creating a world, then all he has to do, according to this idea, is to create the physical world. He will thereby have done sufﬁcient to ﬁx the mental facts, or the ethical facts. There is no second creative task to complete. This idea promises many of the beneﬁts of identity claims, without involving itself in the issues of the previous section. It needs no controversial metaphysics of facts, states of affairs or properties, nor any method of resolving disputed identity claims involving these notions. But it gives the same sense that there is in reality ‘nothing but’ the underlying, physical or natural world. At least it removes one kind of metaphysical distance between the two categories; the distance that would be were there a second creative task to complete, so that we could imagine the physical being as it is, but without the mental overlay, or the natural being as it is, but without the ethical overlay. It is notable that to give us this advantage, the supervenience claim must somewhere involve a necessity. There would be a second creative task to complete if, as well as cre- ating physics, God had to shop for one among several possible ways in which physics is able to ﬁx the mental (in this world). If this relationship were contingent, capable of 80 METAPHYSICS taking different forms in different possible worlds, then clearly there would be a second creative task to complete. It would not be strictly true that physics ﬁxed everything; only physics plus whatever it is as well as physics that links everything else to it. It is, I believe, worthy of note that the best understood examples of supervenience do not quite give us this sort of necessity. Many examples involve a reference to per- ception, as when the face in the picture supervenes upon an array of dots. It is certainly true that to create the face no more is necessary than to create the array of dots. But that seems to be because the world contains perceivers such as ourselves; relative to dif- ferent perceptual powers there would no longer be a face in the picture. So, strictly, God had to do more to create the face in the picture than just arrange the dots; he also had to generate the perceptual sensitivities capable of responding to just that array in just that way. Similarly, to create secondary qualities, such as those of colour, it is not only necessary to create a world in which surfaces reﬂect light in various ways, but also to create the kinds of perceptual systems that detect the variations in just that way. In these cases, instead of the reassuring, physicalist-sounding thesis that everything supervenes on the underlying physics, we only get that it supervenes on the underly- ing physics and the relations between physics and perceivers. Granted, a full-blown physicalist will promise that this relation in turn supervenes upon the physics of the surface and the physics of the perceiver, but, so far, this remains simply a promise. We do not have, in cases like this, reassuring examples of the necessity that supervenience claims require. In the case of ethics it is quite unsatisfactory simply to cite the supervenience of values on natural facts, and then to hope that this makes the intelligible bridge required. The problem is obviously to explain the necessity in question. Granted that any two	Blackwell
worlds that are identical in all natural features are identical in all ethical features, why should this have to be so? What, logically or metaphysically, prevents there being a dimension of freedom, the extra thing that God has to do to ﬁx the fact–value links? Explaining supervenience here marks a constraint on satisfactory theory. Some believe that it can be met by citing the parallel with perceptual supervenience claims such as the ones discussed above; others deny this, but think that attention to the purposes for which we value things can only be met by a value system obeying the supervenience constraint. There arises a kind of pragmatic justiﬁcation for this aspect of the logic of the activity (Blackburn 1993). Once more the issue is one of explanation, and differ- ent metaphysical choices are primarily signiﬁcant because of the different explanatory packages they offer. 5 Conclusion I have structured this survey round the problem of method in metaphysics, and some of the most inﬂuential suggestions about the source of metaphysical method and of metaphysical authority. I have also sketched some of the ways in which authoritative suggestions – analysis, meaning-theory and scientiﬁc reduction – have proved at best aids, but never ﬁnal arbiters of metaphysical success. My own belief is that metaphysics is better regarded as an ocean into which all intellectual rivers ﬂow. When the prob- lems are as abstract and the kinds of explanation required are so difﬁcult to formulate, 81 ROBIN LE POIDEVIN there is bound to be inﬂuence from many different sources: the self-conception of a given period, the scientiﬁc paradigms of that period, the most inﬂuential examples of intellectual success, even the ethical and political agenda, as well as conceptions of the role of the given time. This is not a cause for regret, but it may perhaps arm us with a cautious scepticism when candidates for metaphysical certainty present themselves. 6 Time What is time? We may be at a loss to answer this rather bald, abstract question. But suppose we approach it via the more tractable question of how we become aware of the passage of time. To this, the answer is surely through awareness of change, either in the external world – the ticking of a clock, the movement of clouds in the sky, the setting of the sun – or in our own thoughts. But if awareness of time and awareness of change are the same thing, then perhaps the best answer to our ﬁrst question is this: time just is change. ARISTOTLE (chapter 23) attributed this answer to some of his predecessors. And, as with many of his predecessors’ opinions, he found fault with it. Time could not be the same thing as change, he said, for ﬁrst change can go at different rates, speed up or slow down, but not so time, and secondly change is conﬁned to a part of space whereas time is universal. What are we to make of these objections? Surely time does speed up or slow down, or at least it appears to do so. For people in love, a few hours spent together will pass all too swiftly, whereas time will hang heavily during a labour of unremitting tedium. But such phenomena are easily dismissed as illusory. We can be deceived about spatial matters, such as the shape or size of an object, or its distance from us, so why not also about temporal matters? To see whether it makes sense to suppose that time itself could pass at different rates, consider how we measure the rate of other kinds of change: the speed of a passing bus, for example. We measure the distance it covers against time. Or consider a kettle on a stove. Its rate of heating is given by measuring the rise in temperature against time. So rate of change is variation in some dimension in so many units of time. How, then, would we measure the rate of passage of time? Why, against time, presumably. But this leads to the conclusion that the rate of the passage of time must never vary. For how long could ﬁve minutes take if not ﬁve minutes? But Aristotle’s objection perhaps misses the point. It is true that time could	Blackwell
not be identiﬁed with particular changes, such as the crumbling of a sand castle. But to identify time with change is surely to identify time with change in general. Now it is not at all clear that change in general – that is, the sum of all changes in the universe – could intelligibly be regarded as proceeding at varying rates. Try to imagine every change in the world suddenly doubling in speed. Does that idea make sense? Aristotle would not have thought so. For one thing, we could not possibly notice such a change in rate, for we only notice the change in the rate of some change when comparing it with other changes. We notice the shortening of days with the onset of winter by mea- suring the time between sunrise and sunset against conventional timepieces or our own biological clocks. The idea that time is to be identiﬁed, not with particular changes, but with change in general seems also to avoid Aristotle’s second objection, that change is conﬁned to 82 METAPHYSICS parts of space, whereas time is universal. Only individual changes are spatially con- ﬁned, but the totality of change covers the whole of space. These considerations may have removed one ambiguity in the notion of change, but there remains another. What kind of change do we suppose time to be? Do we think that time is the same as the sum of all the ordinary changes of which we can directly be aware, such as the changing colour of a leaf, and also those which underlie per- ceivable changes, though not themselves perceivable, such as the motion of molecules? Or are we instead thinking of the passage of time itself, the inexorable movement of things once present into the ever-distant past? Of course, a philosopher who said that time was to be deﬁned as the passage of time would not get much of a following, since such a deﬁnition deﬁnes time in terms of itself. We need to have some way of deﬁning the passage of time. This is most vividly described (though some philosophers would object to this way of describing it) as the change in events as they cease to be future, become present, and then increasingly past. One way of capturing the distinction above is in terms of ﬁrst- and second-order change. First-order change is change in the properties of things in the world, where ‘things’ are conceived of as items that persist through time, such as trees, atoms and persons. Second-order change is change in ﬁrst-order changes, namely the shifting degree of futurity or pastness of such ﬁrst-order changes. Second-order change, the changing of changes, is the passage of time. So when we say time is change, is this ﬁrst-order change or second-order change? Suppose we mean ﬁrst-order change. Now, it might seem to us that we could imagine every process in the universe coming to a stop – perhaps after the so-called ‘heat-death’ of the universe, where all energy is perfectly evenly dispersed – and yet time continu- ing to pass. Endless aeons of time might pass in a completely dead, motionless universe. To put it in terms of the distinction above: second-order change need not imply ongoing ﬁrst-order change. But where we have second-order change, we have time. So maybe time can exist in the absence of ﬁrst-order change? Aristotle did not think so, on the grounds that, were all (ﬁrst-order) change to cease, we would cease to notice the passage of time. But we might be more cautious than Aristotle in making perceivabil- ity a criterion of intelligibility. Perhaps there are some states of affairs which we could never even in principle detect. What if instead we deﬁne time as second-order change? This, more subtle, position looks unassailable. How could time exist unless it also passed? To see how things might be otherwise, we need to introduce at this point another distinction, due to the Cambridge philosopher and older contemporary of Russell and Moore, J. E. McTaggart (1866–1925). McTaggart distinguished between what he called the A-series and	Blackwell
