Explanation In Action Theory And Historiography: Causal And Teleological Approaches 0429000642, 9780429000645, 0429000650, 9780429000652, 0429000669, 9780429000669, 042950604X, 9780429506048

Table of contents :
1. Introduction Gunnar SchumannPart I: Causal vs. Teleological Explanation of Action2. Causalism: On Action Explanation and Causal DevianceAlfred Mele3. Why and How? Teleological and Causal Concepts in Action ExplanationGeorge F. Schueler4. Rationalizing Principles and Causal ExplanationGuido Loehrer and Scott R. Sehon5. On an Imaginary Dialogue Between a Causalist and an Anti-causalist Giuseppina D'Oro6. Reasons, Causes, Desires, and DispositionsSeverin Schroeder7. Objectivism and Causalism about Reasons for ActionHans-Johann Glock and Eva Schmidt8. Are Reasons Like Shampoo?Constanine SandisPart II: Causal vs. Teleological Explanation in Historiography9. Counterfactual Causality and Historical ExplanationsDoris Gerber10. Beyond Causalism and AcausalismHarold Kincaid11. Two Methods in HistoryThomas Keutner12. An Anti-causal Theory of Action as Basis for Historical Explanations. A SketchGunnar Schumann13. Meanings and Mechanisms: An Actor-centered Approach to Historical ExplanationDaniel Little14. The Origins of Historiographic CausationAviezer Tucker

Citation preview

Explanation in Action Theory and Historiography

Is the appropriate form of human action explanation causal or rather teleological? While this is a central question in analytic philosophy of action, it also has implications for questions about the differences between methods of explanation in the sciences on the one hand and in the humanities and the social sciences on the other. This question bears on the problem of the appropriate form of explanations of past human actions, and therefore it is prominently discussed by analytic philosophers of historiography. This volume brings together causalists and anticausalists to address enduring philosophical questions at the heart of this debate, as well as their implications for the practice of historiography. Part I considers the quarrel between causalism and anti-causalism in recent developments in the philosophy of action. Part II presents chapters by causalists and anti-causalists that are more narrowly focused on the philosophy of historiography. Gunnar Schumann is Assistant Professor of Philosophy at FernUniversität in Hagen, Germany.

Routledge Studies in Contemporary Philosophy

Good Thinking A Knowledge First Virtue Epistemology Christoph Kelp Epiphenomenal Mind An Integrated Outlook on Sensations, Beliefs, and Pleasure William S. Robinson The Meanings of Violence From Critical Theory to Biopolitics Edited by Gavin Rae and Emma Ingala A Defense of Simulated Experience New Noble Lies Mark Silcox The Act and Object of Judgment Historical and Philosophical Perspectives Edited by Brian Ball and Christoph Schuringa Perception, Cognition and Aesthetics Edited by Dena Shottenkirk, Manuel Curado and Steven S. Gouveia Philosophical Perspectives on Moral and Civic Education Shaping Citizens and Their Schools Edited by Colin Macleod and Christine Tappolet Explanation in Action Theory and Historiography Causal and Teleological Approaches Edited by Gunnar Schumann For more information about this series, please visit: www.routledge.com/ Routledge-Studies-in-Contemporary-Philosophy/book-series/SE0720

Explanation in Action Theory and Historiography Causal and Teleological Approaches Edited by Gunnar Schumann

First published 2019 by Routledge 52 Vanderbilt Avenue, New York, NY 10017 and by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN Routledge is an imprint of the Taylor & Francis Group, an informa business © 2019 Taylor & Francis The right of the editor to be identified as the author of the editorial material, and of the authors for their individual chapters, has been asserted in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. Library of Congress Cataloging-in-Publication Data Names: Schumann, Gunnar, editor. Title: Explanation in action theory and historiography : causal and teleological approaches / edited by Gunnar Schumann. Description: 1 [edition]. | New York : Taylor & Francis, 2019. | Series: Routledge studies in contemporary philosophy ; 121 | Includes bibliographical references and index. Identifiers: LCCN 2019012775 | ISBN 9781138584402 (hardback) Subjects: LCSH: Explanation. | Causation. | Teleology. | Action theory. | Historiography. Classification: LCC BD237 .E864 2019 | DDC 128/.4—dc23 LC record available at https://lccn.loc.gov/2019012775 ISBN: 978-1-138-58440-2 (hbk) ISBN: 978-0-429-50604-8 (ebk) Typeset in Sabon by Apex CoVantage, LLC

