الرئيسية The Death of Argument: Fallacies in Agent Based Reasoning

The Death of Argument: Fallacies in Agent Based Reasoning

The present work is a fair record of work I've done on the fallacies and related matters in the fifteen years since 1986. The book may be seen as a sequel to Fallacies: Selected papers 1972-1982, which I wrote with Douglas Walton, and which appeared in 1989 with Foris. This time I am on my own. Douglas Walton has, long since, found his own voice, as the saying has it; and so have I. Both of us greatly value the time we spent performing duets, but we also recognize the attractions of solo work. If I had to characterize the difference that has manifested itself in our later work, I would venture that Walton has strayed more, and I less, from what has come to be called the Woods-Walton Approach to the study of fallacies. Perhaps, on reflection "stray" is not the word for it, inasmuch as Walton's deviation from and my fidelity to the WWA are serious matters of methodological principle. The WWA was always conceived of as a way of handling the analysis of various kinds of fallacious argument or reasoning. It was a response to a particular challenge [Hamblin, 1970]. The challenge was that since logicians had allowed the investigation of fallacious reasoning to fall into disgraceful disarray, it was up to them to put things right. Accordingly, the WWA sought these repairs amidst the rich pluralisms of logic in the 1970s and beyond.

Categories: Mathematics\\Logic
السنة: 2004
الناشر: Springer
اللغة: english
الصفحات: 388
ISBN 13: 9789048167005
Series: Applied Logic Series 32
File: DJVU, 2.92 MB
تحميل (djvu, 2.92 MB)
You can write a book review and share your experiences. Other readers will always be interested in your opinion of the books you've read. Whether you've loved the book or not, if you give your honest and detailed thoughts then people will find new books that are right for them.
APPLIED LOGIC SERI S The Death of Argument Fallacies in Agent-Based Reasoning John Woods Springer-Science+Business Media, B.V.

The Death of Argument

APPLIED LOGIC SERIES VOLUME 32 Managing Editor Dov M. Gabbay, Department of Computer Science, King's College, London, U.K. Co-Editor Jon Barwisef Editorial Assistant Jane Spurr, Department of Computer Science, King's College, London, U.K. SCOPE OF THE SERIES Logic is applied in an increasingly wide variety of disciplines, from the traditional subjects of philosophy and mathematics to the more recent disciplines of cognitive science, computer science, artificial intelligence, and linguistics, leading to new vigor in this ancient subject. Kluwer, through its Applied Logic Series, seeks to provide a home for outstanding books and research monographs in applied logic, and in doing so demonstrates the underlying unity and applicability of logic. The titles published in this series are listed at the end of this volume.

The Death of Argument Fallacies in Agent Based Reasoning by JOHN WOODS The Abductive Systems Group, University of British Columbia, Vancouver, Canada and Department of Computer Science, King 's College, London, England and Department of Philosophy, University of Lethbridge, Canada Springer-Science+Business Media, B.V.

A CLP. Catalogue record for this book is available from the Library of Congress. ISBN 978-90-481-6700-5 ISBN 978-1-4020-2712-3 (eBook) DOI 10.1007/978-1-4020-2712-3 Printed on acid-free paper All Rights Reserved © Springer Science+Business Media Dordrecht 2004 Originally published by Kluwer Academic Publishers in 2004 Softcover reprint of the hardcover 1st edition 2004 No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording or otherwise, without written permission from the Publisher, with the exception of any material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work.

For their unstinting support, their love and endless good humour, I dedicate this work to Carol Woods Catherine Armstrong Kelly Woods Michael Woods.

Contents Dedication v Preface xiii Acknowledgments xvii Prologue xix Part I Metatheoretical Questions 1. WHO CARES ABOUT THE FALLACIES? 3 1 Questions About Fallacies. 3 2 The Current State of Fallacy Theory. 5 3 Light At The End Of The Tunnel? 7 4 Their Importance. 12 5 Is That All There Is? 15 6 Critical Thinking 19 7 So What? 20 2. THE NECESSITY OF FORMALISM 25 1 Methodological Pluralism 25 2 Formal Logic 26 3 Nomic Systematicity 30 4 Technical Information 31 5 Theory and Practice 32 5.1 Massey 33 5.2 McPeck 34 5.3 McP**k 35 5.4 Scriven 35

viii THE DEATH OF ARGUMENT 3. 6 7 8 9 10 Pedagogy Conclusion Appendix A Appendix B Appendix C THE INFORMAL CORE OF FORMAL LOGIC 1 2 3 4 5 6 7 Artificial Languages The Language of PC The Utility of PC The Irreducible Informality Quantum Logic Further Work for Informal Logic Informal Theories of Implication and Consistency 36 38 39 39 41 43 43 45 48 49 53 55 60 Part II Threats and Intimidation 4. AD BACULUM AND PASCAL'S WAGER 65 1 The Standard Treatment 65 2 Prudential Argument 66 3 Case One: The Heist 67 4 Case Two: The Anti-smoking Commercial 67 5 Case Three: Pascal's Wager 68 6 Doxastic Surrender 69 7 Inaccessibility to Reason 70 8 The State, the Party and the Wager 71 5. APPEAL TO FORCE 75 1 Arguments Prom the Stick 75 2 Case One 76 3 Case Two 76 4 Case Three 76 5 Case Four: Risk Aversion Strategies 77 6 Case Five: The Stick-Up 78 7 Case Six: Negotiations 80 8 Case Seven: Veiled Threats 82

Contents IX 9 10 11 Practical Arguments Ad Baculum Reasoning 10.1 Collective Bargaining 10.2 The Mugger 10.3 Anti-smoking Arguments 10.4 Utilities Description of a Basic Model 11.1 Example 11.2 Example 11.3 Propositions 11.4 Actions Part III Arguments Involving Reference to Persons 6. DIALECTICAL BLINDSPOTS 1 2 3 4 5 Not Practising What You Preach Insincerity and Irrationality Explanatory Opacity Blindspots Trust 7. AD HOMINEM 1 2 3 4 5 Johnstone to Locke Aristotelian Refutations Falsifying Refutations Peirastic Refutations Proof Ad Hominem 8. AND SO INDEED ARE PERFECT CHEAT 1 2 3 4 5 "Fallacies" that aren't Fallacies Bias Popular Appeals and Common Knowledge Preconception and Dogmatism 4.1 Ad Verecundiam 4.2 Conservatism 4.3 Inconsistency Doxastic Loyalty 84 86 87 89 89 90 91 92 92 93 93 97 97 99 104 106 107 111 111 112 118 120 122 125 125 126 131 137 138 139 141 144

X THE DEATH OF ARGUMENT Part IV Pragma-Dialectics 9. PRAGMA-DIALECTICS 151 1 The Pragma-dialectical Approach 151 2 Difficulties 153 3 Unification 155 4 Prospects for Pragma-dialectics 156 10. BUTTERCUPS, GNP'S AND QUARKS 161 1 Theory-dependency 161 2 Fallacies 163 3 Moderate Dependency 165 4 Theory and Analysis 168 11. UNIFYING THE FALLACIES? 171 1 Theory Reductions 171 2 Exemplar Theory 174 3 VEG as Theoretical Stipulation 177 4 Reconciling WW and VEG 179 Part V Intractable Disagreement 12. STANDOFFS IN PUBLIC POLICY 185 1 Degrees of Standoff, Force I-III 185 2 Extremism 187 3 Dealing with Standoffs of Force Five 194 4 Education as Strategy: Persuasion and Thought Control 197 13. STANDOFFS AND DEMORALIZATION 201 1 Demoralization 201 2 Spinning 208 3 Witnessing 211 4 Privatizing the Good 212 5 Indoctrination 214

Contents XI Part VI How to Interpret Arguments 14. CHARITY: CAN WE FIND IT A HOME? 219 219 222 226 232 233 239 239 242 igmatic Optimum 243 246 249 1 2 3 4 5 Slogans Radical Charity Dialogical Charity Radical Charity Again How Reliable is Charity? 15. MISSING PREMISSES 1 2 3 4 5 Cooperative Conditions Validity The Logical Minimum and Genericity Find the Premisses Part VII Analogy 16. BY PARITY OF REASONING 1 2 3 4 5 6 7 Parity of Reason Standoffs The Structure of Analogy Counterexample Relevance Analogical Predication Deep Structure 253 253 257 260 262 265 268 269 17. VERDI IS THE PUCCINI OF MUSIC 273 1 Analogies 273 1.1 The Method of Cases 275 1.2 Analogical Prediction 276 1.3 What the Cases Suggest 279 1.4 Numerical Analogical Prediction 283 1.5 More Theory 285 2 Analogical Semantics 290 3 Concluding Reflections 296

Xll THE DEATH OF ARGUMENT Part VIII Induction 18. SECUNDUM QUID 301 1 Omitting a Qualification 301 2 DeMorgan's Approach 304 3 Extending the Reach of the Fallacy 306 4 Petitio Principii 309 19. HASTY GENERALIZATION 311 1 Hasty Generalization: The Basic Idea 311 2 Primitive Induction 314 3 Strategic Rationality 315 4 Practical Reasoning 317 5 Notional and Behavioural Belief 319 6 The Raven Paradox 320 7 The Problem of Induction Again 324 8 Strategic Rationality Again 331 9 Conclusion 333 20. THE PROBLEM OF ABDUCTION 335 1 Abduction 335 2 Induction Has a Justification 338 3 Objections 341 4 Skepticism 345 Epilogue: The Way Ahead? 349 References 359 Index 373

Preface The present work is a fair record of work I've done on the fallacies and related matters in the fifteen years since 1986. The book may be seen as a sequel to Fallacies: Selected papers 1972-1982, which I wrote with Douglas Walton, and which appeared in 1989 with Foris. This time I am on my own. Douglas Walton has, long since, found his own voice, as the saying has it; and so have I. Both of us greatly value the time we spent performing duets, but we also recognize the attractions of solo work. If I had to characterize the difference that has manifested itself in our later work, I would venture that Walton has strayed more, and I less, from what has come to be called the Woods-Walton Approach to the study of fallacies. Perhaps, on reflection "stray" is not the word for it, inasmuch as Walton's deviation from and my fidelity to the WWA are serious matters of methodological principle. The WWA was always conceived of as a way of handling the analysis of various kinds of fallacious argument or reasoning. It was a response to a particular challenge [Hamblin, 1970]. The challenge was that since logicians had allowed the investigation of fallacious reasoning to fall into disgraceful disarray, it was up to them to put things right. Accordingly, the WWA sought these repairs amidst the rich pluralisms of logic in the 1970s and beyond. The WWA was never intended as a general theory of argument or a comprehensive articulation of an informal logic. It was a dominantly logical approach to the study of fallacies, an approach that emphasized the utility of nonstandard systems of logic, including dialogue logic. Neither was it supposed by the WWA that fallacies were intrinsically dialogical, as is now supposed by Douglas Walton in company with the pragma-dialecticians of the Amsterdam School. Nor was the WWA much interested in finding a unified theoretical structure within which everything bearing the name of fallacy would have a well- elucidated home. This was not a decision borne by failed attempts at