the B-series, representing different ways of deﬁning positions in time. To deﬁne the A-series position of an event is to deﬁne the time of the event as past (by varying degrees), present or future (by varying degrees). To deﬁne the B-series position of an event is to deﬁne the time of the event in terms of its relations with other events, that is whether it is earlier than, later than or simultaneous with those other events. The most striking difference between these two series is that, whereas the A-series position of an event is constantly changing, its B-series position remains ﬁxed. If an explosion takes place, and is followed by a ﬁre, then these two events remain forever in this relation to each other: one is earlier than the other. To say that time 83 ROBIN LE POIDEVIN consists of an A-series is another way of stating the view that time is the same as second-order change. But if time just consists of the A-series, what becomes of the B-series? The natural and plausible answer is that the difference between the two series is one only of lan- guage, not of reality. In reality, B-series positions are a direct consequence of A-series positions. Thus, if my getting out of bed this morning is now past, my having lunch is present, and my taking a stroll in the park is future (A-series positions), then getting out of bed is earlier than the lunch, and both are earlier than the stroll (B-series positions). Once the A-series positions are ﬁxed, so are the B-series positions. Or, as one might put it, it is A-series facts which make B-series statements true or false. McTaggart, however, invites us to consider an intriguing alternative (although he later rejects it): that time in reality consists solely of a B-series. Now this idea is not immediately intelligible, and it has to be said that McTaggart does not help us very much to make sense of it, no doubt because he himself thought that, ultimately, it does not make sense. But we can try to do so in ways which go beyond McTaggart’s own discussion. To raise a similar question to one asked earlier: if time is just a B-series, what becomes of the A-series? Now we cannot reply here that the A-series is ﬁxed by the B-series and leave it at that, because there is a very obvious sense in which the B-series cannot determine the A-series. The fact that the battle of Hastings is earlier than the battle of Trafalgar does not completely ﬁx the A-series positions of these events. The A-series positions are only partially ﬁxed: we can say, for example, that if the battle of Trafalgar is present, then the battle of Hastings must be past, but its position in the B-series does not tell us whether the battle of Trafalgar is present or not. Some philosophers take this as proof that a description of time in purely B-series terms, in terms of what comes before what, leaves out important facts. But perhaps all the above considerations lead to is the conclusion that the relation between the two series is not straightforward. We need to bring in some further component. What could that be? Consider this more promising account of the two series. During a particularly heavy downpour, I say, with characteristic understatement, ‘It’s raining’. I am making an A- series statement, in that I am attributing presentness to the rain. But on the account we are now considering what makes my statement true is the purely B-series fact that my statement is simultaneous with the raining. After the sky has cleared, I say ‘It was raining’. Again, what makes this A-series statement, attributing pastness to the rain, true is the purely B-series fact that my statement occurs after the rain. So what B-series facts – facts, that is, about the B-series positions of events – are able to do is to ﬁx the truth or falsity of A-series statements. This theory is sometimes called the tenseless	Blackwell
theory of time. It is capable, it seems, of reconciling the notion that time in reality simply consists of a B-series, with the intuitive belief that statements such as ‘It’s now raining’, or ‘I had my lunch an hour ago’, can be true or false. One consequence of the view that time consists simply of a B-series is that time does not ﬂow: there is in reality no recession of events into the past. What could motivate such a view? A number of arguments have been put forward in its defence, the best known arising from a famous argument of McTaggart’s, in which he attempted to prove the unreality, not only of the A-series, but of time itself. There is no space to do justice 84 METAPHYSICS to that argument here, but three other considerations in favour of the B-series view (that there is in reality no A-series, only a B-series) can be stated brieﬂy: (a) the plausi- bility of the account given above of how A-series statements can be made true simply in virtue of B-series relations between those statements and the events they describe; (b) the fact that we cannot make sense of the rate of the ﬂow of time varying, which in turn makes us question the idea of a rate of ﬂow, and hence of a ﬂow at all; (c) the apparent unanswerability of the question ‘Why is D now?’ (where ‘D’ stands for today’s date) except in purely B-series terms: D is correctly describable as ‘now’ simply because the question is asked on the date it mentions. The B-series, or tenseless, theory of time faces challenges of its own: how can it account for change, or the direction of time? Such an account may be forthcoming, but there is at least one intuitive belief concerning time with which it comes into conﬂict, namely the belief that the future is unreal: there are no future facts in the way that there are present facts. On the B-series view, in contrast, all times are equally real, and this is perhaps the most striking of its consequences. Further Reading This chapter emphasizes twentieth-century metaphysics, but readers might also wish to consult the metaphysical works of historical ﬁgures discussed in other chapters, especially Hume, A Trea- tise of Human Nature and Kant, Critique of Pure Reason. Craig (1987) explores the great modern change from a divine to a human context of metaphysical thought. Russell (1905) was long considered a model for analytical treatment of metaphysical questions. Wittgenstein (1922), while having a metaphysical character of his own, inspired the strenuously anti-metaphysical attitude of the Vienna Circle. The pragmatism of Peirce, James (1907), and Dewey inﬂuenced many recent American writers, including Quine (1951), Goodman (1955), Putnam (1975) and Rorty (1979, 1989). The criticism of positivism in Quine and Wittgenstein (1953) led to new perspectives on the possibility of metaphysics, especially that of Strawson (1959), whose focus on describing the contours of our ordinary conceptual scheme has been rejected in the eliminativism of Churchland (1989) and Stich (1983). Kripke (1980) and Lewis (1983) have provided different accounts of the importance of modal logic (dealing with necessity and possibility) for metaphysics. Dummett (1978, 1991) has argued that logic and the theory of meaning are the basis for metaphysics, especially for dealing with the fundamental conﬂict between realism and anti-realism, a question also considered in Blackburn (1993). Time HISTORICALLY IMPORTANT TEXTS ON TIME Any list of these would include Book IV of Aristotle’s Physics, Augustine’s Confessions, the Leibniz–Clarke correspondence, and Kant’s Critique of Pure Reason (see especially the Transcen- dental Aesthetic and the First Antinomy of Pure Reason). HISTORY OF THE PHILOSOPHY OF TIME An excellent overview of theories of time from the Presocratics to the later middle ages is provided by Sorabji (1983). There is a useful, if brief, discussion of Aristotle, Leibniz 85	Blackwell
SIMON BLACKBURN and Kant in Van Fraassen (1985). Both ancient and modern writers are discussed in Turetsky (1998). INTRODUCTIONS AND ANTHOLOGIES Apart from Van Fraassen’s book mentioned above, Swinburne (1981) covers a good range of topics in the philosophy of space and time. A lively introduction, in dialogue form, to issues such as the beginning and end of time, the passage of time, the nature of eternity, and the relation- ship between time and freedom is Smith and Oaklander (1995). Le Poidevin and MacBeath (1993) is a collection of important readings, some of which are accessible to the non-specialist. It includes J. E. McTaggart’s attempted proof of the unreality of time. The most comprehensive recent introduction to space and time, which discusses at length the question of whether time ﬂows, is Dainton (2001). SPECIFIC TOPICS The relationship between time and change, and issues concerning the topological structure of time (e.g. whether it has a beginning/end, whether it is inﬁnitely divisible, and whether it is like an open line or a closed circle) are very clearly discussed in Newton-Smith (1980). The direction of time is the subject of Price (1996). The A-series/B-series distinction, the tenseless theory of time, and related topics such as the direction of time, the nature of change and the possibility of time-travel, are presented with considerable depth and originality in the second edition of Mellor (1998). Smith (1993) is a detailed and sustained attack on the tenseless theory. A closely related issue, that of the reality of the future, is the topic of Lucas (1989) and of Faye (1989), which also tackles the relationship between time and causality, as does Tooley (1997). The various inter- connections between time and ethics are pursued in Cockburn (1997). The signiﬁcance of the tenseless theory of time, and that of its rivals, for a variety of debates in philosophy is the subject of a collection of readings: Le Poidevin (1998). PHILOSOPHY OF SPACE-TIME PHYSICS The Special and General Theories of Relativity have had an enormous impact on philosophical thinking about time. An engaging, though not particularly easy, introduction to the interaction of the physics and philosophy of space and time is Ray (1991). An important and wide-ranging text is Sklar (1974). Nerlich (1994) is an important collection of Nerlich’s writings on space, time and relativity. References Aristotle 1983: Aristotle’s Physics, Books III and IV (translated with notes by E. Hussey). Oxford: Clarendon Press. Augustine, Saint 1961: Confessions (translated by R. S. Pine-Cofﬁn). Harmondsworth: Penguin Books. Blackburn, S. 1984: Spreading the Word: Groundings in the Philosophy of Language. Oxford: Oxford University Press. —— 1993: Essays in Quasi-Realism. New York: Oxford University Press. Churchland, P. 1989: A Neurocomputational Perspective: The Nature of Mind and the Structure of Science. Cambridge, MA: MIT Press. Cockburn, D. 1997: Other Times: Philosophical Perspectives on Past, Present and Future. Cambridge: Cambridge University Press. Collingwood, R. 1940: An Essay in Metaphysics. Oxford: Clarendon Press. Craig, E. 1987: The Mind of God and the Works of Man. Oxford: Oxford University Press. 86 METAPHYSICS Dainton, B. 2001: Time and Space. London: Acumen Press. Dummett, M. 1978: The Philosophical Basis of Intuitionistic Logic. In Truth and Other Enigmas, London: Duckworth. —— 1991: The Logical Basis of Metaphysics. Cambridge, MA: Harvard University Press. Faye, J. 1989: The Reality of the Future. Odense: Odense University Press. Gibbard, A. 1990: Wise Choices, Apt Feelings. Cambridge, MA: Harvard University Press. Goodman, N. 1955: Fact, Fiction and Forecast. London: London University Press. Hegel, G. 1991 [1821]: The Philosophy of Right (translated by H. B. Nisbet, edited by A. W. Wood). Cambridge: Cambridge University Press. Hume, D. 1978 [1739–40]: A Treatise of Human Nature (edited by P. H. Nidditch). Oxford: Clarendon Press. James, W. 1907: Pragmatism: A New Name for Some Old Ways of Thinking. New York: Longmans.	Blackwell