Contents

Preface List of Contributors  1 Introduction

vii viii 1

GUNNAR SCHUMANN

PART I

Causal vs. Teleological Explanation of Action

43

  2 Causalism: On Action Explanation and Causal Deviance

45

ALFRED R. MELE

  3 Why and How? Teleological and Causal Concepts in Action Explanation

59

GEORGE F. SCHUELER

  4 Rationalizing Principles and Causal Explanation

78

GUIDO LÖHRER AND SCOTT R. SEHON

  5 On an Imaginary Dialogue Between a Causalist and an Anti-causalist

97

GIUSEPPINA D’ORO

  6 Reasons, Causes, Desires, and Dispositions

112

SEVERIN SCHROEDER

  7 Objectivism and Causalism About Reasons for Action HANS-JOHANN GLOCK AND EVA SCHMIDT

124

vi Contents   8 Are Reasons Like Shampoo?

146

CONSTANTINE SANDIS

PART II

Causal vs. Teleological Explanation in Historiography

165

  9 Counterfactual Causality and Historical Explanations

167

DORIS GERBER

10 Beyond Causalism and Acausalism

179

HAROLD KINCAID

11 Two Methods in History

194

THOMAS KEUTNER

12 An Anti-causal Theory of Action as Basis for Historical Explanations. A Sketch

215

GUNNAR SCHUMANN

13 Meanings and Mechanisms: An Actor-Centered Approach to Historical Explanation

234

DANIEL LITTLE

14 The Origins of Historiographic Causation

249

AVIEZER TUCKER

Author Index Subject Index

269 270

Preface

In March 2016, the editor organized the international philosophical conference “Causalism and Anti-Causalism in Historical Explanations” at the FernUniversität in Hagen, Germany, which provided the stimulus to this volume of papers. The conference discussed the issue of causalism and anti-causalism in philosophy of action and philosophy of historiography and brought together ideas and arguments from both areas of philosophical discourse. I am grateful to the FernUniversität in Hagen, Deutsche Forschungsgemeinschaft (DFG) and the Gesellschaft für Analytische Philosophie (GAP) for funding the conference. I would like to thank Giuseppina D’Oro and Thomas Keutner for proofreading the introduction and for helpful suggestions. I would also like to thank Andrew Weckenmann and Allie Simmons from Routledge for their organizational support in the editing process. Last, but not least, I am especially grateful to Theodor Berwe for his great assistance in the editorial work.

Contributors

Giuseppina D’Oro received her PhD from the University of Essex in 1995 and is Reader in Philosophy at Keele University, UK. She is author of Collingwood and the Metaphysics of Experience (2002) and CoEditor of Collingwood’s An Essay on Philosophical Method (2005) and Reasons and Causes. Causalism and Anti-Causalism in the Philosophy of Action (2013), among other publications. Doris Gerber received her PhD from the Universität Tübingen, Germany, in 2001 and habilitated there in 2010. She is Professor of Political Philosophy at Universität Bayreuth, Germany. Among her publications are: Analytische Metaphysik der Geschichte (2012) and “Causal Explanation and Historical Meaning: How to Solve the Problem of the Specific Historical Relation Between Events” in: Kaiser et  al. (eds.): Explanations in the Special Sciences. The Case of Biology and History. Hans-Johann Glock received his PhD from Oxford University in 1990 and is Professor of Theoretical Philosophy at the Universität Zürich, Switzerland, since 2006. Among his numerous publications are: Quine and Davidson on language, thought and reality (2003), What is Analytic Philosophy? (2008), “Reasons for action: Wittgensteinian and Davidsonian perspectives in historical and meta-philosophical context” (2014). He is also editor of A companion to Wittgenstein (2017). Thomas Keutner received his PhD from the Universität Bonn in 1979 and habilitated at the FernUniversität in Hagen, Germany in 2002. He is Professor Extraordinarius of Philosophy at FernUniversität in Hagen, Germany. He is author of Ignoranz, Täuschung, Selbsttäuschung. Kausalität in den Handlungswissenschaften (2004), co-author of “Ein Wissen das kein Licht ist—Absicht und die Autonomie des Praktischen” (1983). He is also co-translator and co-editor of G.E.M. Anscombe: Absicht. Harold Kincaid received his PhD in Philosophy with Economics minor from Indiana University in 1983. Since 2012, he is Professor at the