XIV THE DEATH OF ARGUMENT federal amity, but rather one that grew out of the conviction that any such unification was destined to end up as theoretically thin. Here, too, as with our friends in Amsterdam, Walton's present inclination is to see in unification prospects of greater theoretical robustness than do I; and I on the other hand remain unconvinced, and so an unrepentant pluralist, at least for now, about the structure of fallacious reasoning. There remains a point on which Walton and I remain in perfect accord. It is that getting the fallacies right is an extremely important task for logic, indeed a central part of its mandate, and that it is a task much more avowed than performed. I hope that I might be forgiven a brief shudder of impatience over the slapdash work that still makes its way into basic textbooks in logic and critical thinking. In one way or another, everything in this book has been influenced by the Netherlands, my intellectual home away from home. Some chapters first took shape at the Netherlands Institute for Advanced Study, where under the direction of Frans van Eemeren I was part of the Research Group on Fallacies, together with Rob Grootendorst, Sally Jackson, Scott Jacobs, Agnes van Rees, Agnes Verbiest, Douglas Walton and Charles Willard. The remaining chapters were all written in whole or part when for portions of most of the years in this period I was Adjunct Professor in Frans van Eemeren's Leerstoelgroep Taalbeheersing, Argumentatietheorie en Rhetorica at the University of Amsterdam. I greatly value my intellectual association with Holland, and am gratified that over the years it has been the occasion of enduring friendships. For their interest and support of my work I warmly thank my Dutch colleagues and friends, Frans van Eemeren, the late Rob Grootendorst, Else Barth, Erik Krabbe, Agnes van Rees, Eveline Feteris, Francesca Snoeck Henkemans, Peter Houtlosser, Bart Garsen, Johan van Benthem, Jeanne Peijnenburg, and Frank Veltman. I am no less grateful to my non-Dutch supporters and critics: In Canada, J.A. Blair, Ralph Johnson, Robert Pinto, Hans Hansen, David Hitchcock, Christopher Tindale, Trudy Govier, Andrew Irvine, Leslie Burkholder, Leo Groarke, Mark Vorobej, Thomas, Hurka, George Englebretsen, J.J. Macintosh, Michael Stingl, Ian Hacking, and Douglas Walton; in the United States, the late Henry Johnstone, Jonathan Adler, Harvey Siegel, Michael Wreen, Lawrence Powers, Mark Weinstein, Michael Scriven, James Freeman, Jaakko Hintikka, William Lycan, Alvin Goldman, Jonathan Bennett, Jonathan Bennett, Julius Moravcsik, Patrick Suppes, Bas van Fraassen, Sally Jackson, Scott Jacobs, Joseph Wenzel, Charles Willard, Isaac Levi, Dale Jacquette, and John Hoaglund; and Dov Gabbay, Jonathan Cohen, Chris Reed, Peter McBurney, Timothy Williamson, Jim Cunningham and Alec Fisher (U.K.); Harald Wohlrapp and Hans Jiirgen

PREFACE xv Ohlbach (Germany); Jean-Paul van Bendegem and Paul Gochet (Belgium); Gabriella Pigozzi (Italy); Lorenzo Pefia (Spain); Graham Priest, Greg Restall, the late Richard Sylvan and Robert Meyer (Australia); and Roderic Girle (New Zealand). Research for these chapters has been supported financially by, in addition to the Netherlands Institute for Advanced Study and the Leer- stoelgroep Taalbeheersing, Argumentatietheorie en Rhetorica, the Social Science and Humanities Research Council of Canada, the University of Lethbridge Research Fund, the Dean of Arts and Science, University of Lethbridge, and King's College, London. My thanks to all. My research assistants in Lethbridge, Dawn Collins, Ethan Toombs and David Graham, have put the book into camera-ready form and prepared the index. I am most grateful for this excellent help. My thanks also to the publisher, Charles Erkelens and his assistant, Lucy Fleet as well as Deborah Doherty, from Author Support, for technical assistance and direction.

Acknowledgments I would like to thank the following co-author, editors, and publishers for their permissions to reprint material in this book. ■ Co-author Brent Hudak for "Verdi is the Puccini of Music", Syn- these, 92 (1992), 189-220; and "By Parity of Reasoning", Informal Logic, 11 (1990), 135-139. ■ Kluwer Academic Publishers for "Verdi is the Puccini of Music", Synthese, 92 (1992), 189-220; "And So Indeed Are Perfect Cheat", Argumentation 9 (1995), 645-668; and "The Necessity of Formalism in Informal Logic", Argumentation 3 (1989) 149-167. ■ Royal Netherlands Academy of Arts and Sciences for "Public Policy and Standoffs of Force Five", in E.M. Barth and E.C.W. Krabbe (eds.), Logic and Politics of Culture, 1992, 9-108. ■ Informal Logic for "By Parity of Reasoning", Informal Logic, XI (1990), 125-139; "Buttercups, GNP's and Quarks: Are Fallacies Theoretical Entities?", Informal Logic, X(1989), 67-77; and "Is the Theoretical Unity of the Fallacies Possible?", Informal Logic, XVI (1994), 77-85. ■ Springer-Verlag GmbH &; Co. KG for "Deep Disagreements and Public Demoralization", in Dov M. Gabbay and Hans Jlirgen Ohlbach (eds.), Practical Reasoning: Springer Notes on Artificial Intelligence, 1996, 650-662. ■ Logique et Analyse for "Missing Premisses in Pragma-Dialectics", Logique et Analyse, 129-130 (1993), 155-168.

XV111 THE DEATH OF ARGUMENT ■ Uitgever Koninklijke Van Gorcum BV, Assen for "The Problem of Abduction", Algemeen Nederlands Tijdschrift voor Wijsbegeerte, jaargang 93 nr. 4, pp. 265-273. ■ Sic Sat Press for "Who Cares About the Fallacies?", in F.H. van Eemeren, Rob Grootendorst, J. Anthony Blair, Charles A. Willard (eds.), Argumentation Illuminated, Amsterdam: Sic Sat Press, 1992, 22-48; "Ad Baculum is Not a Fallacy", Proceedings of the Fourth International Conference of the International Society for the Study of Argumentation, in Frans H. van Eemeren, Rob Grootendorst, J. Anthony Blair and Charles A. Willard (eds.), Amsterdam: Sic Sat Press, 1999, 221-224; and "Ad Hominem: From Johnstone to Locke to Aristotle", in Frans H. van Eemeren, et al. (eds.), Analysis and Evaluation, Amsterdam: Sic Sat Press, 1995, 295-408. ■ Editions Rodopi BV for "Secundum Quid as a Research Programme", in E.C.W. Krabbe, et al. (eds.), Empirical Logic and Public Debate, Amsterdam: Rodopi, 1993, 27-36. ■ Penn State University Press for "Dialectical Blindspots" Philosophy and Rhetoric, 26 (1993), 251-265; and "Appeal to Force" in Hans Hansen and Robert Pinto (eds.), Fallacies: Classical and Contemporary Readings, University Park, PA: Penn State University Press, 1999, 181-196. ■ Communication and Cognition for "Pragma-Dialectics: A Radical Departure in Fallacy Theory", Communication and Cognition, 24 (1991), 43-53. ■ Walter de Gruyter GmbH & Co. KG for "Ad Baculum, Self- Interest and Pascal's Wager", in F.H. van Emeeren et al. (eds.), Argumentation: Across the Lines of Discipline, 1987, 343-349.

Prologue In 1970, the Australian logician C.L. Hamblin published a book entitled Fallacies. It was a bombshell. Hamblin marshalled some very strong criticisms against what he called the Standard Treatment in fallacy theory. Here is part of that attack: We have no theory of fallacy.... In some respects... we are in the position of the medieval logicians before the twelfth century: we have lost the doctrine of fallacy, and need to rediscover it [Hamblin, 1970, p. 12]. Speaking of the treatment of fallacies in logic textbooks at the time of writing, Hamblin insisted that what we find in most cases... is as debased, worn-out and dogmatic treatment as could be imagined — incredibly tradition-bound, yet lacking in logic and in historical sense alike, and almost without connection to anything else in modern Logic at all [Hamblin, 1970, p. 12]. Hamblin's criticisms proved to be a fruitful provocation. To a large extent, three intellectual developments can claim some degree of origination in Hamblin's goading. These three are informal logic, fallacy theory and argumentation theory. A good sense of the development of informal logic can be got from Ralph Johnson's The Rise of Informal Logic [Johnson, 1996], as well as Fundamentals of argumentation theory [van Eemeren et al., 1996]. Fallacies: Selected Papers, 1972-1982 by John Woods and Douglas Walton [Woods and Walton, 1989] renders a similar service for fallacy theory, as does Hansen and Pinto [Hansen and Pinto, 1995]. A useful entre into argument theory is the collection of essays, Studies in Pragma Dialectics, edited by Prans van Eemeren and Rob Grootendorst [van Eemeren and Grootendorst, 1994]. It is hard to know what Hamblin would have thought of the work done on fallacies during the past thirty years, but there can be little doubt that he would not in the year 2003 see occasion to lodge the same criticisms that he

XX THE DEATH OF ARGUMENT levelled in 1970. Fallacy theory may not yet have achieved the theoretical maturity that Hamblin called for, but it has come a long way from the despised Standard Treatment. The present book is a further attempt to take the fallacies research programme in a direction which, as I hope, Hamblin might have approved of. I am a traditionalist about fallacy theory. I accept the traditional notion of fallacy as an argument or piece of reasoning which seems to be good in a certain way but isn't in fact good in that way. I also tend to think of fallacy theory as a branch of logic. I am moreover something of a traditionalist about logic. I yield to no one in the respect I have for modern mathematical logic in its four princely domains of model theory, proof theory, set theory and recursion theory. But no one seriously believes that modern mathematical logic is much good for the analysis of real-life argument and reasoning, nor that it is the best place in which to transact the business of fallacy theory. Mathematical logic has achieved a hegemony that is scarcely a hundred years old, whereas logic itself has been a dominant intellectual force in the tradition that ensues from Athens for two and half thousand years. What logic has always been until comparatively recently is the theoretical core of general theories of argument and inference. It was so with the first logician and remained so, with little variation, until Prege. One thing that the old logic could not do is capture the structure of mathematics, not even Greek mathematics. Centuries later Descartes would celebrate algebra as a universal method of mathematical problem solving. After 1879, adequate logic would inherit this hegemonic character. It would purport to be wholly for the representation of the structure of mathematics. In this fateful transformation, logic lost its customary tie to the agora. There is some evidence that the mathematical hold on logic is relaxing its grip. Informal logic, investigations into critical thinking, argumentation theory and fallacy theory itself are underwriting research programmes of considerable power. Computer scientists have led the way in re-jigging logic in ways that cater to more realistic models of actual human practice. Default logic [Reiter. 1980], nonmonotonic logic [Sandewall, 1972] and [McCarthy, 1980] and autoepistemic logic [Moore, 1985] are all products of work done by computer scientists. Even mainstream logicians have been involved in making logic more realistic, what with developments in dynamic logic [van Benthem, 1994], labelled deductive systems [Gabbay, 1996], situation logic [Barwise and Etchemendy, 1984], paraconsistent logic [Batens et al., 2000], among numbers of other such developments. Nor should we overlook the burgeoning literature