Kant, I. 1933: Immanuel Kant’s Critique of Pure Reason, Second Impression (translated by N. K. Smith). London: Macmillan. Kripke, S. 1980: Naming and Necessity. Cambridge, MA: Harvard University Press. Le Poidevin, R. (ed.) 1998: Questions of Time and Tense. Oxford: Clarendon Press. Le Poidevin, R. and MacBeath, M. (eds) 1993: The Philosophy of Time. Oxford: Oxford University Press. Leibniz, G. W. and Clarke, S. 1956: The Leibniz–Clarke Correspondence (edited by H. G. Alexander). Manchester: Manchester University Press. Lewis, D. 1966: An Argument for the Identity Theory. In Philosophical Papers, vol. 1. New York: Oxford University Press. —— 1983: New Work for a Theory of Universals. Australasian Journal of Philosophy, 61. Lewis, H. D. (ed.) 1963: Clarity is not Enough. New York: Humanities Press. Lucas, J. 1989: The Future. Oxford: Blackwell. Mellor, D. H. 1998: Real Time II. London: Routledge. Moore, G. E. 1903: Principia Ethica. Cambridge: Cambridge University Press. Revised edition edited by T. R. Baldwin, Cambridge: Cambridge University Press, 1993. Nerlich, G. 1994: What Spacetime Explains. Cambridge: Cambridge University Press. Newton-Smith, W. H. 1980: The Structure of Time. London: Routledge and Kegan Paul. Price, H. 1996: Time’s Arrow and Archimedes’ Point. New York: Oxford University Press. Putnam, H. 1975: The Meaning of ‘Meaning’. In Mind, Language and Reality: Philosophical Papers, vol. 2. Cambridge: Cambridge University Press. Quine, W. V. O. 1951: Two Dogmas of Empiricism. In From a Logical Point of View, New York: Harper. Ray, C. 1991: Time, Space and Philosophy. London: Routledge. Rorty, R. 1979: Philosophy and the Mirror of Nature. Princeton, NJ: Princeton University Press. —— 1989: Contingency, Irony and Solidarity. Cambridge: Cambridge University Press. Russell, B. 1905: On Denoting. Mind, 14. Sklar, L. 1974: Space, Time and Space-Time. Berkeley: University of California Press. Smith, Q. 1993: The Language of Time. New York: Oxford University Press. Smith, Q. and Oaklander, L. N. 1995: Time, Change and Freedom. London: Routledge. Sorabji, R. 1983: Time, Creation and the Continuum. London: Duckworth. Stich, S. 1983: From Folk Psychology to Cognitive Science. Cambridge, MA: MIT Press. Strawson, P. 1959: Individuals: An Essay in Descriptive Metaphysics. London: Methuen. Swinburne, R. G. 1981: Space and Time, 2nd edn. London: Macmillan. Taylor, B. 1993: On Natural Properties in Metaphysics. Mind, 102. Tooley, M. 1997: Time, Tense and Causation. Oxford: Clarendon Press. Turetsky, P. 1998: Time. London: Routledge. 87 SIMON BLACKBURN Van Fraassen, B. 1985: An Introduction to the Philosophy of Time and Space, 2nd edn. New York: Columbia University Press. Wittgenstein, L. 1922: Tractatus Logico–Philosophicus. London: Routledge. —— 1953: Philosophical Investigations. Oxford: Blackwell. Discussion Questions 1 How is metaphysics possible? How do metaphysical questions arise? 2 Does metaphysics discover the broad structure of reality or describe how our conceptual scheme structures our thought about reality? 3 Must all possible conceptual schemes have certain features in common? 4 How might a programme of revisionary metaphysics be justiﬁed? 5 Can metaphysics be assessed without considering its ethical and political implications? 6 Is metaphysics the science of Being? 7 Does metaphysics have an acceptable method? 8 Why is metaphysics preoccupied with its own possibility? 9 Is metaphysics a function of what we ﬁnd familiar and what we ﬁnd puzzling? Can what we ﬁnd familiar change? 10 Are there limits beyond which metaphysics should not go? Does the whole of meta- physics attempt to go beyond these limits? 11 If metaphysical questions are unanswerable, should metaphysics be abandoned? 12 Are metaphysical claims meaningless? 13 Are we able to determine whether the absolute presuppositions of our thought are true or false? 14 Can philosophical reﬂection exist only with hindsight? What does your answer tell us about the nature of metaphysics?	Blackwell
15 Can we eliminate all residual reference to ourselves when we consider the world? 16 Does any conceptual scheme have a unique and special relationship with the way things are? 17 What kind of justiﬁcation can there be for any way of describing the world? 18 Have philosophers been right to move away from analysis as a method for metaphysics? 19 Should we accept Russell’s theory of descriptions as a paradigm of philosophy? 20 In a reductionist programme, how do we determine what requires reduction and what it should be reduced to? 21 Do common-sense concepts need elimination, not analysis? 22 Can semantics be metaphysically neutral? 23 Can logic and a theory of meaning settle metaphysical controversies? 24 Is metaphysics continuous with science? 25 Should metaphysicians search for identity of things and properties rather than for identity of meaning? 26 Why do philosophers and scientists have different questions to ask about identities? 27 Are the basic questions of metaphysics questions of ontology? 28 Should metaphysics deal with supervenience rather than identity? 88 METAPHYSICS 29 To what extent should explanatory success determine our choice of metaphysical position? 30 Should we identify time with change? 31 Does a description of time purely in terms of whether events are earlier than, later than or simultaneous with other events leave out important facts about time? 32 Does time ﬂow? 33 Are the past, present and future equally real? 89 3 Philosophy of Language M A RT I N DAV I E S Philosophy of language deals with questions that arise from our ordinary, everyday conception of language. (Philosophy of linguistics, in contrast, follows up questions that arise from the scientiﬁc study of language.) But saying this does not yet give a clear idea of the sorts of questions that belong distinctively in philosophy of language. Wittgenstein (1953, §119) said, ‘The results of philosophy are the uncovering of one or another piece of plain nonsense and of bumps that the understanding has got by running its head up against the limits of language.’ On this conception, philosophy is about the ways in which we understand and misunderstand language, about how we come to mistake plain nonsense for something that is intelligible, and about what cannot be expressed in language. So, on this view, virtually all of philosophy is concerned with questions about language. It is, indeed, true that language has loomed large in the phi- losophy of the last hundred years or so. But there is still a speciﬁc, recognizable area of the discipline that is philosophy of language. It begins from one absolutely basic fact about language, namely, that expressions of a language have meaning, and can be used to talk about objects and events in the world. For philosophy of language, the central phenomenon to be studied is linguistic meaning. This chapter introduces some of the ways in which that study proceeds. Readers might also like to look at the closely related chapters on PHILOSOPHY OF LOGIC (chapter 4), PHILOSOPHY OF MIND (chapter 5), FREGE AND RUSSELL (chapter 37), and WITTGENSTEIN (chapter 39). 1 Introduction: Questions of Meaning Questions about meaning are central in the philosophy of language. These questions are of two kinds. On the one hand, there are questions about the meanings of partic- ular linguistic expressions (words, phrases and whole sentences); on the other hand, there are questions about the nature of linguistic meaning itself. Questions of the ﬁrst kind belong to semantics (section 4); questions of the second kind belong to meta- semantics (section 5). The business of semantics includes questions about the meanings of subject expres- sions – including PROPER NAMES (chapter 37) (Theaetetus, Fido) and DESCRIPTIVE PHRASES PHILOSOPHY OF LANGUAGE (pp. 798–9) (‘the man in the gabardine suit’, ‘the present king of France’) – and of PRED- ICATE (chapter 4) expressions like ‘is sitting’, ‘barks’, ‘is a spy’ and ‘is bald’. It also	Blackwell
includes questions about the meanings of complete subject–predicate sentences (‘Theaetetus is sitting’, ‘The present king of France is bald’). There are important philo- sophical questions about the meanings of other subject terms, including pronouns (‘I’, ‘you’, ‘she’, ‘he’, ‘it’, ‘they’) and demonstratives (‘this’, ‘that’, ‘this knife’, ‘that butter’), and also about the meanings of expressions that go beyond the terms in the basic com- bination of subject and predicate. Expressions in this latter category include adjectives (‘large’ as it occurs in ‘large ﬂea’, ‘small’ as it occurs in ‘small elephant’, ‘good’ as it occurs in ‘good person’ and in ‘good philosopher’, ‘false’ as it occurs in ‘false sentence’ and in ‘false nose’), modal adverbs (‘possibly’, ‘necessarily), manner adverbs (‘slowly’, ‘clumsily’), and many more. Still within the domain of semantics, and closely related to questions about the meanings of words, phrases and sentences, are questions about the ways in which the meanings of words determine, or at least constrain, the meanings of the phrases and sentences in which they occur. Even if we know what kind of thing the meaning of a sentence is (the meaning of ‘Fido barks’ or of ‘The man in the gabardine suit is a spy’), we also need to understand how component words and phrases make their contribu- tions to the meanings of complete sentences. On the other side of the semantics versus meta-semantics divide are questions about the nature of linguistic meaning itself. Some of these questions are ontological. Are meanings entities; and, if so, what kinds of entities are they? One putative answer might be that the meanings of complete sentences are propositions, and that answer would lead, in turn, to questions about the nature of propositions themselves. An alternative answer might be that, if the meanings of sentences are entities, then they are states of affairs. Someone following up this alternative might say that in order for a sentence to be true, the state of affairs that is the sentence’s meaning needs to be a state of affairs that obtains. Facts might then be identiﬁed with states of affairs that obtain, and true sentences would be said to be true in virtue of facts. In this case, the answer to the onto- logical question about meanings would lead to a version of the correspondence theory of truth. Other meta-semantic questions concern the elucidation or analysis of the concept of meaning. Can we, for example, give any kind of philosophical analysis of the concept of linguistic meaning; and, if we can, what kinds of ideas can legitimately be used in the analysis? How, in general, is the meaning of a linguistic expression related to its use? How is the concept of meaning related to the concept of truth? In particular, for a complete sentence, what is the relation between the meaning of the sentence and the conditions under which an utterance of the sentence would be true? The everyday idea of meaning or signiﬁcance is related to the idea of what is con- veyed or communicated in the use of language. In recent philosophy of language, a standard assumption in meta-semantics has been that there is such a thing as the literal meaning of linguistic expressions, and that the total communicative signiﬁcance of a linguistic act is the product jointly of the literal meanings of the expressions used and of contextual factors. According to that meta-semantic view, semantics is the study of the literal meanings of expressions, and of the way that the literal meanings of complex expressions (phrases and sentences) are determined by the literal meanings of their 91 MARTIN DAVIES component words. Strictly speaking, questions about the interaction between literal meaning and contextual factors belong, not to semantics, but to pragmatics (section 6). 2 Theories of Meaning Answers to semantic questions and answers to meta-semantic questions can be given by propounding what might be called theories of meaning; but that phrase has two very	Blackwell