Contributors  ix School of Economics, University of Cape Town, South Africa. From 2014–16 he was Director of the Research Unit for Behavioral Economics and Neuroeconomics, 2014–16. Selected Books and papers: Philosophical Foundations of the Social Sciences, Cambridge University Press 1996. “Philosophies of Historiography and the social sciences”, 2006. “Causation in Social Sciences”, 2009. He is co-editor of The Oxford Handbook of the Philosophy of Economics, Oxford University Press 2009, and editor of Oxford Handbook of the Philosophy of Social Science, Oxford University Press 2013. Daniel Little received his PhD from Harvard in 1977 and is Professor of Philosophy and Chancellor at University of Michigan-Dearborn, USA. Selected Books and papers: New Contributions to the Philosophy of History, Springer. 2010. “Action in History and Social Science” 2012. “Philosophy of History”, The Stanford Encyclopedia of Philosophy. New Directions in the Philosophy of Social Science, 2016. Guido Löhrer received his PhD from the Universität Freiburg, Germany, in 1993 and habilitated at the Universität Bern, Switzerland, in 2001. Since 2008, he is Professor of Practical Philosophy at Universität Erfurt, Germany. Selected Books and papers: Praktisches Wissen. Grundlagen einer konstruktiven Theorie menschlichen Handelns, Mentis 2003. “Action, Reason Explanations, and Values”, 2016. “The Davidsonian Challenge to the Non-Causalist” (with Scott R. Sehon), American Philosophical Quarterly 53.1. He is also co-editor of Gründe und Zwecke. Texte zur Handlungstheorie, Suhrkamp 2010. Alfred R. Mele received his PhD from University of Michigan in 1979 and is William H. and Lucyle T. Werkmeister Professor at the Department of Philosophy of Florida State University, USA, since 2000. Among his numerous publications are: Motivation and Agency (2003), Effective Intentions: The Power of Conscious Will (2009), “Teleological Explanations of Actions: Anticausalism vs. Causalism” (2010), “Actions, Explanations, and Causes” (2013). Constantine Sandis received his PhD from University of Reading in 2005. In 2013 he was appointed to Full Professor at University of Oxford Brookes. Since 2015 he is Professor for Philosophy at University of Hertfordshire. Selected Books and papers: The things we do and why we do them. Palgrave Macmillan 2012. “One Fell Swoop: Small Red Book Historicism Before and After Davidson”. Journal of the Philosophy of History 2015, “Period and Place: Collingwood and Wittgenstein on Understanding Others”. Collingwood and British Idealism Studies 2016. He is editor of New Essays on the Explanation of Action, Palgrave Macmillan 2009 and co-editor of Reasons and Causes. Causalism and Anti-Causalism in the Philosophy of Action, 2013, as well as of Philosophy of Action: An Anthology. Wiley-Blackwell 2015.

x Contributors Eva Schmidt received her PhD from Universität des Saarlandes in 2011 and is Assistant Professor at University of Zürich, Switzerland. Selected Books and papers: Modest Nonconceptualism: Epistemology, Phenomenology, and Content. Studies in Brain and Mind, Bd. 8., Springer 2015. “Does Perceptual Content Have to Be Objective? A Defense of Nonconceptualism”, Journal for General Philosophy of Science 2015 and “New Trouble for Reasons as Evidence: Means that Don’t Justify the Ends”, Ethics 2017. Severin Schroeder is Associate Professor at University of Reading, UK. He received his PhD from Universität Hamburg, Germany, in 1998. He is author of “Are reasons Causes? A Wittgensteinian Response to Davidson” 2001; Wittgenstein: The Way Out of the Fly Bottle 2006. “Wittgenstein sur l’explication de l’action” 2014; “Intuition, decision, compulsion” 2016. He is also editor of Wittgenstein and contemporary philosophy of mind. Palgrave Macmillan 2001. George F. Schueler received his PhD from University of California, Berkeley in 1973. Since 2007 he is Professor of Philosophy at University of Delaware. Among his numerous publications are: The idea of a reason for acting 1989. Desire: Its Role in Practical Reason and the Explanation of Action, MIT Press/Bradford Books 1995. Reasons and Purposes: Human Rationality and The Teleological Explanation of Action, Clarendon Press/Oxford University Press 2005. Articles include: “Interpretative Explanations” 2009; “The Humean Theory of Motivation Rejected” 2009; “Motivational Internalism and Externalism” 2010. Gunnar Schumann is Assistant Professor of Philosophy at FernUniversität in Hagen, Germany. He received his PhD from Universität Halle-Wittenberg, Germany, in 2011. He is author of Epistemische Rechtfertigung und Wahrheit als Empfehlung, mentis 2013; ­“Practical reasoning as normative reasoning” 2014; “Von Wright’s theory of action and the ‘Logical Connection Argument’ ”, 2017. Scott R. Sehon is Professor of Philosophy at Bowdoin College. He received his PhD in Philosophy from Princeton University in 1994. Selected Books and papers: “Deviant Causal Chains and the Irreducibility of Teleological Explanation”, Pacific Philosophical Quarterly 1997; “An Argument Against the Causal Theory of Action Explanation”, Philosophy and Phenomenological Research 2000; “Goal-Directed Action and Teleological Explanation” 2007; and “The Causal Theory of Action and the Commitments of Common Sense Psychology” 2013. His books include: Teleological Realism: Mind, Agency, and Explanation, MIT Press 2005 and Free Will and Action Explanation: A NonCausal, Compatibilist Account, Oxford University Press 2016.