xxi on dialogue logic ([MacKenzie, 1990; Girle, 1993; Girle, 1996; Walton and Krabbe, 1995; Lorenzen and Lorenz, 1978; Gabbay and Woods, 2001c; Barth and Krabbe, 1982], to name just a few). Mine therefore is both a more traditional and a more liberal view of logic. (A more detailed account of this conception than I have space for here can be found in Woods, Johnson, Gabbay and Ohlbach [Woods et al., 2002].) One of the virtues of this liberalism is that it makes wholly unnecessary (in fact, wholly impossible) fruitless wrangles about whether fallacy theory is logic or something else (dialectic, for example). And if we heed Hamblin's wish that fallacy theory achieve at least some sort of connection with modern logic,1 we have taken a stand against those who think that ordinary reasoning and everyday argument should not be subjected to accounts that are technically theoretical. Though I favour the more traditional conception of logic, I am not inclined to be greedy about it. I am far from believing that everything good in argumentation theory and fallacy theory is logic. Though my notion of logic is already something of an interdisciplinary concept, the disciplines required for the production of deep and mature theories of human cognitive and argumentative practice significantly outreach the stretch of anything deserving of the name of logic. I welcome the need for interdisciplinary cooperation, linking up all branches of logic with computer science, cognitive psychology, neurobiology, forensic science, linguistics (including conversation analysis and discourse analysis) and argumentation theory. It is important that we be aware that there has been a considerable conceptual shift in fallacy theory since Aristotle's original classification in On Sophistical Refutations. In some cases the name persists while what it denotes has changed. Aristotle defined a fallacy as an argument that appears to be a syllogism, but isn't a syllogism in fact. But no one today, even those who take the traditional approach, would define a fallacy in just this way. This is because hardly anyone is much inclined to think the theory of syllogisms as a serious rival of modern systems of logic. (Perhaps the dismissal is too abrupt. The logic of the syllogism was the first intuitionistic, relevant, nonmonotonic and paraconsis- tent logic ever. See below). Similarly, present-day accounts of the ad hominem fallacy have nothing to do with Aristotle's discussion (interestingly the ad hominem did not make Aristotle's original list of thirteen) or with Locke's treatment of it [Locke, 1961]. Nor do our accounts of begging the question have the same breadth as Aristotle's, or our ac- 1A case in point is Woods and Walton's modelling of question-begging in the Kripke-semantics for intuitionistic modal logic [Woods and Walton, 1989, ch. 10].

xxii THE DEATH OF ARGUMENT counts of the complex question fallacy bear the slightest resemblance to what Aristotle said about it. Then, too, the inductive and statistical fallacies are comparatively new, arising in the 17th century with the birth of probability theory and the emergence of inductive methods in science. (See here [Woods, 1999b; Woods, 2002a; Woods, 1999d; Woods, 1999a; Woods, 1999c].) 'Dialectic' is also a word with a history. In Aristotle's On Sophistical Refutations, a dialectical argument is a question-and-answer exercise with respect to some received view or dialectical proposition. A dialectical proposition is one that expresses an endoxon. An endoxon is an opinion held by everyone, or by most people, or by experts. Today a dialectic argument is generally understood as one in which an issue is in dispute, with the parties attempting to prevail on each other with regard to that issue. However, in some cases, modern theorists understand a dialectical argument as little more than a dialogue in which the parties are having a rational discussion. (See here [van Eemeren and Grootendorst, 1984] and [Hamblin, 1970, p. 320].) When names don't change, underlying conceptual change is likely to go unnoticed, except for investigators who have tried to master the history of fallacy theory. Since these are patently in the minority, bloopers are bound to be made. One is that John Stuart Mill charged that every deductively valid argument committed the fallacy of begging the question. Aristotle considered the possibility, and then rejected it; and Sextus Empiricus made the complaint and stuck by it. But not only did Mill not make the claim, he actually and emphatically denied it. Another blooper is one in which Locke is credited with originating the terms ad verecundiam and ad ignorantiam (and borrowing the term ad hominem), which he then used to denote the fallacies that have borne those names ever since. But Locke in fact does not think that arguments of these kinds are fallacious at all. Conceptual shifts and historical inaccuracies can mar a theorist's best efforts. I have tried to avoid these difficulties here, but I can't claim a perfect result. Conceptual shifts and historical mistakes are a little like fallacies. They don't in the general case announce themselves. A word or two would be in order about the title of this book. At 176a10 — 12 of Sophistical Refutations Aristotle writes as follows: For it is possible for it to be true to say 'Yes' or 'No' without qualification to countless different questions; but still one should not answer them with a single answer for that is the death of argument ([Barnes, 1984, p. 299]; emphasis added). Aristotle is here discussing the fallacy of several questions or, as it has come to be called, the fallacy of many questions or complex ques-

XX111 tion. In its modern sense, the many question fallacy is exemplified by queries such as, "Have you stopped beating your dog?" in the context of interrogation arguments in which questions must always be answered by 'Yes' or 'No'. On some accounts, the fallacy involves the question's presupposition, "You have beaten or used to beat your dog". If the question is allowed to stand, then since it must be answered 'Yes' or 'No', each answer presupposes that the answerer has beaten his dog. In this way, the questioner commits the respondent to a proposition which the respondent has not conceded (and in most real-life cases would emphatically reject if he could). This is not, however, Aristotle's analysis. His view is that [t]hose fallacies that depend upon the making of several questions into one consist in our failure to articulate the account of a proposition. For a proposition predicates a single thing of a single thing .... Now since a deduction [= syllogism] starts from propositions and a refutation is a deduction, a refutation, too, will start from propositions [Barnes, 1984, pp. 286;169a6 - 14]. Aristotle's example is the question "Are Coriscus and Callias at home or not at home?" But what goes wrong with this question? To understand Aristotle's position it is necessary to emphasize that he is discussing the thirteen fallacies in the context of a kind of argument called refutation. In a refutation there are two participants, a questioner and an answerer. The answerer puts forward a thesis T (e.g., that all virtues can be taught), which he then defends by answering critical queries put by the questioner. Aristotle requires that the answerer's responses be either 'Yes' or 'No' and therefore that the questioner's questions be answerable in this way. The further task of the questioner is to take the concessions made by the answerer as premisses and to construct from those premisses a syllogism whose conclusion is not-T, i.e., the contradictory of the thesis that the questioner was attempting to defend. Syllogisms for Aristotle are a special kind of valid argument. They are valid arguments that satisfy additional constraints. One is that the argument contain no redundant premisses (which make syllogistic logic a kind of nonomontonic relevance logic); another is that no premiss may occur as the conclusion, and derivatively, that premiss-sets must always be consistent (which makes syllogistic logic a paraconsistent logic); a further requirement is that syllogisms not have multiple conclusions (which makes the system a kind of intuitionistic logic). (See [Woods, 2001].) A refutation, then, is an argument that satisfies these conditions (among others) and whose conclusion is the contradictory of the questioner's original thesis. A further requirement imposed by Aristotle on syllogisms is that they be constructed from a special kind of statement, for which Aristotle reserves the technical term "proposition". As we have

XXIV THE DEATH OF ARGUMENT seen, a proposition is a statement that predicates a single thing of a single thing. Propositions therefore cannot be molecular constructions of pairs of statements. If we examine the question that Aristotle discusses, "Are Coriscus and Callias at home or not at home?", it is clear that it admits of a 'Yes' or 'No' answer. The 'Yes' answer concedes to the questioner the statement, "Coriscus and Callias are at home or are not at home". A negative answer concedes to the questioner the statement, "It is not so that Coriscus and Callias are at home or not at home". Neither of these statements is of any use to the questioner. They are not propositions. The first is equivalent to "Coriscus is at home and Callias is not at home, or Coriscus is at home and Callias is at home, or Coriscus is not at home and Callias is at home, or Coriscus is not at home and Callais is not at home". And the second is equivalent to the negation of the first. Since they are not propositions in Aristotle's technical sense, they are ineligible for use in constructing syllogisms; hence they cannot be used in constructing a refutation. This leaves us with two rather pressing questions. Why did Aristotle think that he was justified in requiring that syllogisms be constructed exclusively out of propositions? And what is it about the concession of non-propositions in the context of refutation-arguments that justifies Aristotle's passionate denunciation? ("the death of argument"). The first question is more easily answered than the second. Aristotle imposes the proposition-only condition on syllogisms because he holds the thesis of propositional simplification. According to this thesis, anything stateable in any natural language is also stateable without relevant loss in the sublanguage of propositions. Propositions in turn are statements of the form "All A are B", "No A are B", "Some A are B" and "Some A are not-B". If the thesis of propositional simplification is true, this works a substantial economy into Aristotle's logic. For Aristotle can now claim that he can capture a complete theory of deduction in this very economical sublanguage of propositions. Aristotle states the theory at On Interpretation 17a13,18a19ff. and 18a24 [Barnes, 1984, pp. 26-28]. There is no known extant proof. Our second question asks for what justifies Aristotle's claim that allowing non-propositions into refutation-arguments would be the death of argument. Aristotle does not answer this question, or anyhow does not answer it in any direct way. Even so, an answer can be conjectured. Take as a second example the rather more interesting (and modern-sounding) "Have you stopped beating your dog?" It is equivalent to "Do you have a dog you have beaten and, if you have a dog you have beaten, have you stopped beating it?" As Aristotle correctly notes, the question is answerable 'Yes' or 'No'. If the answer is affirmative it concedes the