different senses. In the ﬁrst sense, a theory of meaning answers semantic questions by specifying the meanings of linguistic expressions. In the second sense, a theory of meaning answers meta-semantic questions by providing an elucidatory account of the nature of linguistic meaning. Consider semantic questions ﬁrst. These are questions about the meanings of words and phrases, and about the ways in which these contribute to the meanings of whole sentences. For the words, phrases and sentences of a particular language, someone might seek to answer these questions in a very explicit way, by providing a certain kind of formal, axiomatized theory. The idea would be that the axioms of this theory should specify explicitly the meanings of the words of the language under study, and that rules of inference should then permit the derivation, from those axioms, of theorems speci- fying the meanings of phrases, and ultimately of whole sentences in that language. In the case of any given sentence (say, ‘Fido barks’), the derivation of a meaning- specifying theorem would make use of the axioms of the theory that specify the meanings of the words occurring in that sentence (in this case, the words ‘Fido’ and ‘barks’). This derivation of a theorem from axioms could reasonably be said to display how, in the language under study, the meanings of the component words contribute to the meaning of the complete sentence. A theory that shows how the meanings of sentences depend on the meanings of their parts is sometimes said to be compositional. The construction of theories of the kind envisaged here is not a trivial matter. We might start off by supposing that a typical theorem would say something like: The meaning of the sentence ‘Fido barks’ is the proposition that the particular dog Fido engages in the activity of barking. or, avoiding the explicit talk about propositions: Th1 The sentence ‘Fido barks’ means that the particular dog Fido engages in the activity of barking. We might also suppose that the axioms from which this theorem is to be derived would say things like: Ax1 The word ‘Fido’ means a particular dog, namely, Fido. Ax2 The word ‘barks’ means the activity of barking. But, even in the context of this extremely simple example, we can see that we would immediately confront at least two important issues. The ﬁrst issue concerns the differ- 92 ence between subject terms and predicate terms (here, between ‘Fido’ and ‘barks’); the second issue concerns the derivational route from the axioms to the theorem (here, from Ax1 and Ax2 to Th1). PHILOSOPHY OF LANGUAGE 2.1 The difference between subject terms and predicate terms The issue about the difference between subject terms and predicate terms is this. The two axioms that we proposed have the same form. In each case, the axiom says that an expres- sion stands in the meaning relation to something in the world – in one case, a dog, in the other case, an activity. So, those axioms do not really explain why these two words – ‘Fido’ and ‘barks’ – can go together to make up a sentence, whereas other pairs of words – such as two nouns – cannot together yield a sentence, but merely constitute a list. We can make this point more vivid if we consider, for a moment, nouns that are closely related to predicate terms. Consider, for example, the noun ‘barking’. This noun can, of course, occur in sentences: ‘The barking of the neighbourhood dogs kept me awake’. Indeed, it can function as the subject term in a sentence: ‘Barking is fun’. But the noun ‘barking’ cannot be juxtaposed with the word ‘Fido’ to make a sentence. Similarly, con- sider the noun ‘baldness’. If we juxtapose this noun with a subject term, then what we get – say, ‘The present king of France baldness’ – is not a sentence. (In order to see that it is just a list, we can imagine it as an account of the topics discussed at a meeting.) The point, in short, is that there seems to be an important difference in meaning	Blackwell
between predicate terms, such as ‘barks’ and ‘is bald’, and other closely related expres- sions. The meanings of the predicate terms somehow ﬁt them for combining with subject terms to form sentences, while the meanings of those other expressions (‘barking’, ‘baldness’) do not so ﬁt them. Now, suppose that we were asked to offer a meaning-specifying axiom for the noun ‘barking’ in the style of Ax1 and Ax2. We would be virtually bound to say: The noun ‘barking’ means the activity of barking. But this says just the same about the meaning of the noun ‘barking’ as Ax2 says about the meaning of the verb ‘barks’. Similarly, the model for axioms that we have estab- lished thus far would suggest that for the verb phrase ‘is bald’, and for the noun ‘bald- ness’, the axioms should be: The phrase ‘is bald’ means the property of baldness. The noun ‘baldness’ means the property of baldness. But, as we just saw, while ‘barks’ and ‘is bald’ can function as predicate terms in sen- tences, ‘barking’ and ‘baldness’ cannot. There must be a difference in meaning between ‘barks’ and ‘barking’, and between ‘is bald’ and ‘baldness’; but the axioms that we have proposed do not reveal what this difference might be. 2.2 The derivational route from axioms to theorems The issue about the route from axioms to a theorem is this. The derivation of a theorem from axioms is supposed to have the status of a logical PROOF (pp. 347–9), and it 93 MARTIN DAVIES should make use of well-understood, and well-behaved, logical resources. So, suppose that we apply these requirements to the theorem that we have highlighted (Th1) and the two axioms (Ax1 and Ax2). Certainly the theorem does not follow from the axioms just by way of the logical resources of the propositional calculus or the PREDICATE CALCULUS (p. 367), for example. The inference from Ax1 and Ax2 to Th1 is not a logically valid one. In order to obtain a valid inference, we would need to add an extra premise, in the form of another axiom. Clearly, if we were to add the hypothetical statement: If the word ‘Fido’ means a particular dog, namely, Fido, and the word ‘barks’ means the activity of barking, then the sentence ‘Fido barks’ means that the particular dog Fido engages in the activity of barking. then we could proceed from this statement plus Ax1 and Ax2 to the conclusion Th1 by way of familiar logical rules of inference. The form of the inference would be: A, B, if A and B then C; therefore, C. which is clearly valid. But there would be good reasons to aspire after something more general than this hypothetical statement about the speciﬁc words ‘Fido’ and ‘barks’. We should seek an axiom that speaks in general terms of the effect of putting together two terms – subject and predicate – to make a simple sentence. Provided that we can ﬁnd some account of what a subject term is and what a predicate term is, something along the following lines might suggest itself: Ax3 If a subject term M means something, say, X, and a predicate term N means something, say, Y, then the sentence made up of M followed by N means that X engages in, or exempliﬁes, Y. This is pleasingly general in the way that it talks about M and N and their meanings X and Y, and it is a generalization that we can, it seems, instantiate in order to yield the more speciﬁc hypothetical statement about the subject term ‘Fido’ and the predicate term ‘barks’. If that is right, then the conclusion Th1 can be validly derived from Ax1, Ax2 and Ax3. So, by making explicit the generalization in Ax3, we have made some progress with the requirement that the derivation of a meaning-specifying theorem from axioms is supposed to have the status of a logical proof. Furthermore, we could instantiate the same generalization, Ax3, to give a hypo- thetical statement about the subject term ‘Theaetetus’, the predicate term ‘is sitting’, and the sentence that they go together to make up. Taken in conjunction with speciﬁc	Blackwell
axioms about the meanings of ‘Theaetetus’ and ‘is sitting’, this hypothetical statement would enable us to derive a theorem specifying the meaning of the sentence ‘Theaete- tus is sitting’. But, although this looks promising (apart from the issue about subject terms and predicate terms), there are still legitimate causes for concern as to whether statements like Ax1, Ax2 and Ax3 interact with standard logical resources in a 94 well-behaved way. In fact, most work in semantics proceeds on the assumption that, in order to be sure of the logical status of our derivations, we should not make use of axioms and theorems that talk explicitly about meaning. We shall return to this point later (section 4.1). PHILOSOPHY OF LANGUAGE The Notion of Meaning and Standard Logic In standard logical systems we have a principle – in essence, Leibniz’s Law – saying, among other things, that if Fido = Rover then from any statement about Fido we can infer the corresponding statement about Rover, and vice versa. If we now imagine that one and the same dog has two names, ‘Fido’ and ‘Rover’, we can ask whether from the premise: Fido = Rover plus: Ax1 The word ‘Fido’ means a particular dog, namely, Fido. it follows that: Ax1¢ The word ‘Fido’ means a particular dog, namely, Rover. Either answer – ‘yes’ or ‘no’ – is potentially problematic. If Ax1¢ does follow, then we can see – by a few more steps – that everything that can be said about the meaning of the word ‘Fido’ can be equally truly said about the meaning of the word ‘Rover’. Furthermore, the same will go for any pair of words that pick out the same object in the world, such as the pair ‘Hesperus’ and ‘Phosphorus’, which both pick out the planet Venus. But we should want to leave it open, at this early stage of our enquiry, whether such pairs of words can be distinguished in point of their meaning. Indeed, ahead of detailed theoretical considerations, we might reasonably expect that such pairs of words would sometimes differ in meaning, despite picking out the same object. So perhaps we should prefer the alternative that says that Ax1¢ does not follow from Ax1 plus the premise that Fido = Rover. But, in that case, we shall have to admit frankly that we are departing from what is familiar, and moving to a kind of theory whose logical behaviour is not so well understood. We have been considering theories of meaning, in the ﬁrst of two possible senses of that phrase. These are theories that seek to answer semantic questions. A semantic theory may be a theory in a quite formal sense (with axioms and rules of inference by means of which theorems can be derived) and we have noted that there is an issue about the role of the notion of meaning in formal derivations or proofs. But, apart from the worry about logical good behaviour, a semantic theory might take the concept of meaning as an unanalysed, and unexplained, primitive notion. After all, it is the job of a semantic theory to tell us what linguistic expressions mean. A theory of meaning in the second sense – a meta-semantic theory – will, in contrast, set out to explain the concept of meaning in other terms. A meta-semantic theory will provide an analysis, or some other kind of philosophical elucidation, of the notion of linguistic meaning, perhaps by plot- ting connections between that notion and the notion of use, or the notion of truth. 95 MARTIN DAVIES talk expresses thought about about world Figure 3.1 The relationship between talk, thought and the world, which is explored in the philosophy of language, the philosophy of mind and metaphysics. 3 Language, Mind and Metaphysics: Questions of Priority A semantic theory relates pieces of language to pieces of the world. We use language to talk about the world, and to express our thoughts, which are also about the world. (The aboutness of thoughts is often called intentionality.) Talk, thought and world form a triangle, and in philosophy of language, PHILOSOPHY OF MIND (chapter 5), and META-	Blackwell