Contributors  xi Aviezer Tucker received his PhD from University of Maryland. He is Associate Researcher at the Davis Center for Russian & Eurasian Studies, Harvard University, USA. Selected Books and papers: Our Knowledge of the Past: A Philosophy of Historiography, Cambridge Univ Press 2004. “Causation in Historiography” 2008; “Sciences of Tokens and Types: The Difference between History and the Social Sciences” 2012; “Biology and History. What makes the difference?” 2014; “Historiographic Counterfactuals and the Philosophy of Historiography: An Introduction to a special issue of the Journal of Philosophy of History” 2016. He is editor of A Companion to the Philosophy of History and Historiography, Boston: Wiley-Blackwell 2008.

1 Introduction Gunnar Schumann

This volume deals with the philosophical quarrel about the nature of action explanation; in particular, whether explanations of human actions, be they past or present, are a form of causal explanation or teleological explanations. This introduction consists of three parts: The first generally introduces the central issue that the volume attempts to tackle and provides a brief historical overview of the debate to help the reader understand the relevance of this philosophical quarrel as well as how both philosophical fields, action theory and philosophy of historiography, are interrelated. The second part describes the aims of this volume, and the third part provides an overview of the contributions to this volume and draws some connections among the topics they discuss.

I.  What Is the Debate About? Actions belong to the life of humans. We encounter actions in countless forms in everyday life, either actively as agents or as patients or as observers. Actions come in a variety of forms and have several aspects. There are voluntary bodily motions (raising an arm, nodding one’s head), omissions, inner activities (calculating in one’s head, reciting a poem inwardly), emotional actions (shouting because one is angry, crying because one is in great fear) and rational actions (going to the store in order to buy food), spontaneous and planned actions, habitual actions, verbal and nonverbal actions, activities (walking, shooting, playing) and success actions (arriving, hitting, scoring), short-term actions and those the performance of which may take years, present and past actions, individual and collective actions, cooperative (conversing, trading), and competitive actions (fighting). Sometimes the question arises as to why a particular action was performed and the answer is not always self-evident. Then we ask for the explanation of an action: “Why is Rebecca going to the mall?”; “What purpose was behind Peter’s purchase?”; “Why did the company end the collaboration?”. And sometimes action explanations even require a thorough and systematic investigation. Human actions and their results are