XXV statement "I have a dog that I have beaten and I still beat it". But now consider what a negative reply gives us: "It is not the case that I have a dog that I've beaten and that I still beat it", or symbolically, not-((D and B) and S). Let me say again that Aristotle rightly recognizes that this statement could be true, that it could be a fact that not-((D and B) and S). But it is not a fact that the questioner can make use of as long as he is interested in determining the questioner's status as a (potential) past and present dog-beater. Imagine a dialogue in which every question were complex in this sort of way, and every answer were (as might well be the case in real-life interrogation-argument) in the negative. Then it would be impossible for the questioner to get to the particular facts of the case at hand. This indeed would be the death of argument with regard to any thesis which itself is a proposition (e.g.,"The questioner is not a violent person"). If every fact of the case is complex in the kind of way that the fact not-((D and B) and S) is, then no fact will imply any statement in the form "The questioner is a violent person" which is the contradictory of the thesis we are presently considering. Hence no refutation can be constructed. It is arguable that Aristotle would have been prepared to extend the death-of-argument metaphor to all fallacies. There is no doubt that Aristotle thinks that fallacies are serious mistakes. This is also my own view, but it is subject to a tautologous seeming qualification: Fallacies are serious mistakes when they are indeed fallacies*. Whether we are thinking of Aristotle's original thirteen or what I call the gang of eighteen (see chapter 1), it is apparent that rarely are arguments of these various types fallacies just because they are of that type. So, for example, when an ad hominem argument or an ad verecundiam argument is a fallacy, it is not so merely because it has the form of an ad hominem or an ad verecundiam argument. That is to say, fallaciousness is not intrinsic to arguments of these kinds; and the same is true for nearly them all, whether Aristotle's thirteen or the gang of eighteen. This is both a setback and an attraction. The setback is that the fallacy theorist must be able to specify the varying conditions under which an argument of a given kind is and is not fallacious. This is more easily said than done; but doing it, even so, is filled with interesting challenges, which are the potential occasion of theoretical depth. And that is the attraction of fallacy theory. A word or two about my subtitle would also be in order. Since all reasoning is agent-based, Fallacies in Agent-Based Reasoning might strike the reader as rather pleonastic. Of course. But not all theories of reasoning make express provision for agents in their working vocabularies. In

XXVI THE DEATH OF ARGUMENT some circles, it is still widely held that the predicate calculus is the general theory of reasoning, but not even the proof theory of this approach has the means to establish theorems about what reasoning agents do. What the subtitle of this book is intended to signal is that an account of fallacies needs to be set in a more general theory of cognitive agency, a theory of how agents prosecute their alethic and strategic agendas.2 I do not here have the space to propose a complete theory of agent-based reasoning. For the most part, it looms in the background, with occasional partial articulations as may be needed, to fill out a point in the emerging account of fallacies. Some of the standard fallacies have something to do with real-life dialogues. To the extent that this is so, it falls to the fallacy theorist to have a clear-eyed view of what real-life dialogues are and of the conditions under which they are properly transacted. One of the earliest (and best) of such contributions is Aristotle's account of refutations, in On Sophistical Refutations. Mediaeval writers were logicians of considerable sophistication, and they made important headway with precise analyses of disputation games such as Obligation [Sherwood, 1963]. In the present-day dialogue logic is a growth industry. At the heart of these developments is a conception of dialogical propriety captured in what has been called the Goody Two-Shoes Model of Argument [Gabbay and Woods, 2001c]. As applied here the name is a somewhat light-hearted reflection of the idea of a person who makes too much of being good. (The ancient word "goody" is a contraction of "goodwife".) In appropriating this name, I intend no mere levity, but wish rather to dramatize the model's distinctive features. In its most basic sense, Goody Two-Shoes argument is a project of open, objective enquiry in which the parties behave (or are expected to behave) reasonably, honestly, helpfully and equably. On one reading, Grice's logic of conversation is an instance of the Goody Two-Shoes model, as is the pragma-dialectical approach to critical discussions. Much, too, of modern dialogue logic approximates to this model. In the chapters that follow, I shall explain to the unacquainted reader details of Grice's and the pragma-dialectical approaches. For the present, it suffices to say that in a number of places in this book I shall attempt to demonstrate that a robust and imaginative fallacy theory requires that we should not accord hegemonic standing to the Goody Two-Shoes model; and that there are types of argument that deviate significantly from Goody Two-Shoes and yet are objects of serious theoretical attention. 2 "Alethic" derives from the Greek and here means "pertaining to what is true" or "truth- oriented" .

XXV11 Strange as it might initially strike us, such arguments, though they are in one straightforward sense of the term non-cooperative, are not, just so, irrational or stupid. Part of their importance lies in the emphasis they lend to a distinction between what might be called alethic rationality and strategic rationality, the one an intellectual virtue and the other a practical virtue. This is a distinction which we shall repeatedly find it profitable to engage.

PART I METATHEORETICAL QUESTIONS In the interests of inter-chapter coherence, I have tried to give the book a kind of thematic organization. The three chapters that make up Part 1 are devoted to the following questions, among others. What is a fallacy? What is it that we want fallacy theory to do (what are its targets)? What resources are needed for fallacy theory and what are its principal methods? What is the connection between fallacy theory and informal logic? What is the connection, or lack of it, between informal logic and formal methods generally? What, in addition to its possible contributions to fallacy theory, is the proper business of informal logic?

Chapter 1 WHO CARES ABOUT THE FALLACIES? The main business of this chapter is to motivate the systematic study of fallacious reasoning and argument. In its most usual meaning, a fallacy is a common misconception; that is, a false statement that is widely believed. Examples of such statements are, "Handling frogs causes warts", "You'll catch a cold if you sit in a draft", and — at a more academic level — "John Stuart Mill thought that every valid argument begs the question". (Recall the Prologue!) This is not, however, the meaning of "fallacy" that has occupied logicians and argumentation theorists since the time of Aristotle. In this second meaning, a fallacy is something that characterizes the reasoning or the line of argument that gets one to assert or assent to or believe a given statement. And there is, in this second sense, no requirement that the statement that the reasoning has led to be a fallacy in the first sense. If the initial task of the chapter is to say why studying fallacies in this second sense is a valuable thing to do, I also wish to call attention to the importance of the distinction between alethic and strategic rationality, mentioned briefly in the Prologue. 1. Questions About Fallacies. There are mistakes. And there are mistakes that people actually make. The difference between them is that mistakes that are actually committed are mistakes that seem not to be mistakes to those who commit them. They are errors, blunders, and missteps whose mistakenness is unapparent. Some theorists of reasoning reserve their particular attention for a particular group of mistakes that are actually committed — or anyhow are committed with a notable frequency. They call these par-

4 THE DEATH OF ARGUMENT ticular mistakes "fallacies", a term that has been used since antiquity. In the traditional meaning originating with Aristotle in the fourth century B.C., a fallacy is a bit of reasoning which seems to be good in a certain respect but is not in fact good in that respect. The deficiency being unapparent, a fallacy then would be a mistake of some sort that people actually make, or anyhow are disposed to make. Three questions immediately arise. 1 How does a fallacy differ from other mistakes that people are inclined to make? I call this the boundary question. 2 Do fallacies have something in common, apart from the fact that they are mistakes that people make or are inclined to make with a notable frequency? I call this the common factor question. The third question is the most important of the three. 3. Apart from the fact that they are mistakes that people make or are inclined to make, do fallacies have any particular importance for human life? Are fallacies mistakes whose commission matters, and if so how? This I call the who-cares question. My aim in the present chapter is to make some headway in answering these questions. The very fact that I have set myself this task is enough to indicate to the reader my conviction that the three questions are important, and that the last is especially important. This suggests an answer to the query embedded in the title of the chapter, "Who Cares About the Fallacies?" The answer is "Everyone does (or should)". The burden of the chapter will be to explain and justify this reply. The fallacies I am thinking of in this chapter are collectively referred to as "the gang of eighteen". The term is a loose convenience of course. [Copi, 1986] treats seventeen fallacies, [Carney and Scheer, 1980] eighteen, [Schipper and Schuh, 1959] twenty-eight, [Black, 1946] eleven. Woods, Irvine and Walton discusses the following fallacies. ■ ad baculum ■ ad hominem ■ ad misericordiam ■ ad populum ■ ad verecundiam ■ affirming the consequent ■ denying the antecedent ■ amphiboly ■ begging the question

Who Cares? 5 ■ biased statistics ■ complex question ■ composition and division ■ faulty analogy ■ equivocation ■ hasty generalization ■ secundum quid ■ gambler's ■ ignoratio elenchi [Woods et al, 2000] It is worth noting that though these lists are obviously not identical, they do display a rather striking overlap. 2. The Current State of Fallacy Theory. The fallacies are as usual up to their ears in definitional and analytical dissensus. People who are distrustful of anything so portentous as fallacy "theory" often wonder why it is that the more or less standard list of them, eighteen or so, usually of Latin name and dubious origins in a tired tradition, should still command a place in contemporary effort. After all, the scope for error, misstep and misconception is vast,1 and who is to say that many a more deserving blunder does not call out at least equally for the celebrity of a name? So why these? (The boundary question). A further objection is that for all the endless attention that the fallacies attract, there seems to be no settled and credible idea of what fallaciousness is. "Even when one learns to recognize alleged examples of the various 'fallacies', it is difficult to see what common factor makes them [so]" [Lambert and Ulrich, 1980, p. 25]. (The common factor question.) Is it distinctive of them that the fallacies are bad arguments (or argument instances)2 which occur commonly or with a notable or striking frequency?3 Are the fallacies always arguments involving incorrect inferences, or is such a view, common though it is, seriously misconceived ^ee, for example, [Massey, 1975a]. Shades of DeMorgan: "There is no such thing as a classification of the ways in which men arrive at an error: it is much to be doubted whether there ever can be", [DeMorgan, 1926, p. 276]. 2 "We should retain the historical nucleus of the idea of a fallacy as a logically bad argument" [Johnson, 1987a, p. 245]. 3 Johnson "we shall introduce the notion of frequency; because a fallacy is not just any mistake in an argument, but one that occurs with some frequency" [Johnson, 1987a, p. 245]. Cf. Scriven "Fallacies are the attractive nuisances of argumentation, the ideal types of improper inference. They acquire labels because they are thought to be common enough or important enough or to make the cost of labels worthwhile" [Scriven, 1987a, p. 333].

6 THE DEATH OF ARGUMENT and itself a kind of fallacy?4 Are the fallacies always arguments (or strategic mismanoeuvres, dialectical infelicities, or whatever else they might be) which exhibit a peculiar disposition to trick or deceive?5 If so, whatever are we doing in exposing first-year students the wide world over to purported paradigms of the fallacies which would dupe no one (thus "There is an afterlife, hence there is an afterlife")? This is not asked frivolously. If a fallacy is a bad argument which appears not to be bad, then how is it possible to identify an argument as a fallacy or to give an example of a fallacy? For the example to work, the person for whom the example is intended must recognize that the argument is bad, but it must also be an argument whose badness is not apparent to him or her. So, on the face of it, to see something as a fallacy is to see it as bad and not to see it as bad. But this makes it impossible to recognize fallacies. This is the Fallacy Dilemma. It imposes on the theorist and the student of fallacies a burden of some subtlety. It is the burden of getting people to see the badness of arguments they are not inclined to see as bad. Undoubtedly among the strongest contemporary criticisms that have been launched against the Standard Treatment of the fallacies are, as we saw in the Prologue, those of Hamblin in his influential book, Fallacies. The truth is that nobody, these days, is particularly satisfied with this corner of logic... We have no theory of fallacy at all.... In some respects... we are in the position of the medieval logicians before the 12t/l century: we have lost the doctrine of fallacy, and we need to rediscover it [Hamblin, 1970, pp. 11-12]. And ...what we find in most cases, I think it should be admitted, is as debased, worn-out and dogmatic a treatment as could be imagined — incredibly tradition- found, yet lacking in logic and historical sense alike, and almost without connection to anything else in modern logic at all. This is the part of his book in which a writer throws away logic and keeps his reader's attention, if at all, only by retailing traditional puns, anecdotes, and written examples of his forbears [Hamblin, 1970, p. 12]. Hamblin is making two complaints. One is, in effect, that informal logicians, and other people who were writing about fallacies in the late 4Treating fallacies as inference-mistakes is "a colossal mistake, a super-fallacy..." [Hintikka, 1987, p. 211]. 5See Govier, "By definition, a fallacy is a mistake in reasoning, a mistake that occurs with some frequency in real arguments and which is characteristically deceptive" [Govier, 1987, p. 177], emphasis added. Cf. Woods and Walton, "a fallacy is an argument that is a tricky deception, because it is incorrect even while it has a tendency to seem correct" [Woods and Walton, 1982, p. 2]. And, more carefully, Fogelin and Duggan, "the term 'fallacy' is our most general term for criticizing any general procedure used for the fixation of beliefs that has an unacceptably high tendency to generate false or unfounded beliefs" [Fogelin and Duggan, 1987, p. 262].