PHYSICS (chapter 2) we move around this triangle (ﬁgure 3.1). Thus, for example, we might try to give a philosophical account of some distinctions in reality – say, between OBJECTS AND PROPERTIES (pp. 726–7), or between particulars and universals – in terms of differences among words, or in terms of differences in the realm of thought, provided that we already had some understanding of those linguis- tic or mental differences. Or, going the other way about, we might assume some account of the metaphysical differences, and use it in our philosophical work in the domains of talk or thought. We shall shortly consider just such a question of relative priority between philosophy of language and metaphysics. There are also important questions of priority between philosophy of language and philosophy of mind. Indeed, any strategy for elucidating the concept of linguistic meaning will inevitably depend on our general view of the order of priority as between talk and thought. We need to be clear, ﬁrst, just what notion of priority is at issue here. Then we shall consider three possible views about language and mind. 3.1 Philosophical priorities: language and mind The kind of priority that concerns us here is priority in the order of philosophical analy- sis or elucidation. To say that the notion of X is analytically prior to the notion of Y is to say that Y can be analysed or elucidated in terms of X, while the analysis or elucida- tion of X itself does not have to advert to Y. Thus to say that the notion of BELIEF (chapter 1) is analytically prior to the notion of KNOWLEDGE (chapter 1), for example, is to say that knowledge can be analysed in terms of belief, while a good analysis of belief does not need to reintroduce the notion of knowledge. (This is just to say what the claim would amount to, not whether it would be correct.) Analytical priority should be distinguished from ontological priority and from epis- temological priority. To say that X is ontologically prior to Y is to say that X can exist 96 PHILOSOPHY OF LANGUAGE without Y, although Y cannot exist without X. For example, it might plausibly be said that individuals are ontologically prior to nations. To say that X is epistemologically prior to Y is to say that it is possible to ﬁnd out about X without having to proceed via knowl- edge about Y, whereas ﬁnding out about Y has to go by way of ﬁnding out about X. So it is plausible, for example, that the positions and trajectories of medium-sized mater- ial bodies are epistemologically prior to the positions and trajectories of subatomic par- ticles. We can ﬁnd out about material bodies without investigating subatomic particles but, it might be said, our route to knowledge about subatomic particles has to go via observations of material bodies. Having distinguished these three kinds of priority, we can make the working assumption that they are logically independent of each other. According to that assumption we can, for example, suppose that the notion of X is analytically prior to the notion of Y, without being obliged to hold that X is either ontologically or episte- mologically prior to Y. Our question about the order of analytical priority as between language and mind relates, particularly, to the notion of linguistic meaning and the notion of intentionality (aboutness) for mental states. The three possible views that we need to be aware of are these. Mind ﬁrst: This is the view that it is possible to give a philosophical account of the intentionality of thoughts without essentially adverting to language, and that the notion of linguistic meaning can then be analysed in terms of the thoughts that lan- guage is used to express. Language ﬁrst: This is the view that an account of linguistic meaning can be given without bringing in the intentionality of thoughts, and that what a person’s thoughts are about can then be analysed in terms of the use of language. No priority (both together): This is the view that there is no way of elucidating the	Blackwell
notion of what a person’s thoughts are about without bringing in the notion of lin- guistic meaning, nor the other way around. The two notions have to be explained together. (There is in fact a fourth possible view, which involves a different kind of no priority claim, namely that the two notions are quite unrelated. But this view, while possible, is quite implausible.) All three views have adherents. The no analytical priority view is characteristic of the work of Donald Davidson (b. 1917), who couples it with ontological and episte- mological no priority claims. Thus, Davidson (1975) argues that there can be no thought without language nor language without thought, and that there is no ﬁnding out in detail what a person believes without interpreting the person’s speech (Davidson 1974). The language ﬁrst view ﬁnds expression in Michael Dummett’s (b. 1925) writ- ings (see Dummett 1973, 1991, 1993), while the mind ﬁrst approach is taken by Paul Grice (1913–88) (Grice 1989; see also Schiffer 1972). Given what we have assumed about the logical independence of the three kinds of priority, we can see that one option that is available to us is to agree with Davidson in denying the ontological or episte- mological priority of mind over language, and yet to follow Grice in trying to analyse the notion of linguistic meaning in terms of the thoughts that language is used to express. 97 MARTIN DAVIES 3.2 Philosophical priorities: language and the world Our example of a question of relative priority as between philosophy of language and metaphysics comes from the work of Peter Strawson (b. 1919) (Strawson 1959, Part II; 1970a) on subject terms and predicate terms. The distinction in language between subject and predicate terms is intuitively closely related to the metaphysical distinction between objects or particulars, on the one hand (corresponding to subject terms), and properties or universals, on the other (corresponding to predicate terms). Can we use the metaphysical distinction to help us understand the linguistic one? (As we have already seen, we certainly need some account of the subject–predicate distinction.) Or should we, alternatively, seek a more purely logico-linguistic account of the difference between subject terms and predicate terms, and then try to understand the meta- physical distinction in terms of the linguistic one? One aspect of the subject–predicate distinction to which Strawson gives considerable attention is what he calls the asymmetry of subjects and predicates regarding negation. We can explain the basic idea like this. When a subject–predicate sentence (‘Theaetetus is sitting’) is negated (‘It is not the case that Theaetetus is sitting’, or ‘Theaetetus is not sitting’), the negation can be taken together with the predicate term to yield a new expression (‘not sitting’) that is of the same kind – a new predicate term. But the negation cannot be taken together with the subject to yield a new expression (‘not Theaetetus’) that is of the same kind – a new subject term. In short, predicate terms have negations while subject terms do not. In order to provide (at least the beginnings of) an explanation of this fact about lan- guage, Strawson invites us to consider propositions in which, as he puts it, a general character or kind (or property or universal) is assigned to, or predicated of, a particu- lar, or spatio-temporal individual. The important point is that these general characters come in incompatibility groups vis-à-vis such empirical particulars, while the particu- lars do not conversely come in incompatibility groups vis-à-vis the general characters. Thus, for example, consider the various colours (general characters) and the sundry items of furniture furniture in the room (spatio-temporal individuals). An item of cannot be more than one colour (all over); whereas more than one item of furniture can be the same colour. Or consider the various postures that people can adopt (general	Blackwell
characters) and the philosophers in a courtyard. Perhaps Theaetetus is sitting, and so not standing or lying, while Theodorus is also sitting and Socrates is standing. As both examples illustrate, general characters come in groups that ‘compete’ for spatio- temporal individuals; whereas the individuals do not come in groups that ‘compete’ for the general characters. There is at least the prospect, here, of an explanation of the linguistic distinction between subject terms and predicate terms that appeals to the individuals and the general metaphysical distinction between spatio-temporal characters that they may exemplify. This approach exhibits an order of explanation that is the opposite of the one that is usually associated with the German mathematician and philosopher Gottlob FREGE (1848–1925) (chapter 37). On the Fregean approach, the metaphysical or ontological category of objects is to be read off from an antecedently ﬁxed linguistic category of logical subject terms (roughly, names). So someone adopting this approach would need to be able to give a purely logico-linguistic criterion for an expression to be a name. It 98 is not an easy matter to discharge this obligation without smuggling back in something like the distinction between objects and properties. Strawson himself, after reviewing some of the logico-linguistic marks of the subject–predicate distinction that might be drawn on in the Fregean order of explanation, remarks: PHILOSOPHY OF LANGUAGE The general relative natures of the terms combined in the fundamental combination of predication – the differences or alleged differences between subjects and predicates – have so far been discussed in almost exclusively formal terms, in terms belonging to formal logic itself or to grammar. To understand the matter fully, we must be prepared to use a richer vocabulary, a range of notions which fall outside these formal limits. We assume that the subject–predicate duality, and hence the differences so far remarked on, reﬂect some fundamental features of our thought about the world. (Strawson 1974a: 13–14) Now, it would not, in fact, be quite accurate to say that Strawson awards absolute pri- ority to metaphysics; but we can, at least, say that he does not award a pre-eminent role to philosophy of language. In this, Strawson’s position contrasts with that of Dummett. As we have already noted, when it comes to the question of priority as between lan- guage and mind, Dummett adopts the language ﬁrst option. As we can now add, he also takes a Fregean view on the question about language and metaphysics. So, language as pre-eminent. Indeed, he once Dummett does regard the philosophy of wrote: We may characterize analytical philosophy as that which follows Frege in accepting that the philosophy of language is the foundation of the rest of the subject. (Dummett 1978: 441) This may not be an entirely happy characterization of analytical philosophy as a whole – since it would leave many analytical philosophers on the wrong side of the classiﬁcation – but it does indicate one inﬂuential line of thought within analytical philosophy. We should not, however, think that the only possible reason for interest in philoso- phy of language is that it might hold the key to questions in philosophy of mind and metaphysics. Questions about meaning are no less deep, and no less important, for the fact that, in order to investigate them fully, we may well need to draw on resources coming from other areas of philosophy. 4 Semantic Theories: Davidson’s Programme We have already met the idea of a formal, axiomatized theory that permits the systematic derivation of theorems specifying the meanings of complete sentences drawn from some language (section 2). The sentences might be simple or complex, and the language might be a formal language or a natural language. If we were interested	Blackwell