2  Gunnar Schumann a subject of human sciences in the widest sense, such as ethics, law, politics, sociology, economics, psychology, anthropology, and ethnology, or all forms of human historiography. “Why did the Mayans build their temples this way?”, “Why did Europe’s great powers go to war with each other in 1914?”, “Why did so many Germans vote for Hitler in the early 1930s?” Usually, when we ask for an explanation of an action, we do not want to know how the action came to be. We do not want an account of e.g. all the physiological factors involved in the physical aspect of the action. What we usually ask for when we ask “Why was this action done?” is an explanation of the rationale of the action, of what it was that spoke in favor of the action, what may justify the action (maybe only from the standpoint of the agent). In short, we ask for reasons the agent had or presumably had, i.e. the agent’s purposes, goals, motives, intentions, and their means-ends beliefs. This is called a teleological explanation (from Ancient Greek: τέλος—“goal, purpose, end”). Teleological action explanations are often just given in an abbreviated form, i.e. only either the agent’s goal (their volitional premise) or her means-end-belief (their doxastic premise) are explicitly stated. That is because the other premise is supposed to be so obvious or common knowledge that it need not to be stated explicitly when an action is explained. The concept of teleological action explanation is closely linked to that of intentional action because in explaining the action teleologically we present the action as intentional. When we can state what the purpose, goal, or intention of an agent was, we can be sure that her action was intentional. Philosophers are interested in the concept of intentionality and try to give an account of its exact meaning, because it is the criterion by which they distinguish mere behavior (which is a mere bodily movement, as e.g. the movement of one’s leg when the knee-jerk is tested) from proper action (the raising of one’s arm; voting). Intentional actions (or actions in the narrower sense—in opposition to mere behavior) are those happenings for which it is appropriate to ask for what reason they were done (although there need not always be a definite answer). “Action”, respectively “agency”, is a concept that is interwoven with other concepts of great philosophical importance like “mind”, “person”, “freedom”, “responsibility”, “good” and “bad”, “culture” (as opposed to “nature”), and its connections to “cause”, “event”, and “regularity” are matter of philosophical debates. There is a philosophical quarrel about the appropriateness and sufficiency of teleological explanations of actions. Causalists claim that the purpose-stating explanations of human actions are, taken by themselves, scientifically suspicious and at best superficial explanations, for proper explanations of human actions ought to state their causes—just as physical events are explained in the sciences. According to causalists, the true form of an explanation of a human action is that of a causal explanation.

Introduction  3 In contemporary analytical action theory, causalism is the position that, as Mele puts it, “our actions are, essentially, events [. . .] that are suitably caused by mental items, or neural realizations of those items. Causalists traditionally appeal, in part, to such goal-representing states as desires and intentions (or their neural realizers) in their explanations of human actions [. . .]” (Mele 2000: 279). The causalist’s claim is that intentional action is just a form of behavior caused (in the right way) by a desire or intention or other appropriate mental state. There is a sense in which causalism is uncontroversial: When one takes “cause of the action” in its minimal sense as anything which is to explain the action. However, the causalist explicitly or implicitly assumes that the only proper form of explaining an action is to state its causes. According to the anti-causalists, this is not self-evident. The causalist might be under the false impression that “to explain why” would be tantamount with “to state the causes of an event” and that “Why?”-questions would always have the role of asking for the causes of something. However, anti-causalists stress that there are (at least) two different senses of “Why?” and “because”—the causal one and the teleological one.1 And they insist that intentional human actions are to be explained by their reasons and that reasons are not reducible to causes. In everyday language and in humanities we do not differentiate strictly between “cause” and “reason” and sometimes use both expressions interchangeably. (We sometimes speak of the reasons of physical events such as e.g. an explosion or of causes of collective human actions like the French Revolution. But we also use both expressions very differently: We speak of a spark as the cause of the exploding barrel, but we would not speak of a spark as the reason for the barrel to explode.) What causalists and anti-causalists disagree on is whether teleological explanations of actions are only causal explanations in disguise, in the sense that reasons and purposes do not in fact explain the action but underlying causes do. According to the causalist, the elements of an action explanation in terms of reasons, as purposes, goals, intentions, and beliefs (that allegedly might be identified with neural states or events in the agent’s brain in the future) are taken to be causes—the anti-causalist denies this. For him, purposes, goals, intentions, and beliefs are the determinants of action—but this relation is not one of causation but of another kind, mainly of justification or of normativity or of commitment, for reasons render actions intelligible. The connection of this quarrel between action theoretical causalists and anti-causalists to the problem of explanations in historiography lies in the question: What is the appropriate form of explanation of past human actions? Explanations of past human actions are at least an important sub-set of explanations in historiography.2 The question is especially virulent and far-reaching because it concerns the further and prominent question whether the method of historiographic explanation is the same as in the natural sciences and whether this marks a decisive difference