Who Cares? 7 1960s, had not done a good job in solving the Fallacies Dilemma. The other is, in effect, that these same people had not done a good job in answering the three questions that I set out in the introduction to this chapter: the boundary question; the common factor question; and the who-cares question. Since, as I have already said, my aim is to answer these questions, and also to deal with the Fallacies Dilemma, it should be evident that I think that Hamblin's objections can be met. 3. Light At The End Of The Tunnel? My answer to the third question, "Why should we care about the fallacies, why are they so important?", will have something to do with rationality. Rationality for me is a down-and-dirty affair, the more or less efficient management of life's dynamic turnings with the aid of low- finite cognitive resources in low-finite time against persisting invitations to err. It seems to me quite clear that part of the main business of human rationality lies in the contribution it makes to our survival and our prosperity. The human agent is genetically and culturally endowed with skills, some of which may have been naturally selected for, and these are the basic equipment in what we might with a certain looseness call his rational survival kit It is, of course, a good question how far to press the idiom of natural selection. The human agent is a global maximizer, that is, a being who knows how to endure local setbacks in return for longer term advantages. It is debatable whether "the mechanism of natural selection... can [always] simulate rational human action directed toward a purpose or aim" [Popper, 1972, p. 267]. This is not to say that human agents wait passively by for new mutations to occur; in fact "one can talk of many, if not most, new cultural elements occurring by virtue of the fact that they are needed" [Ruse, 1974, p. 435]. When I speak here of our evolutionary endowments, I intend reference to rational skills that either have been naturally selected for or have come to us by way of cultural evolution, that is, because they "were needed". What has this to do with fallacies? I am struck by something that Michael Scriven said, that the fallacies are a kind of ideal type of attractive nuisance or impropriety [Scriven, 1987a]. They are, I think, at least that. And as they have come down to us in the tradition they are also a kind of caricature of their associated improprieties, which lie deeply embedded in human practice. Certainly, "There is an afterlife; therefore, there is an afterlife" is nothing if not a caricature, and it is a perfect illustration of the Fallacy Dilemma. Good caricatures have a certain memorable vividness, and this is a helpful feature of them when it comes to what I think is the most distinctive aspect of the fallacies

8 THE DEATH OF ARGUMENT that the tradition has preserved for us at mid-century. For I think that the fallacies are idealized caricatures that are also symptoms of the sorts of things that can go wrong with those skills that make up our rational survival kits. So my contention will be that the fallacies that the tradition has thrown up are a highly idealized, vivid symptomatology of basic rational misperformance. If this is right, then it won't suffice to say that the fallacies are worthless as they stand — overdrawn, unrealistic, and unanalyzed (though often they are indeed these things). Thus construed, the fallacies would fare, well or badly, with other idealized symptomatologies, such as the one-month and two-month commercial bill discount rates that are loosely symptomatic of Japan's money supply rising too quickly. It follows from this that the importance of the fallacies is derivative. They owe their importance to the importance of survival and to the contribution thereto made by those skills with which they are associated by caricatured misperformance. What are these skills? Without any pretence to completeness, it seems that they will include at least the following (however imperfectly they may be realized in practice): ■ The ability to see the significance of what goes on and, in particular the ability to draw consequences. ■ The ability to generalize from positive instances. ■ The ability to make primitive inductions. ■ The various routines of epistemic cooperation and, in particular, knowing when to rely on the say-so of others. ■ The capacity to generate (not just receive) information. ■ The means of discerning type-ambiguities in nature. ■ The rudiments of cost-benefit protections of self-interest, including measures for conflict resolution. No doubt other candidates could be furnished for inclusion in this list; nor are all the skills there displayed strictly disjoint. I do think, however, that these are basic in the sense that they are necessary, that human survival would be deranged without them. Supposing this to be so, the question recurs: "What has all this to do with the traditional fallacies?" I am going to propose that the traditional fallacies are errors or missteps which involve the misperformance of one or other of these basic skills; and, correspondingly, that the importance of fallacies is in part at least parasitic on the importance of these skills. Of course, it

Who Cares? 9 would be ludicrous to conceive of fallacies as mistakes that are intrinsically fatal or some near thing. (No coroner will give "fallacy" as the cause of death!) Not every misperformance of a fundamental skill has the importance that attaches generically to the skill itself. The connection that I am suggesting is subtler and more indirect. The items that make up our rational survival kits are largely constitutive of what it is for beings like us to be cognitive agents; and they go some way toward defining our rationality. And they are topic-neutral. So, on the present suggestion, one thing that fallacies have in common is that they are violations of comparatively primitive, topic-neutral cognitive norms. Fallacies in turn imbibe this topic-neutrality. A fallacy is not a mistake in physics or political science. It is a mistake period}. When it is committed in physics or political science, it is not an error intrinsic to those disciplines. I should note in passing that there are other skills whose misperformance lands us in fallacies, and other norms whose violation does so as well. Consider in this regard the important work of Robert Ennis (e.g., in his classic textbooks [Ennis, 1969b; Ennis, 1969a]). My list of skills is not a rival of Ennis', but rather a complement. Let us turn to some examples. A. In the business of survival, we will fare badly unless we know how to draw consequences. If we cannot draw consequences at all, we cannot avert disastrous consequences. And associated with this is the fallacy of the non sequitur. B. We will not survive if we cannot generalize from samples, that is, generalize in confirmation-theoretic ways. The inability to project the "right" correlation is a fundamental failure. Equally, an over-zealousness for projecting coincidences can be ruinous, as in the generalization from a single observed case of food-poisoning with attendant epic discouragement of nutrition across the board. Thus, concerning generalizations, the gamut goes from null to hasty, for which latter the tradition offers us the slovenly caricature, "Cynthia is a bad driver; so women are bad drivers". C. Not every generalization from a single case is a fallacy. In fact it is sometimes a blunder and worse not to generalize in this way, such is the role of "primitive inductions". The toddler on touching the element on his mother's kitchen stove learns in one case never to do that again. It is a remarkable feat and, it may seem, a full-blown generalization. 'Never' quantifies and 'again' projects. Here is a situation in which instantial experimentation would be, to the say the least, ill-advised, as if the toddler's confidence in

10 THE DEATH OF ARGUMENT the generalization had to await fifty or so repetitions of the encounter (and the early retirement of his poor hand from the pool of his motor resources). A mother noticing any such affection for the experimental method would at once summon the assistance of her pediatrician. And she would, of course, be nothing but right to do so. Here, too, the tradition reminds us of the importance of analogy-recognition in rational inductive practice, and intensifies the reminder with a caricature (oftener than not of misperformance rather than non-performance). Thus the silly but harmless, "It's perfectly all right for moose to go about without clothes, so why not us?" (It may, by the way, be preferable to detach primitive inductions from induction proper and assimilate them outright to analogical inference. Even so, it is doubtful that the link with induction would be altogether severed; for "[a]nalogy thus pictured is an inferential leap, whereof the top of the trajectory is a slurred-over induction") ([Quine and Ullian, 1970, p. 61], emphasis added). Before going on with examples, it might be helpful to pause and to say again what I am trying to establish. We notice, first of all, that in virtually every treatment of the fallacies in the last hundred and fifty years or so, there is substantial evidence of a consensus that the traditional eighteen are possessed of sufficient wherewithal to grant them a non-trivial place in our speculations. Why, given that there are, as DeMorgan has said, many more ways of going wrong than any taxonomy can capture, do these recur in the literature, generation-in and generation-out? Is it simply a matter of intellectual conventions which we find lazily congenial? It is also noticeable that, though there are some exceptions, the traditional lore about the gang of eighteen is dominated by brief and next-to- empty descriptions, together with examples of striking simple-mindedness These I have been calling caricatures. Caricatures are meant to be simple-minded, and my point is that this is their principal function in the traditional accounts of the fallacies, and that in its most recurring aspects traditional fallacy "theory" is exemplary rather than analytic. It has largely been the case that traditional fallacy theory teaches whatever it does teach by example; and simple examples are more easily remembered than complex ones. So much for fallacious caricature. But what explains the importance and persistence of the gang of eighteen? Again, this is explicable as follows: the standard examples caricature (because they over-simplify) improprieties of rational performance of which they themselves can be considered a kind of idealized symptom. Those mis-

Who Cares? 11 performances in turn are failures of one or another of our basic rational survival skills. If this is a correct picture, it should be discernible in further examples. Let us see. D. It is imperative that we make causal inferences. Crossing the streets of Amsterdam is a guaranteed disaster if the physics of traffic is construed in accidental, so to speak, correlations. Thus, post hoc, ergo propter hoc. E. We need to make judgements of relevance. If the gunman has fired his pistol six times, should Dirty Harry try to infer whether the gun is empty (is it a Luger, perhaps?), or is he free to infer the completeness of the propositional calculus? The failure of relevance would be lunatic (though the implication, if he lived long enough to draw it, would be just fine). Well, it would be truth-preserving. Thus, fallacies of relevance. F. We need to engage in routines of cooperative rational acceptance, of learning when to accept and act on the say-so of others. Was it Peirce who said that I know when my birthday is and who my parents are by hearsay? Who can pretend to any knowledge of current affairs without newspapers, or of science without scientific textbooks, or of the whereabouts of Central Station without asking Suzanne? So we have the ad verecundiam, the ad hominem, and the ad populum (that version of it in the form, "People say ... ", or "It is common knowledge that ..."). And we have caricatures: "Fifty million Frenchmen can't be wrong, so policies in Micronesia are fine". G. It is necessary to know how to command or summon up information in a timely way. An efficiently managed life cannot be one of ungoaded informational passivity. We need to engage in techniques of information-seeking, including the asking of questions. Bleeding from an artery, it is insufficient to wait and see whether one's perceptual field will eventually register a sign reading "Hospital fifty metres ahead". One needs to ask. And so we have asking-fallacies such as many questions: "Have you stopped beating your dog?" H. We must be able to recognize and to adjust to type ambiguities in nature. If our toddler, on burning his finger on the kitchen stove, quits the kitchen forever, and the house, and the neighbourhood, and the county ("Sussex is too rough for me!"), he is defeated