in actually constructing theories of this type, then we might do well to proceed by stages. The ﬁrst and simplest stage would involve just a ﬁnite stock of sentences from a formal language, and would treat each of the sentences as an unstructured unit. 99 MARTIN DAVIES Perhaps there are just three sentences, ‘A’, ‘B’ and ‘C’, where ‘A’ means that Theaetetus is sitting, ‘B’ means that Fido barks, and ‘C’ means that the man in the gabardine suit is a spy. In this case, the idea of an axiomatized semantic theory is simple to the point of triviality. For we can just take three axioms, one stating the meaning of each of the three sentences, and then the project of deriving a meaning-specifying theorem for each sentence of the language quite literally involves taking no steps at all. The second stage would involve subject–predicate sentences from a formal language. We might have three predicate terms, ‘F’, ‘G’ and ‘H’ and three subject terms, ‘a’, ‘b’ and ‘c’, allowing us to construct nine subject–predicate sentences, among which, let us suppose: ‘Fa’ means that Theaetetus is sitting ‘Gb’ means that Fido barks and ‘Hc’ means that the man in the gabardine suit is a spy. If we want to be able to derive a meaning-specifying theorem for each sentence in a systematic way, then we might hope to formulate a theory in which there is an axiom talking about the meaning of each subject term and an axiom talking about the meaning of each predicate term (rather along the lines of Ax1 and Ax2 in section 2 above). At these ﬁrst two stages, the total number of sentences under consideration is ﬁnite. The third stage would involve an inﬁnite collection of sentences built from some ﬁnite stock by applying constructions over and over. We might take the three sentences, ‘A’, ‘B’ and ‘C’, from the ﬁrst stage, and allow the application of a negation operator ‘ÿ’. This would give three more sentences, ‘ÿA’, ‘ÿB’, ‘ÿC’ (meaning that Theaetetus is not sitting, that Fido does not bark, and that the man in the gabardine suit is not a spy, respectively), and then three more, ‘ÿÿA’, ‘ÿÿB’, ‘ÿÿC’, and then three more, indeﬁ- nitely. Alternatively, we could allow the application of a conjunction operator to two sentences at a time to give ‘A&B’, ‘A&C’, ‘B&C’ and (if we mark the order of the con- juncts) also ‘B&A’, ‘C&A’, ‘C&B’ and (if we allow repetition of conjuncts) ‘A&A’, ‘B&B’, ‘C&C’ (thus, nine new sentences to add to the original three). Then, applying the con- junction operator again, we get ‘(A&B)&C’, ‘(A&C)&C’, ‘(B&C)&C’ and ‘(A&B)&(A&C)’, and many more – in fact, 144 new sentences to add to the twelve. We could, of course, allow the application of both the negation operator and the conjunction operator, to give us sentences like ‘A&ÿB’. Suppose that we do that, and then try to provide an axiomatized semantic theory for this simple, but inﬁnite, collec- tion of sentences. Along with axioms stating the meanings of our three building blocks, ‘A’, ‘B’ and ‘C’, we shall need an axiom saying something about the meaning of ‘ÿ’ and one saying something about the meaning of ‘&’. A ﬁrst thought in this direction might be to suggest these two axioms: ‘ÿ’ means negation ‘&’ means conjunction. 100 PHILOSOPHY OF LANGUAGE But it is quite unclear how, from this axiom about ‘ÿ’, say, and the axiom stating the meaning of ‘A’, we could derive a theorem specifying the meaning of ‘ÿA’. What looks more promising, for the negation operator, is something like: If S is any sentence at all, then the sentence made up by putting ‘ÿ’ together with S means the negation of whatever S means. This builds into the speciﬁcation of the meaning of the operator an indication that it is indeed an operator that can be applied to any sentence to yield a new sentence. Some- thing similar can be done for the conjunction operator; but we shall not pause any longer over these details. (In fact, this also suggests how we might make some progress	Blackwell
with the issue about the difference between subject terms and predicate terms, dis- cussed in section 2.1. We could build into the axiom for a predicate term an indication that it is indeed a predicate – an expression that can go together with a subject term to make a sentence.) A formal language with subject terms and predicate terms, plus a couple of sentence- building operators, is clearly very far indeed from the rich complexity of natural lan- guage. But, apart from the difference of scale, the step from formal languages to natural languages might seem simple in principle. If we know how to provide an axiomatized semantic theory for a formal language with subject terms and predicate terms, for example, then we also know how to provide a theory for a similarly structured frag- ment of natural language that uses real words and phrases instead of letters (‘Theaete- tus’ instead of ‘a’, ‘is sitting’ instead of ‘F’). However, in many cases, the relationship between sentences of a formal language and sentences of a natural language is much less clear. There is a tiny indication of this already in the case of negation, where the formal language operator attaches at the front of a complete sentence whereas natural language negation usually occurs within the predicate term (‘is not sitting’, ‘does not bark’). And, quite generally, the complex- ity of the relationship between formal and natural languages is shown by the difﬁculty of the task (familiar to most philosophy students) of regimenting into logical notation arguments that are expressed in natural language. So, one question that has to be faced is just how to regard the relationship between the superﬁcial forms of natural language sentences, on the one hand, and the regimented forms (sometimes called logical forms; see Sainsbury 1991) to which an axiomatized semantic theory could be applied, on the other hand. Even if we ignore these differences between formal languages and natural lan- guages, it is still the case that we have so far only envisaged a semantic theory for a simple subject–predicate language with sentence operators. Subsequent stages of the project of constructing semantic theories would involve bringing further types of expression within the scope of the enterprise. We should, however, pause here to notice that, while it is clear that axiomatized semantic theories could be of interest to logicians, and – to the extent that they display the meanings of particular kinds of expressions such as pronouns, demonstratives, adjectives and so on – also to linguists, still it may not be evident why they should merit a philosopher’s attention. In fact, the project of constructing compositional semantic theories has been a central concern in recent philosophy of language, and Davidson’s 101 MARTIN DAVIES work in particular reveals several reasons for focusing philosophical attention on semantic theories (Davidson 1967a; and many other papers collected in Davidson 1984). We shall review ﬁve of these reasons. 4.1 The format of semantic theories The ﬁrst reason for paying attention to axiomatized semantic theories concerns the proper format for the theorems that such a theory is supposed to yield. Since these the- orems are supposed to be meaning-specifying, the initially obvious format would be one that relates a sentence to its meaning. If meanings are themselves regarded as entities of some kind, then we might expect a format like: The meaning of sentence S is m. If meanings are not regarded as entities, then we might expect instead: Sentence S means that p. Thus, on the ﬁrst alternative, if meaning-specifying theorems are explicitly to relate sentences to propositions, say, then an example of such a theorem might be: The meaning of the sentence ‘Theaetetus is sitting’ is the proposition that Theaete- tus is sitting. On the second alternative, similarly, an example might be: The sentence ‘Theaetetus is sitting’ means that Theaetetus is sitting.	Blackwell
If, for some general reason, those formats have to be rejected – if the concept of meaning cannot, after all, ﬁgure in the theorems of a compositional semantic theory – then that would be a signiﬁcant meta-semantic result. Davidson (1967a) does indeed reject both those formats, and argues that the target theorems of a semantic theory should, instead, exhibit the format: Sentence S is true if and only if p (where ‘if and only if ’ (‘iff ’) expresses the material biconditional). Since theorems like this specify conditions under which a sentence is true, Davidson is said to favour truth- conditional semantics. Davidson’s argument for this conclusion comes in two steps. The ﬁrst step is intended to rule out the idea that, to each word, each phrase and each sentence, there should be assigned some entity as its meaning. In this step, the so- called Frege argument (Frege 1892: 62–5; cf. Quine 1960: 148–9) is used to show that, under certain assumptions, all true sentences would be assigned the same entity. Clearly, no such undiscriminating assignment of entities could be an assignment of meanings, since it is certainly not correct that every true sentence has the same meaning. (Essentially the same line of argument allows Davidson (1969a) to conclude 102 that there is no point in saying that true sentences correspond to facts. The Frege argu- ment shows that, under certain assumptions, there is only one fact. For further discussion, see Neale 1995.) PHILOSOPHY OF LANGUAGE The Frege Argument The Frege argument shows that if an equivalence relation, E, on sentences meets two conditions then it classiﬁes all true sentences as equivalent, and likewise all false sentences. The two conditions are these. If two sentences are logically equivalent then they are equivalent according to E. If two sentences are the same except for the fact that, where one contains a subject term i, the other contains another subject term, j, referring to the same thing, then they are equivalent according to E. Suppose that we take any two true sentences. Let us choose, say, ‘Penguins waddle’ and ‘The earth moves’. Then, ﬁrstly, the sentence (1) Penguins waddle is logically equivalent to (2) The number that is 1 if penguins waddle and is 0 if penguins do not waddle is 1. So, by the ﬁrst condition, (1) and (2) are equivalent according to E. Similarly, (3) The earth moves and (4) The number that is 1 if the earth moves and is 0 if the earth does not move is 1 are equivalent according to E. But secondly, the subject term in (2), namely ‘The number that is 1 if penguins waddle and is 0 if penguins do not waddle’, refers to the number 1, since penguins do indeed waddle. And the subject term in (4), namely ‘The number that is 1 if the earth moves and is 0 if the earth does not move’, also refers to the number 1. So, sentences (2) and (4) are the same except for the fact that, where one contains a (rather lengthy) subject term, the other contains another (equally lengthy) subject term referring to the same thing. By the second condition, then, sentences (2) and (4) are equivalent according to E. Putting all this together, (1) is equivalent to (2), and (2) to (4), and (4) to (3). So, (1) is equivalent to (3) according to E; and that is what we needed to show. The second step in Davidson’s argument for the truth-conditional format points out that the ‘means that p’ construction presents logical difﬁculties, so that the formal derivations of meaning-specifying theorems will be highly problematic (see section 2.2). In contrast, the truth-conditional format is logically well understood. The way in which theorems specifying truth conditions for complete sentences are derived from axioms assigning semantic properties to words and phrases can, to a considerable 103 MARTIN DAVIES extent, be carried over from the work of Alfred Tarski (1902–83) on certain formal	Blackwell