4  Gunnar Schumann between the sciences and the humanities, respectively the sciences and the social sciences. Hardly anyone would deny that (relevant) past intentional actions and their consequences belong to the subject matter of historiography in principle. Philosophers of historiography have thus also contributed to the question of how to explain (past) human actions and what role causes and reasons play in this.3 In fact, one of the roots of the reason vs. causes debate as it is discussed in contemporary analytic theory of action is philosophy of historiography resp. philosophy of social science (represented by Collingwood, Winch, Dray, the other root is represented by late Wittgenstein, Ryle, Anscombe (cf. Hacker 1996: 619)). Still, sometimes it is argued that intentional action and consequently their explanations do not play a decisive role in historiography at all. For one may think of two objections: 1.  The idea that the explanation of past actions is relevant to historiography may stem from the old-fashioned picture of historiography as the historiography of “great deeds of great men”. But, there is also a historiography of social or economic conditions, of institutions, of law systems, of cultural factors, a.s.o. such that the subject of history in fact are e.g. economical classes, social environments, cliques, castes, political camps and parties, organizations, ethnic groups, nations, cultures, religious sects, artistic and intellectual movements, lobby groups a.s.o. Accordingly, historical explanations are not explanations of great individual’s actions. This objection can be met, though: The problem whether historical explanations are causal or teleological is not only relevant for past actions of individual agents, but also for collective actions of groups of agents. The explanation of collective action is dependent on the more general issue of the appropriate form of action explanation. It seems safe to say that collective actions can be analyzed in terms of individual actions. Hardly anyone would be willing to accept that collective subjects exist— in the sense that they are irreducible, non-analyzable agents sui generis. It seems that only an unnecessary literal understanding of expressions referring to collective agents (“The protestants declined the offer . . .”) necessitates the assumption of group subjects as acting agents in their own right that could not be taken to be in some sense the sum of single agents. In what sense collective agency emerges from individual agency is sure enough the topic of another debate. 2.  The other objection goes back to the fact that historiography is not, or need not to be, the historiography of “great men and their deeds”. Sometimes it is held that human history does not even consist of actions and what men mutually intend but that it is determined by social “structures”, institutions, and frameworks. This objection can be met as well: Admittedly, historical developments are not the direct result of intended actions in the sense that not everything that happened in human history was intended by agents. But

Introduction  5 that is trivial and not presupposed by the claim that human history is made up from intentional actions or even only of the claim that historical explanations rest on action explanations (Gerber 2012: 28f.). (E.g. the [re-]discovery of the Americas was not intended by Columbus, nonetheless his [re-]discovery was an intentional action.) The claim that history does not consist of actions (or of results of intentional actions) is wrong, when taken in the sense that there is no connection between history and human actions. The claim might originate from the impression that sometimes the realm of voluntary influence of man in history is very limited, for there are processes, structures, institutions, and frameworks that drive human history—more or less detached from the intentional actions of humans: constitutions and political institutions, forms of government, the political culture of a country, customs and habits, unconscious forms of behavior, collective mentalities, systems of beliefs and values, sequences of generations, entrenched friend-foe constellations, linguistic differentiations of class- or region-specific nature; but also geographic or climate conditions, productive forces and relations of production, business organizations, school systems, stable international relations and international organizations; etc. (Kocka 1986: 77). Sometimes historians speak of things like a “logic of imperialism” or of an autonomous “logic of escalation” which led more or less inevitably to World War I (a process that might even have run against the declared intentions of prominent political agents). But there are two problems with this anti-intentional view of history: i. It is undeniable that such structures as e.g. institutions or processes may develop dynamics on their own. Intentional actions may have unintended side effects or consequences which again may provide or determine the framework of further human actions. But the thesis that such cybernetic processes govern history presupposes the validity of the distinction between intentional actions and non-intentional effects of actions—and intentional actions are those for which reasons can be given in principle. Thus, on the one hand, even an antiintentionalist view of human history rests on the intelligibility of the distinction between intentional and non-intentional actions and thus would also require a conceptual clarification of the concept of intentional action. ii. All these structures can neither be explained nor described without referring to the intentional actions of individuals or social groups in one or the other way. “Social structures” do not exist independently from human making but are the intended or unintended consequences of them. What are so-called “structural explanations” if not extremely abbreviated, implicit intentional explanations which have intentional collective actions as their basis (Gerber 2012: 30f.)? So-called “structural explanations” rest on intentional explanations.