12 THE DEATH OF ARGUMENT by part-whole asymmetries. The point is that even a house can't be figured out at all until the basic part-whole asymmetries are cottoned onto. Our toddler needs to see that though the house is safe and cozy, the kitchen is differentially tricky. And so, we speak of the fallacies of composition and division. I. Elementary routines of cost-benefit analysis are needed for the protection of self-interest. This is prudence, seriously underestimated by Oscar Wilde as "that bitch-goddess of incapability", the recognition of where one's interests lie and of what will and won't conduce to their realization. Whereupon the ad baculum enters the picture: "Your money or your life!" 4. Their Importance. The fallacies that the tradition has given us have been victimized and traduced for failing every manner of task that it was not theirs to perform. They are not decision procedures, and not complete analyses of the actual reasoning practice. It is perfectly all right if someone wants to pursue such analyses — encouraged by the idealizations to muck in empirically and find out how induction, for example, actually does work, and in what ways it can systematically misfire (see below). But this has not been the mandate of the traditional typological accounts of the fallacies. The propositional calculus allows us to identify certain invalidity- schemata which might interest us for one reason or another. Some of these have names. So, "If A then B, not-A, not-B" is the fallacy of denying the antecedent; and "Not (A and B), therefore not-A", though it hasn't a name, so far as I know, does have a description: "the fallacy of distributing negation through conjunction". Though structurally more lucid than many of the fallacy schemata we have been discussing, these too are idealizations, caricatures of the real thing. They are no more an account or a theory of actual deductive malpractice than a traditional description of the petitio is a determination of the admissibility of certain patterns of explanation in paleontology.6 So I conclude that the traditional fallacies have been unfairly knocked, though sometimes, as we have seen, their associated descriptions are over-stated or just wrong. We also now have an answer to the boundary question, "Why these!" Because they are symptoms of errors that afflict the basic rational skills needed for survival. And, in so saying, the common factor question can likewise be put to rest. 6See Harper in reaction to the Woods-Walton treatment of the petitio in Kripke-structures [Harper, 1980].

Who Cares? 13 The idea of rational misperformance covers at least two different kinds of cases, what we might call misfirings and abuses. Misfirings indicate malfunctions of the cognitive apparatus: not being able to manage the distinction between necessary and sufficient conditions; processing difficulties with nested negations; confusions (sometimes) over conjunctive probability; confusions of primitive inductions with instantial inductions; and so on. Abuses lie closer to the plane of conscious manipulation. Thus resorting to an irrelevant ad hominem rather than being taken in by one; exploiting another's confusion of necessary and sufficient conditions; tripping up an adversary with incorrect conjunctive probability determinations; and so on.7 With the distinction at hand, we can now capture another intuition about the fallacies, the idea that they are tricks or deceptions. So they are, dialogically speaking, when one speaker commits a fallacy knowingly with a view to having the other speaker accept it innocently, with intended contamination of his commitment store. I have also been assuming without argument what some readers may not be prepared to grant, namely, that the cognitive skills required for human survival involve rules and strategies which unfailingly qualify as rational. It may be, however, that sometimes such manoeuvres turn out to be in some interesting sense incorrect For example, hasty generalization is sometimes a prudent strategy, especially when the risks are high, even though the haste of the generalization might attract a charge of fallaciousness. If fallacies are violations or failures of rationality, then we could expect to have it that survival skills are sometimes exercised successfully but not rationally. Robert Fogelin seems to be making a similar point when he says: "human beings seem endowed with innate capacities for messing things up above the simplest levels of complexity. (This is particularly true of inductive inferences where, for example, the tendency for hasty and unfounded generalization seems to be hardwired into the human brain)" [Fogelin and Duggan, 1987, p. 4]. But, as Thomas Hurka has suggested, in conversation, sometimes it is the better strategy to conjoin probabilities non-multiplicatively, even though it violates the multiplicative axiom of the probability calculus. The conjunction axiom of the probability calculus provides that the probability of a conjunctive proposition "A &; B" is the probability of A multiplied by the probability of B (provided that A and B are probabilistically independent). But consider the following problem. Suppose that Lisa graduated in sociology from an Ivy League university, and that 71 am grateful to Scott Jacobs for discussion of these matters.

14 THE DEATH OF ARGUMENT she is an active feminist and environmentalist. Which of the following two statements is more probable? ■ Lisa is a Republican ■ Lisa is a socially moderate Republican If you selected the second instead of the first, your choice agrees with that of test subjects in psychological experiments. But it violates the conjunction axiom of the probability calculus. To see how, suppose that the probability of Lisa's being a Republican is 0.3 and of being a social moderate is 0.6. Then the probability that she is both is 0.3 x 0.6 = 0.18. Jonathan Cohen has argued in several places that although non-multiplicative conjunctive probability behaviour does violence to a concept of probability with which we associate the name of Pascal, it nevertheless honours a notion of probability with which we associate the name of Bacon and that sometimes this latter is the concept of probability which it is more appropriate for people to act upon.8 As for hasty generalization, it may be that it is not always fallacious (see item C, Section 3), and so it is likely that a condition on rational generalization will be something like this: generalizations may be made more or less hastily depending, among other things, on the risk that attaches to inductive caution. Stephen Stich points out that rats are notorious hasty generalizers about toxicity [Stich, 1985]. See also, [Garcia et al., 1972]. They will refuse food tasting like any food ingested shortly prior to any sickness. Stich concludes that doing so probably equips that rat with more false beliefs than true, yet nevertheless the haste and breadth of the generalization will be indulged by evolution. It seems clear that there are two ways of looking at such situations. We could say, as Stich himself does, that there is, in evolutionary terms, nothing particularly crucial about truth. Or we could say that there are situations in which it is best to risk or make certain sorts of errors so as to avoid even more critical errors. 8See, for example, Cohen. Pascalian probability (also called aleatory) is the sort of probability embedded in games of chance, such as roulette. One of the founders of the mathematical theory of this concept was Blaise Pascal (1623-1662), hence the name. "Baconian" probability is named after Francis Bacon (1561-1626), one of the earliest, and staunchest, inductivists of modern science. Baconian probability is the degree of likelihood conferred on general propositions by their true positive instances. Cohen's submission is that these are different concepts of probability, and that each has (a contextually qualified) place in our cognitive lives [Cohen, 1989]. I also note that among the alternative formalisms for representing uncertainty is what some AI theorists call possibility theory (and which makes essential use of fuzzy set theory) as well as Dempster-Shafer theory (which permits the probabilities of disjoint sets to sum to more than 1.)

Who Cares? 15 The second seems to me the better lesson to draw; and if this is right, it is some encouragement for thinking that in certain cases broad and hasty generalization are the more rational strategies, never mind that in some perfectly allowable ways they are (or might be) errors. Whether we decide to align the idea of fallacy with such errors is basically a question of whether to interpret fallacies as prudential or strategic failures rather than cognitive ones, and when to do so. In any case, whatever the better option might be, it leaves undisturbed the fact that sometimes — often in fact — the strategic error and the cognitive error coincide, that the strategic error inheres in the falsehood of the generalization. And so we can continue to say that the hasty generalizations of the Standard Treatment are reminders that things can go wrong with inductions and that sometimes what goes wrong is the embrace of cognitive (or epistemic) error for the sake of misplaced strategic caution. We shall note, however, that what we have just said does not determine that our toddler's one-shot generalization, though correct strategically (as it certainly is), is incorrect, or anyhow not as correct, cog- nitively or epistemically. Our description of the case does not require that we postulate for the toddler any more epistemic confidence in the generalization than that it might be true. This suffices to ground his determination not to act on its contradictory. Before leaving this matter, there is a related issue raised by Isaac Levi, also in conversation. Fallacies, I say, are idealized symptoms of misperformance of rational skills necessary for human survival. Levi's concern is that I leave the account underdeveloped. For how are we to understand the idea of misperformance without a background theory of rationality which is constructable independently of any particular account of the fallacies? Fair enough, but it is not strictly to the point of the present endeavour. I mean to explain the importance and persistence of the gang of eighteen. For this I need the concept of misperformance. For present purposes, I am prepared to take it as primitive. This would be a bad choice were misperformance, unanalyzed, obscure to the point of explanatory impotence. I don't think that such is the case, nor do I think that Levi thinks so either. So for the task at hand I shall let things be. (Hasty generalization is also the topic of chapter 19.) 5. Is That All There Is? I have been suggesting that the importance of the gang of eighteen derives from the importance of the cognitive skills whose failure or misperformance they caricature. If this were the whole story, it could fairly be said that it wouldn't make the fallacies interesting enough to justify

16 THE DEATH OF ARGUMENT the attribution of importance to them. I think that this is right. The case for the importance of the fallacies needs to be strengthened. It would be strengthened — in fact made — if we were able to justify a further claim, namely, that the traditional fallacies are snares and delusions. Let me turn to this now. It is surely no stretch of the term to say that people are often deluded by complexity. There is evidence galore, some of it contentious, that human beings handle complexity badly. Nested negations, quantifier iterations,9 the compounding of probabilities and the routine manoeuvres of induction all tend to go wrong in interesting and experimentally discernible ways.10 What is striking about these errors is that everybody makes them, they are confidently made and they are hard to correct. Fo- gelin posits a genetic defect, a default in the cognitive hard-wiring. The idea of deludedness, thus explained, catches in its net a goodly selection from the gang of eighteen: all the inductive and statistical fallacies, affirming the consequent and denying the antecedent. In the Standard Treatment we find caricatures of this kind of cognitive malpractice and it is interesting and important because it is delusional. We now have the resources to hand to cash Scriven's engaging metaphor, in which fallacies are "attractive nuisances of argumentation" and "ideal types of improper inference". Such misperformances are attractive because they are natural to commit and they are confidently committed; they are nuisances because they are errors. They are ideal types of error exemplified in their caricatures; and they are improper because, again, they are errors. So much for delusions. Snares are another matter. If some of the traditional fallacies are delusional because, and in the sense that, they caricature delusional performance, others are snares because, and in the sense that, they caricature cognitive effort of a kind that entraps the theorist, not the quotidien reasoner. All of the inductive fallacies saddle the theorist with the necessity to say something helpful about the problem of induction. The causal fallacy theorist must stand ready with solutions to various issues in the theory of causality. Unless these things are done, the theorist is all but guaranteed to have little of theoretical interest to say about 9Those familiar with the basic AI literature will be aware that von Neumann has a celebrated argument about the complexity of functions, an impediment to information-processing by humans which computers manage to overcome. Nested functions are the problem. The problem extends to existential quantifiers by way of Skolem functions with which they can be replaced [von Neumann, 1958]. 10See the references herein to [Fogelin and Duggan, 1987; Cohen, 1989; Stich, 1985; Garcia et al., 1972]. See also [Nisbett et al., 1983; Nisbett and Ross, 1980; Tversky and Kahneman, 1983; Cohen, 1981; Holland et al., 1986].