languages (Tarski 1944, 1956). This second step in the argument is technical in character, and it is a matter of dispute whether a compositional semantic theory making direct use of the ‘S means that p’ format is feasible. (See Schiffer 1972: 162; Taylor 1982; Davies 1984.) As for the status of the ﬁrst step in the argument, it is now widely agreed that there are reasons for rejecting Davidson’s use of the Frege argument. The Frege argument can only establish that all true sentences would be assigned the same entity as their meaning if it is legitimate to suppose that the equivalence relation of having the same meaning would meet the two conditions that ﬁgure in the argument. But this is very far from being obvious. So, pending further argument to the contrary, we can admit that it would be possible for a compositional semantic theory to work by assigning an entity to each sentence as its meaning or semantic value. In particular, one possibility would be to assign to each sentence a certain kind of structured entity – a state of affairs with objects and properties as constituents. The situation semantics programme of Barwise and Perry (1983) constitutes one development of this possibility (see also Taylor 1976; Forbes 1989). 4.2 The correctness of semantic theories The second reason for philosophical interest in semantic theories relates to the condi- tions of adequacy or correctness on any such theory. Whatever is the right format for a semantic theory, we need some account of the conditions under which a theory in that format is the correct theory for the language of a particular speaker or group of speakers. We need to know, for example, what makes it correct to say that, in our lan- guage, the particular string of letters ‘p’–‘e’–‘n’–‘g’–‘u’–‘i’–‘n’–‘s’ . . . and so on means that penguins waddle. In order to be able to consider a possible reformulation of this question, we need to make use of the notion of an abstract or possible language. In the abstract, a language can be considered to be a collection of sentences together with a stipulated assignment of meanings to those sentences. So, for example, there are possible languages in which there is a sentence made up of those same letters – ‘Penguins waddle’ – but with the meaning that the earth moves. Given the notion of a possible language, the question whether a semantic theory is correct for the language of a given group of speakers can be reformulated as the question whether the possible language for which the semantic theory is stipulated to be correct is the actual language of a given group (Lewis 1975; Peacocke 1976; Schiffer 1993). What is sometimes called the actual language relation is thus a relation between languages (in the abstract) and groups of language users. Under the reformulation that we are envisaging, conditions of adequacy on semantic theories become constraints on the actual language relation. Any philosophical elucidation of the key semantic concept used in semantic theo- ries, such as meaning or truth, can be transposed into a condition of adequacy on those theories (or, equivalently, into a constraint on the actual language relation). Thus, suppose for example that an elucidation of the concept of meaning says that any sentence S has meaning m in the language of a group G if and only if some condition C(S, m, G) holds. This can be transposed into a condition of adequacy as follows: 104 If a semantic theory for the language of a group G delivers a theorem saying that the meaning of sentence S is m then it should be the case that C(S, m, G). PHILOSOPHY OF LANGUAGE Similarly, it can be transposed into a constraint on the actual language relation: A possible language in which S has the meaning m is the actual language of a group G only if C(S, m, G). This kind of transposition can be carried out in the opposite direction too. Any condition of adequacy on semantic theories (or any constraint on the actual language	Blackwell
relation) can help us elucidate the key semantic concept used in those theories. Thus, for example, consider semantic theories that adopt the truth-conditional format: Sentence S is true if and only if p. Tarski’s (1956) Convention T imposes a condition of adequacy on such theories; namely, the condition that the sentence that ﬁlls the ‘p’ place should translate (or else be the very same sentence as) the sentence S. This condition of adequacy on truth-conditional semantic theories constitutes a partial elucidation of the semantic concept of truth in terms of the concept of trans- lation. Intuitively, the concept of translation is very closely related to the concept of meaning; and what Convention T requires, in effect, is that the sentence that ﬁlls the ‘p’ place should have the same meaning as the sentence S. If a truth-conditional seman- tic theory meets that condition, then the truth condition speciﬁcations that it yields are guaranteed to be correct. Thus, Convention T’s elucidation of the semantic concept of truth involves a connection between that concept and the concept of meaning: If a sentence S means that p then S is true iff p. But Convention T provides no further help with the concept of meaning itself. One way to shed further light on the concept of meaning – and so, via Convention T, on the concept of truth – would be to spell out other conditions of adequacy on spec- iﬁcations of meanings (or of truth conditions). The concept for which we seek elucida- tion here is the concept of meaning (or truth conditions) in the language of any group. The concept of meaning is the same whether we consider a group of English speakers or a group of Chinese speakers; so we expect that a condition of adequacy should relate to groups in a very general way. One quite general thing that we can say about speciﬁ- cations of meaning is that they help us to describe members of a group as engaging in linguistic acts. The theorems of a semantic theory for the language of a group, G, can license the redescription of utterances of sentences by a member of G as acts of saying or assertion. For example, if the semantic theory says that a sentence S means that Theaetetus is sitting, then we might reasonably construe an utterer of S as saying or asserting that Theaetetus is sitting. Construing a person’s utterances as particular lin- guistic acts is one aspect of interpretation, and what we have just seen is that we can make a link between the theorems of a semantic theory and the project of interpreting the members of a language community. 105 MARTIN DAVIES A semantic theory can play a role in facilitating an overall interpretation of the behaviour of members of a group, by offering a way of understanding their speciﬁcally linguistic behaviour. If there are legitimate constraints on ways of interpreting people, then these may yield constraints on the particular aspects of interpretation that are licensed by a semantic theory. In short, the needs of interpretation may lead us to con- ditions of adequacy on meaning speciﬁcations, or truth condition speciﬁcations, and so to elucidations of the concepts of meaning and truth. In Davidson’s programme, this idea of a link between semantic theories and interpretation is implemented by describing the constraints on the project of radical interpretation. This is the imaginary project of constructing an overall scheme of interpretation for language users about whom we know nothing at the outset. One putative constraint on radical interpretation is that speakers should be so interpreted that what they say about the world – and, presumably, what they believe about the world – turns out to be by and large correct. This is the Principle of Charity (see Davidson 1967a, 1973). Thus, for example (Davidson 1984: 169): A theory of interpretation cannot be correct that makes a man assent to very many false sentences: it must be generally the case that a sentence is true when a speaker holds it to be.	Blackwell
The Principle of Charity has signiﬁcance outside philosophy of language, since it appears to offer some prospect for ANTI-SCEPTICAL ARGUMENTS (pp. 51–6), in particular for arguments to the conclusion that most of what we ourselves say and think is correct (Davidson 1983). But in later work by Davidson and others (Grandy 1973; McDowell 1976; Wiggins 1980) the Principle of Charity has given way to a principle that focuses on intelligibil- ity, rather than on correctness: speakers should be so interpreted that what they say and believe about the world turns out to be by and large reasonable or intelligible. As Davidson (1984: xvii) himself says: ‘The aim of interpretation is not agreement but understanding’. Sometimes a speaker’s being wrong is quite understandable, while being right would be almost miraculous. An eloquent statement of this principle – sometimes called the Principle of Humanity – is provided by Wiggins (1980: 199): Let us then constrain the theory . . . that provides sentence by sentence interpretations of the language L by the requirement that [it] should combine with a plausible anthropology . . . in such a way that in concert the two theories make the best sense possible . . . of the total life and conduct of L-speakers. If the semantic theory that provides sentence by sentence interpretations were to make use of the ‘S means that p’ format, then the Principle of Humanity could be conceived as a condition of adequacy on that semantic theory, roughly along the following lines: If the semantic theory delivers a theorem saying that S means that p then it should be the case that interpreting utterances of S as expressions of the proposition that p contributes to making the best sense possible of the total life and conduct of speakers. 106 PHILOSOPHY OF LANGUAGE This condition of adequacy would help us to elucidate the concept of meaning and thence, via the meaning–truth connection, the notion of truth. But it is also possible to regard the Principle of Humanity as providing direct elucidation of the concept of truth, without explicitly going via the concept of meaning. One way to do this is to introduce a purely formal predicate ‘T’ that applies to sentences and then to consider theories that deliver theorems of the form: Sentence S is T if and only if p. We can impose a condition of interpretational adequacy on such a theory, roughly along the lines of: If the semantic theory (canonically) delivers a theorem saying that S is T iff p then it should be the case that interpreting utterances of S as expressions of the proposi- tion that p contributes to making the best sense possible of the total life and conduct of speakers. (We need the restriction to canonical proofs of T-condition specifying theorems because, for any given sentence S, a theory will deliver many different theorems of the required form: see Davies 1981: ch. 2.) And then we can see that if a theory meets this condi- tion of adequacy its predicate ‘T’ will apply to precisely the true sentences of the lan- guage under study. Wiggins (1980) regards this as showing the way to a ‘substantial theory of truth’. Truth simply is the property that plays the ‘T’-role speciﬁed by the con- dition of adequacy that the Principle of Humanity furnishes (see also McDowell 1976: section 1; Wiggins 1997). 4.3 Semantic theories and mental states Some aspects of natural languages pose particular problems for a semantic theorist working within the Davidsonian framework. One of these aspects is the use of natural language to report on people’s mental states. So a third reason for being interested in semantic theories is that reﬂection on these theories may yield insight into the mean- ings of sentences about mental states, and so, also, insight into the nature of those mental states themselves. To see how the problems arise, consider a belief report sentence like: (BRep) Theaetetus believes that Fido barks.	Blackwell