Who Cares? 17 his fallacies. The theorist is thus entrapped by difficult and intractable philosophical questions. It is not for nothing that the inductive and causal fallacies did not have much of a place in the tradition until after Hume's ransacking of induction and causality. For example, not much was made of inductive fallacies by Aristotle. Aristotle does discuss secundum quid, but in an desultory way, as the mistake of not qualifying a generalization. It was probably not until Antoine Arnauld and Pierre Nicole in their Port Royal Logic [Arnauld and Nicole, 1996]n that faulty inductions were categorized as a species of fallacy, and it was not until Mill's A System of Logic (Book V) that these faults acquired their central importance in the evolving tradition. Composition and Division give the theorist no particular worry about whether to bet on the Calgary Flames just because all of its members are good players. Rather, the theorist is faced with the necessity of getting clear the part-whole relationship [Woods and Walton, 1989, ch. 8]. He might also be held to the task of helping with, "Is my nose the mereological sum of its atoms?" [Putnam, 1988, p. 12]. Putnam imagines Kripke's reply as "since your nose could have consisted of different atoms, it has a modal property the group of atoms does not. So your nose is not identical with the group of atoms" [Putnam, 1988, p. 12]. And more inexorably still, when does the term "thing" compose and when does it divide? To ask this is to be well on the way to asking for a theory of (Aristotelian) substances. The ad ignorantiam appears as the problem of inference to the best explanation. We might agree that Max Planck's postulation of quanta produces an entirely good derivation of blackbody radiation; and we might further agree that this is a reason to infer the truth of Planck's conjecture. But, then, why not also infer the existence of those traditionally despised ghosts, for which there is also no evidence, as theoretical entities which explain poltergeist phenomena? And if inference to the best explanation is all right, the theorist can be expected to explain why (a) P explains Q (b) Q (c) A P is fine, while (d) If P then Q (e) Q (0 ••• P 11 Similarly, the statistical fallacies came to light with the emergence of probability theory. We find the gambler's fallacy, for example, in [Laplace, 1951, p. 163].

18 THE DEATH OF ARGUMENT is a notorious fallacy. Mind you, lots of people think that inference to the best explanation isn't all right, after all [Duhem, 1969; Laudan, 1981; van Praassen, 1980].12 Even so, I hold that the fallacy theorist cannot pretend to a good account of the ad ignorantiam without taking a principled stand on inference to the best explanation. The ad verecundiam faces the theorist with the task of explaining the legitimacy of expertise in our communal epistemic practices. He or she will need, among other things, a theory of rational co-operation [Axelrod, 1984; Woods, 1989a]. He or she might also be expected to pronounce upon the Principle of Charity (co-operation theorists having appropriated the term from the precincts of radical translation). (See chapter 14 below.) All these cases, and more, answer well to the idea of entrapment. It is the theorist who is trapped, not the everyday practical reasoner, and in every case the theorist will have been snared by a hard problem bequeathed to him by the philosophical tradition which oftener than not is only guide to such matters. In so saying we have the basis for resisting Arnauld's words which are a kind of rallying cry for legions of writers for whom the notion of fallacy "theory" is anathema. If what I've been saying is true, then not only is the gang of eighteen important, but Arnauld misses the point, and those who follow him are deeply misled by his having done so. But reason does not find its principle use in science [=theory]; to err in science is not a grave matter, for science has but little bearing on the conduct of life [Arnauld and Nicole, 1996, pp. 265-266]. Let me sum up the argument to date. The traditional fallacies caricature epistemic (and sometimes strategic) misperformance of skills necessary for our survival. Thus far, they are important because those skills are important. Moreover, many of the gang of eighteen portray defects which apparently are so embedded in human nature as to make such malpractice delusional, natural, universal, confident, wrong and incorrigible. Further still, several of the traditional lot turn on philosophical issues which one finds entrenched in even their most casual traditional descriptions, never mind that these descriptions are often wrong. The fallacy theorist is snared by these problems, for they stand menacingly in the way of his saying something sensible and deep about his target fallacies. He cannot make his way without presenting himself on in- 12See also [Cartwright, 1983, 87 ff] and [Hintikka, 1989, p. 4]. See [Gabbay and Woods, 2004] for further discussion.

Who Cares? 19 duction, causality, rational co-operation, charity, inference to the best explanation, Aristotelian substances and much more besides. Of course, this is not to say that theoretical headway with, say, the post hoc fallacy encumbers the researcher with the need to produce a complete and comprehensive theory of causality. But at a minimum, he must (1) attend to Hume's contention that there is no objective difference between and no fact of the matter about whether a correlation is also a causal correlation; and (2) integrate his remarks about the post hoc with his contribution to the prior question. So the requirement is that the fallacy theorist must make such contributions to central philosophical problems as make possible their integration with and reinforcement of his assertions about the fallacies in question. Another way of saying this (and only slightly overstated), is that each of the traditional fallacies is an entire research-programme in its own right. The further importance of the fallacies, then, is that they reflect in their caricatures performance-limitations which may be congenital in all reasoners, and/or they collect in their traditional casual descriptions philosophical problems of great difficulty, some success with which is an inescapable condition of advancing the theory of the fallacies in question. For fallacy theory is continuous with philosophy itself, and yet it branches into all the disciplines. 6. Critical Thinking One of the more interesting contentions among contemporary critical thinking theorists has to do with the question of field-specificity. Some hold with McPeck that there are few, if any, interesting canons of successful critical thought shorn of disciplinary attachment and, in any event, differential disciplinary attachment will always make for differential critical procedures.13 This sounds right, though it is an interesting question whether it forecloses upon all domain-neutral principles. (Cf. [Paul, 1985].) It is quite true that we shouldn't appraise paleontology in just the way that we do quantum field theory, but that may not entirely settle the hash of the generalization that denying the antecedent is always and everywhere a critical error. Some people (for example, [Weinstein, 1990]) have also come to the opinion that philosophy — "pure epistemology" — is conspiciously disabled for the analysis of critical thinking. It is thought, apparently, to 13The modern locus classicus of this view is [Toulmin, 1969]. McPeck may have taken Toulmin's insight somewhat further than originally intended [McPeck, 1981]. An excellent response to McPeck is [Ennis, 1996].

20 THE DEATH OF ARGUMENT be too general and too abstract, leaving unattended the peculiarities of domain-specificity. It also stints on the supply of application models. For all its allure, the domain-specificity thesis produces for its enthusiasts a huge encumbrance. If the critical thinking theorist is now expected to plow the fields of all of human knowledge, how is he to do this with low finite cognitive resources and within his biblical allotment of three score and ten years (or twenty, if he doesn't smoke and watches his diet)? Even our most robust and tenacious polymath will stagger under the cruel burdens of such learnedness. Of course, it is always open to the critical thinking theorist to make his task manageable by leaving most of it undone. Thus, "You do physics; Sarah law; Harry art history; and I'll take a crack at policy-formation theory". When you, Sarah, Harry and I reconvene in Amsterdam at the ISSA Conference in 2006, whether or not we will be able to exchange our results will be a direct function of how deeply true the domain-specificity thesis was in the first place. If the respective investigative canons of those disciplines do inhere in their dark centres, we will have nothing to say to one another. Even if subject depth doesn't completely foreclose upon chitchat across the lines of discipline, the proposed division of labour is not really what my critical thinker wants. He wants his theories to be, if not general, at least fairly comprehensive. He wants to outdo even Gerald Ford, who is said to be able to do two things at once. So he must be prepared to spread himself around. At this juncture the fallacy theorist has something useful to propose. He can invite the critical thinker to appropriate the gang of eighteen and let it take him where it will. If what I have been saying is so, the gang of eighteen will take him very many places and will make of the theorist a formidable cosmopolite. It is difficult to imagine that its tendrils would fail to reach any significant suburb of critical effort, as I hope the earlier discussion will have made plain. The particular virtue of the suggestion is that it gives to our polymath some useful procedural structure. He will have some settled idea of where to begin, and of which issues have been settled and of which are outstanding. The gang of eighteen furnishes the theorist with a rich and challenging research programme in medias res. It is a programme with an ancient philosophical pedigree, but it is also a programme that insists upon application models and tough conditions of empiricial adequacy. The truth in the contemporary complaint against the Standard Treatment is not that the research programme is no good, but rather that too many a textbook has not taken it seriously. So there is a rapprochement between the traditional fallacies and the critical thinking agenda begging to get underway. I hope that it will get underway, and soon, since such is the farther importance of the fallacies. The present

Who Cares? 21 proposal supplies us with an interesting, in fact a crucial, byproduct. It tests in an orderly way the suggestion that the differential critical canon lies deep in the disciplinary heart of each subject. If we did meet with this radical eccentricity, my sunny optimism of just lines above would have been called into question. But now the doubts would be principled ones. We would have demonstrated a Limitation Theorem, to speak tendentiously. And that too would be important. 7. So What? Part of what answers the boundary question is that fallacies are mis- performances of topic-neutral cognitive skills of a very basic kind. Another part of the answer is that fallacies are snares and delusions. They are snares because they embed theoretically elusive concepts. They are delusions in the sense that they are confusions that appear to have a genetic origin (we appear to be, in Fogelin's words "hard-wired" for these errors). They are attractive mistakes, they are made by everyone, and they are hard to correct. The two answers together also go some way towards answering the common factor problem, for not every type of mistake carries the appearance of a genetic etiology, not every type of mistake is attractive, universal and incorrigible; not every type of mistake embeds a research programme; and so on. This leaves the who-cares question. What is the importance of studying the fallacies? I have suggested that their importance is derivative upon the importance of the skills of which they are misperformances. I haven't yet discussed what may strike some people an the most obvious evidence of the importance of fallacies and fallacy theory: Fallacies damage cognitive performance, (both personally and interpersonally); and fallacy theory gives instruction in how to identify and avert fallacious reasoning and fallacious argument. The reason I have not yet cited this presumed fact is that it is not a fact. Books and articles on fallacy theory are not manuals of self-help for the cognitively (or argumentatively) insecure. I won't take the time to review the empirical and experimental record. Interested readers could consult, with profit, the combined e-mail lists, the OSSA list for the study of argumentation and ARGTHRY list at York University, from December 10, 1999 to February 10, 2000, for a fascinating discussion of the real-life efficacy of courses in critical thinking and informal logic. For something hard-printed see also [Manktelow, 1999]. On the face of it, this is a horrific admission. It is tantamount to conceding the uselessness of fallacy theory, and of converting our title into a merely rhetorical question.