Intuitively, the meaning of this sentence depends on the meaning of the name ‘Thea- etetus’, the meaning of the verb ‘believes’, and the meaning of the contained sentence ‘Fido barks’. So we might have some expectations about the way in which a meaning- specifying theorem for (BRep) would be derived in a compositional semantic theory. We might expect that the derivation would involve, ﬁrstly, proving a meaning-specifying theorem for the contained sentence ‘Fido barks’, and then using that, plus axioms about the meanings of ‘Theaetetus’ and ‘believes’, in order to prove the target theorem for the belief report sentence. 107 MARTIN DAVIES In the framework of truth-conditional semantic theories, similarly, we might expect that the proof of a truth-condition-specifying theorem for the belief report sentence would make use of a theorem for ‘Fido barks’ plus axioms about ‘Theaetetus’ and ‘believes’. But we can see – at least, in an impressionistic kind of way – why this is liable to be problematic. Statements about people’s mental states display some of the same logical properties as statements about meaning (see the box in section 2). With statements about meaning, it is not usually thought to be the case that from: Fido = Rover plus: Sentence S means that Fido barks it follows that: Sentence S means that Rover barks. This logical behaviour is a departure from standard logical systems, and it raises a ques- tion about the logical resources that would be needed in a semantic theory using the ‘Sentence S means that p’ format. It also helps to motivate use of the alternative ‘Sen- tence S is true iff p’ format, since that is logically well behaved. Thus, for example, from: Fido = Rover plus: Sentence S is true iff Fido barks it certainly does follow that: Sentence S is true iff Rover barks. But while this well-understood logical behaviour is one of the attractions of truth- conditional semantic theories, it also makes it hard for truth-conditional theories to give an adequate treatment of sentences that report on people’s mental states. With statements about belief, it is not usually thought to be the case that from: Fido = Rover plus: Theaetetus believes that Fido barks it follows that: Theaetetus believes that Rover barks. 108 The ﬁrst belief report might be true, while the second was false, for example. This is one aspect of what is called the intensionality of belief reports. The problem posed for truth-conditional semantics is that it is hard to see how a truth-condition-specifying theorem for the sentence: PHILOSOPHY OF LANGUAGE (BRep) Theaetetus believes that Fido barks can be derived in the way that we envisaged, without also involving the denial of the aspect of the intensionality of belief reports that we just highlighted. The derivational route that we envisaged goes via a truth-condition-specifying theorem for the contained sentence ‘Fido barks’, to the effect that the contained sentence is true iff Fido does indeed bark. Given the assumption that: Fido = Rover we can move from this theorem: (FF) ‘Fido barks’ is true iff Fido barks to: (FR) ‘Fido barks’ is true iff Rover barks. But now, consider whatever route leads from the original theorem (FF) to a truth- condition-specifying theorem for the belief report sentence (BRep), to the effect that it is true iff Theaetetus does indeed believe that Fido barks. Given the assumption that Fido = Rover, an exactly similar route will also lead via (FR) to the conclusion that the same sentence (BRep) is true iff Theaetetus believes that Rover barks. So, given the assumption that: Fido = Rover we are driven to the overall conclusion that: Theaetetus believes that Fido barks iff Theaetetus believes that Rover barks. But this goes ﬂatly against the presumed intensionality of belief reports. The issues here are controversial, and there have certainly been proposals for bring- ing within the compass of truth-conditional semantics sentences that are usually reck-	Blackwell
oned to exhibit intensionality (see Davidson 1969b). For now, however, it is enough to notice that attention to compositional semantic theories could bring questions about mental states into sharper focus. 4.4 Semantic theories and epistemology The fourth reason for directing philosophical attention onto the construction of seman- tic theories is epistemological. We have indicated the way in which a ﬁnite set of axioms 109 MARTIN DAVIES can permit the derivation of a meaning-specifying theorem for each of inﬁnitely many sentences. So in the case of a given language with inﬁnitely many sentences, someone who knew the facts stated by the ﬁnitely many axioms of a compositional semantic theory would thereby be in a position to come to know what each sentence of the language means. The existence of ﬁnitely axiomatized semantic theories thus sheds light on an epistemological problem: how is it possible for a ﬁnite being to master an inﬁnite language? So far as the purely formal notion of an axiomatized theory goes, the set of axioms of a theory does not have to be ﬁnite. But Davidson (1965) uses the requirement that a language should be learnable (and learnable by a ﬁnite being) to motivate a ﬁnite axiomatization constraint on semantic theories. Since Davidson favours truth- conditional semantic theories, this yields the idea that the aim of a semantic theorist is to construct ﬁnitely axiomatized theories of truth conditions for (regimented frag- ments of) natural languages – building, so far as possible, on the work of Tarski. The sorts of theories envisaged are sometimes spoken of as Tarski–Davidson truth theories, and there is a very considerable body of work – much of it quite technical in character – in pursuit of this aim. We should note, however, that there is some dispute over the ﬁnite axiomatization language (Schiffer 1987: ch. 7) constraint. At least one prominent philosopher of argues that it may be possible for a speaker to master a language, even though no ﬁnitely axiomatized truth-conditional semantic theory can be provided for it. Even if there is a ﬁnitely axiomatized semantic theory for a language, there is still a pressing question about the explanatory relevance of the existence of such a theory. Knowledge of the facts stated by the axioms of a compositional semantic theory would sufﬁce, in principle, for knowledge as to what each sentence of the language means. But ordinary speakers of a natural language usually lack conscious knowledge of any compositional semantic theory for their own language. So the question that presses is how the mere existence of a ﬁnitely axiomatized semantic theory, unknown to ordinary language users, can solve the epistemological problem that they apparently face. One approach to answering this question begins by considering a wider range of knowledge that ordinary language users have. Along with knowing what the sentences of their language mean, ordinary speakers of English also know, for example, that in the sentence: Nigel shaved him the word ‘him’ cannot refer back to Nigel, while in the sentence: Nigel wanted Bruce to shave him the word ‘him’ can refer back to Nigel. They know that the sentences: Less than two books were on the table More than one book was on the table 110 PHILOSOPHY OF LANGUAGE sound ﬁne, while: Less than two books was on the table More than one book were on the table sound somehow wrong – despite the fact that ‘Less than two books’ is equivalent to ‘At most one book’, and ‘More than one book’ is equivalent to ‘At least two books’. And we could multiply examples of speakers’ linguistic knowledge indeﬁnitely. Ordinary language users have conscious access to these pieces of knowledge. But they do not have conscious knowledge of any set of axioms or rules from which these particular pieces of knowledge follow. Theoretical linguists articulate sets of linguistic	Blackwell
(especially, syntactic) rules or principles called grammars. But the explanatory claims of theoretical linguistics do not end with the mere existence of grammars from which particular pieces of linguistic knowledge could be derived. In a famous passage, Noam Chomsky (b. 1928) says: Obviously, every speaker of a language has mastered and internalized a generative grammar that expresses his knowledge of his language. This is not to say that he is aware of the rules of the grammar or even that he can become aware of them . . . Any interest- ing generative grammar will be dealing, for the most part, with mental processes that are far beyond the level of actual or even potential consciousness. (Chomsky 1965: 8) According to Chomsky (1965, 1986, 1995), then, ordinary language users possess a body of linguistic knowledge which is, for the most part, inaccessible to consciousness. For this reason, it is often spoken of as tacit knowledge. It is because they have this body of tacit knowledge of linguistic rules that they are able to know the vast host of par- ticular things about their language. While Chomsky is concerned primarily with knowledge of syntax, it seems that the idea that ordinary language users possess a body of tacit knowledge could also be applied in the area of semantics. Thus, one way of answering the question about the explanatory relevance of the existence of compositional semantic theories would be to credit ordinary language users with tacit knowledge of the axioms of such a theory (see below, section 5.3). Some philosophers of language are critical of any appeal to the notion of tacit knowledge, claiming that it embodies conceptual confusions (Baker and Hacker 1984). But accounts of the notion of tacit knowledge, particularly as it applies to semantic theories, have been offered (see Evans 1981a; Davies 1986, 1987, 1989; Peacocke 1986, 1989), and the proposal that ordinary language users have tacit knowledge of the axioms of truth-conditional semantic theories has been developed in some detail (see Higginbotham 1985, 1986, 1988, 1989a, 1989b). We should note, however, that there are other proposals for answering the episte- mological question about compositional semantic theories and ordinary language users’ knowledge about the meanings of complete sentences. For example, Dummett (1976, 1991) makes use of a notion of implicit knowledge that is importantly different from Chomsky’s idea of tacit knowledge, and Wright (1986) sees the construction of compositional semantic theories, not as an articulation of language users’ actual knowledge, but as a matter of RATIONAL RECONSTRUCTION (p. 386). 111 MARTIN DAVIES 4.5 Semantic theories and metaphysics The ﬁfth reason for philosophical interest in semantic theories is that the task of bring- ing speciﬁc linguistic constructions within the scope of a compositional semantic theory sometimes sheds light on issues in other areas of philosophy, and particularly in metaphysics. We have already discussed the relationships between philosophy of language, philosophy of mind and metaphysics, so we shall be rather brief here. Semantic theories for natural languages have to deal with pronouns (‘I’, ‘you’, ‘she’, ‘he’, ‘it’, ‘they’), demonstratives (‘this’, ‘that’, ‘this knife’, ‘that butter’) and other index- ical expressions (‘now’, ‘then’, ‘today’, ‘tomorrow’, ‘here’, ‘there’). Questions about the semantic contributions of these expressions (see Kaplan 1989) inevitably highlight issues in ontology, as well as in philosophy of mind. For, when we consider the mental states that are typically expressed using these expressions, we ﬁnd that they are mental states that involve thinking about persons, things, times and places in particular ways. So an adequate account of the thoughts expressed by the use of these expressions seems to involve a commitment to talking about ways of thinking or, in Fregean terminology, about SENSES (pp. 793–6) (see Frege 1918; Perry 1977, 1979; Evans 1981b, 1982;	Blackwell