22 THE DEATH OF ARGUMENT Upon reflection, we see that the expectation that a theory of fallacies would make us adept at avoiding their commission was both naive and careless. It was naive because, like most investigations of the human condition, there is little by way of a direct connection between disclosure and efficacy (in this respect, the traditional classification of logic as a humanities discipline is nothing but appropriate). It was careless in not taking seriously how fallacies have been characterized. They are snares and delusions. They are mistakes that are easy to make, that are widely made, and are difficult to correct, and difficult to detect. The value of fallacy theory is the value of any of the human sciences generally. A good account of the fallacies will specify the conditions under which certain inclinations to err cash out in human performance. Understanding that is understanding a part of the human condition and a nontrivial part, as has been the burden of this chapter to show. Bearing on the question of the utilitarian value of fallacy theory is a principle which I shall dub the Easy-Easy Principle. The principle does not strictly derive from the utilitarian expectation we have just been discussing, but it resembles it in spirit. To explain the principle, we need the concept of easiness as applied to human tasks. A task is easy when a human being can perform it competently without formal tutelage, and without noticeable effort. Having a conversation is easy for any pair of co-linguists, but producing a grammar of their language is not. Reading a timetable at the airport and inferring whether there is a flight from Vancouver to London on Saturday is easy; generating a theory of negation-as-failure (on which that inference in part rests) is not. Riding a bicycle is easy; plotting the underlying kinesiology is not. The Easy-Easy Principle asserts that easy behaviour has in principle an easy theory. An easy theory is similarly one that can be understood by an arbitrarily selected competent individual without tutelage and without noticeable effort. Easy theories include common sense theories, but are not restricted to them. What is surprising about the Easy-Easy Principle is the attraction it holds for people who should know better (briefly discussed in [Woods, 2000a]). It takes little reflection to see that it is a hopelessly mistaken maxim. Language is easy, but linguistics is hard. Inference is easy, but logic and cognitive science are hard. Arguing is easy, but there is no chance that even some of the simpler theories of argumentation are easy. A case in point is the pragma-dialectical approach to argumentation. For many people it is an attractive and accessible theory. But it is not easy. Fallacy theory fails the Easy-Easy Principle hands down. Getting a deep and principled description of a fallacy turns out to be a far more complex thing than is superficially evident in the ease with which it is

Who Cares? 23 committed in real life. Determining, for example, whether it is fallacious to judge that the probability of "Lisa is a socially moderate Republican" is greater than "Lisa is a Republican" is, as we saw, something that takes us into the thick of inductive logic and probability theory; and these are far from easy. The failure of the Easy-Easy Principle for the investigation of fallacious reasoning and argument serves as an indispensable guide to the fallacy theorist. It bids him not to seek for easy answers. This is hugely important guidance. It implies that a few confident intuitions on the part of the would-be theorist will not get the job done. It follows from this that if, for example, formal deductive logic is not the best theoretical instrument for modelling the fallacies, it is because it is the wrong model (concerning which see [Woods, 2000a]), not that it is difficult or technical or even that it is formal. It is here, I think, that we can best appreciate Hamblin's hostility to what he calls the Standard Treatment found in many of the logic textbooks of the late 1960s. Hamblin appears to be saying that the brief and largely unanalyzed descriptions of the fallacies that we find in such works cannot possibly be theoretically adequate accounts of them. This matters, according to Hamblin, since at the time he was writing, these textbook treatments were by and large all there was. It is as if the research community had given up on fallacies. It is also likely that Hamblin was responding to another concern. In most disciplines, textbooks have what I call the trickle-down property. They give accounts, albeit in simplified form perhaps, of the best theoretical work currently done in that discipline. This is true of textbooks on economics, biochemistry, physics, and banking and finance. Hamblin's point seems to be that there was in the late sixties nothing to trickle-down into textbooks or chapters on fallacies. Hamblin might agree with me that one of the attractions of the Standard Treatment is that it provides an idealized caricature of the fallacies, a kind of shortcut symptomatology. But if that is all there is to fallacy theory, there is no fallacy theory, and the caricatures are unavailing. If what I have been saying in the present chapter is so, we may hope that the would-be fallacy theorist has a reasonable conception of how he or she must proceed, if Hamblin's dream of a regenerated research programme is to be realized. Of course, much water has flowed under many bridges since 1970, the year in which Hamblin's book was published. It is gratifying to see how much of Hamblin's dream has indeed been realized; and, if I may make so bold to say so, realized in ways not obviously incompatible with the account of fallacies sketched in these pages.

Chapter 2 THE NECESSITY OF FORMALISM In this chapter and the next I want to take notice of the special relationship between fallacy theory and informal logic. Most of fallacy theory has nothing to do with the logical forms investigated by systems such as FDL (formal deductive logic). So it is reasonable to see most of fallacy theory as informal logic. At one time, I was inclined to the skeptical view that there was nothing to informal logic but fallacy theory. Now I am pretty sure that this is wrong. As will become clear, the fundamental question for informal logic is its connection or lack of it to the idea of logical form. 1. Methodological Pluralism This chapter is a sequel to an earlier attempt to find and justify uses of formal methods in informal logic and argumentation theory. Here I wish to give greater attention than I was able to previously to various motivations of anti-formalist skepticism in (much of) argumentation theory [Woods, 1980], reprinted in [Woods and Walton, 1989, ch. 17]. I shall also attempt to set these remarks in the context of the complex web of relations between and among theory, pedagogy and classroom practice. I acknowledge that my title is something of a provocation. So, too, was my conclusion in What is Informal Logic?: "What is Informal Logic? Nothing is? [Woods, 1980, p. 62]. But the present title's purpose is not merely rhetorical. For I hope that it can be said without offence that fueled by disapproval, a good deal of informal logic is crotchety, suspicious, anti-theoretical, doctrinaire and apocalyptic and thus in many ways wrongheaded and precipitate. There is, I think, a disturbing amount of methodological anxiety and self-scrutiny in argu-

26 THE DEATH OF ARGUMENT mentation theory, which suggests rather modest levels of conceptual and operational maturity in the subject. This being so, I rather think that what is required is a good deal of pragmatic toleration of methodological pluralism, and it is in this spirit that I offer my thoughts here. I propose to come at the question of the motivation of anti-formalist skepticism under five headings: formal logic; nomic systematicity; technical information; theory and praxis; and pedagogy. 2. Formal Logic Here I have in mind the modalized variant of first order logic known as the (generic) system of strict implication. It is this system against which Ralph Johnson blasphemed so mischieviously when he coined the acronym FDL, pronounced 'fuddle' [Johnson, 1987a, p. 47]. It is a wonderfully funny turn. But why should it be? The fact is that the disapproval of FDL among nonformalist argumentation theorists is immense; and it can be accounted for somewhat along the following lines. Among the leading complaints are these: (1) FDL is too narrow to model adequately anything as rich and complex as actual argumentation. FDL is still basically the theory of deduction restrictively conceived in simple sentential, first order and modal environments. (2) FDL is too abstract, as witness first order model theory in infinite domains. (3) FDL squanders too much of its energies on such merely technical questions as the representation of truth-functionality by the Sheffer stroke. (4) FDL is too metatheoretical, as witness its affection for such things as the Compactness Metatheorem. And (5) FDL is too counterintuitive, sacrificing credibility to algorithmicity, as witness its treatment of implication and equivalence. Each of these observations has a justification of sorts and each makes a point of greater or less importance. Each however is answerable, more or less convincingly (see Appendix A); and so I am minded to think that such criticisms pass their targets in the dark. (See also the Prologue.) What is truly surprising, however, is that the standard list of anti-formalist complaints does not contain two of the strongest attacks on the relevance of formal logic to the study and modelling of argument. It has been around awhile. I proposed it in 1974 at the St. Louis International Conference on Relevant Logic [Woods, 1989b], and the same point had been made even earlier and independently by Gilbert Harman; so it is Harman's point by precedence [Harman, 1970]. It is

The Necessity of Formalism 27 simple to state: It is said that the rules of inference of formal logic are more or less canonical for human reasoning; but this is provably untrue. The so-called "rules of inference" actually fail for inference. Harman, and I too, take inference to be a kind of reasoning which involves the adjustment of one's beliefs under the dynamic press of new stimuli. Inference can be likened to a function / from a set Gamma of belief-expressing statements to a not necessarily distinct set Delta of belief-expressing statements such that for any statement, 5, it may typically be the case that either Delta=GammaU{5} or Delta=Gamma-{5}, where B is some member of Gamma inconsistent with A. Where the possessor of Gamma is not sure about 5, then Delta might be GammaU{"I am not sure whether ^4"}; and so on. ('U here denotes the operation of set-theoretic sum, or union; '-' denotes set-theoretic subtraction.) The function / thus works typically either as an extension or a restriction of Gamma. A possessor of a stock of beliefs who is exposed to life's dynamic turnings and who has access to this function is able to modify his belief-stock by adding a new statement or by deleting an old statement, or by withholding either or both.1 It is a serious error to think (what virtually every elementary textbook in formal logic asserts with such lofty confidence) that the rules that govern /are (a subset of) the "inference rules" of classical logic. For, given that a belief-stock Gamma contains A and ^4—>5, Delta may be set in at least three ways: either Delta=Gamma U {B} or Delta=Gamma-{^4} or Delta=Gamma-{;4—>5}. But the detachment rule for formal logic allows for just one manoeuvre: 'to infer' B. Hence the detachment rule is not a valid rule of inference. Similar considerations disqualify the other rules of formal logic. They also, by the way, disqualify (modalized) ex falso quodlibet, which is the classical theorem that says that a contradiction entails any statement whatever. But it is important to notice two things about the unseating of this rule. First, there is nothing peculiarly or exotically wrong with ex falso quodlibet; it is just one of several rules which fail for inference. Secondly, the demonstrable failure of ex falso quodlibet is a failure for inference, not for entailment. C.I. Lewis was, I think, seriously wrong when he said the system of strict implication has a "primary advantage over any present system", namely, "... that its meaning of implication is precisely that of ordinary inference..." [Lewis, 1912, p. 531]. 1 Of course not every adjustment of one's belief-stock is accomplished inferentially. A change in retinal stimuli may well lead to a change in belief (of what you see), but this is not inference (pace Peirce). For more on the type of belief-adjustment that inference is, see [Gabbay and Woods, 2003].

28 THE DEATH OF ARGUMENT Well, then, the rules of formal logic misregulate inference, and this is a devastating body blow to strict implication as a theory of inference. It does not mean, however, that the system of strict implication is of no imaginable use to the argumentation theorist. For the system of strict implication still stands as a theory of entailment (Most of the more celebrated attacks on disjunctive syllogism and ex falso quodlibet, confuse inference with entailment in some or other lethal way!). And so, the answer to our question would seem to be: if a knowledge of entailment theory would be of use to argumentation-analysis, then formal logic, i.e., FDL, still has a contemporary raison d'etre within argumentation theory. And surely some knowledge of entailment is of imperative interest to the argumentation analyst. For one thing, he mu