6  Gunnar Schumann 3.  There is another sense in which not everything that happened in human history was the result of human planning and intending. Sometimes, human actions are done for no particular reason, but they result from affects, emotions, urges; in one word they are non-rational. Actions may be the expression of strong emotions like joy, hate, pain, desperation, bloodlust, or fear. There might be no particular reasons involved there. Explanations with the help of these factors may also be appropriate of some collective actions of historical significance. This is certainly correct. But, insofar as we may categorize these actions as voluntary and intentional, it still can be said that there might be given reasons for them. Even emotional actions are intentional in the sense that they are not something passively experienced and for which we might not hold the agents responsible. It is not always the case that an intentional action is preceded by deliberation or calculation. But this is not necessary in order for it to be called an intentional action. All that is required for a behavior to count as intentional (i.e. as an action) is that there could be given reasons for it by the agent in principle. Furthermore, the amount of unconscious behavior is typically of no importance in historiography; for political, military, economical, moral, scientific, artistic, religious actions which make up a great deal of history cannot be done unintentionally. It belongs to the concept of these actions that they can only be done on purpose, because they require deliberation and calculation (Collingwood 1946: 309f.).

II.  Historical Overview II.1  Early Modern Age to First Half of 20th Century To illustrate the controversy between teleologists4 and causalists in action theory, a short historical survey of it will be instructive. The debate about the appropriate form of human action explanation reaches far back in the history of philosophy (Milesian Philosophy of Nature/Atomists vs. Plato/Aristotle). Aristotle was the first who drew an explicit distinction between “efficient cause” (= “cause” in our modern sense) and “final cause” (purposes, aims, and goals). (This does not mean, as it is sometimes taken, that he simply regarded explanations in terms of purposes and ends as causal explanations, for Aristotle’s use of αἰτία is not the same as that of our modern word “cause”, rather it can be translated with “explanation”).5 He also applied “final causation” to the realm of nature and nonsentient beings, and regarded it as the appropriate form of explanation of the behavior of natural things like falling bodies or seeds. The issue about the appropriate form of human action explanation in the narrower sense started in early modern philosophy. With the rise and success of the sciences from the 17th century on, the new mechanistic worldview replaced old teleological explanations of physical phenomena

Introduction  7 and processes. Teleological explanations fell into disrespect.6 The temptation was strong to expand the methods of explanation of physics to all other fields, including the realm of men. Like in physics, all explanations in terms of goals and purposes were thought to be substitutable by explanations in terms of efficient causes, such that human actions become scientifically explainable. René Descartes, Thomas Hobbes, John Locke, George Berkeley, David Hume, Jeremy Bentham, John Austin (the 19th-century legal philosopher), and John Stuart Mill all agreed that mental states (be it a metaphysical state sui generis or a physiological state) are the causes of voluntary actions (cf. Hacker 1996: 543–547). Goals, aims, and intentions were construed as “conatus” or “impulse” in order to be taken as causes (this idea is illustrated by the word “motive” which comes from Latin motivum [“moving cause”]). Descartes and his dualist followers (Rationalists and Empiricists alike) believed that these mental concepts are abstract, invisible, ethereal entities belonging to a realm in its own right (the so-called res cogitans) and have causal interactions with the physical realm (the res extensa). This Cartesian picture envisaged the world as bifurcated into mind(s) and bodies. According to this prevalent tradition of early modern thinkers, we know perfectly what is going on in our own soul, but have only very limited access to the mental states of others. Every subject would have optimal epistemic access to these elements of her res cogitans qua introspection whereas it would be always doubtful whether and what mental states others have or the other’s mental states would only be derivable by analogy to one’s own mind. Nonetheless, it would not have been doubted that volitional mental states are the causes of the actions of men (except for e.g. Leibniz with his doctrine of pre-established harmony). By the same token, could not animals and ultimately even human beings be grasped as machines, only of a very complicated sort? If so, the workings of these machines, human actions (taken as bodily movements), could and should be explained causally as well. Indeed nobody was able to describe in detail the causal chains that led from volitional mental states to human actions—but that just seemed because human beings are such a complex system with many interacting factors such that concrete and complete causal explanations were hard to state. But it would not be doubted that causal explanations of human conduct were possible, at least “in principle”, such that their precise formulation could be given in the course of future scientific development, especially of neuroscience. The great success of causal explanations in the early modern sciences made it tempting to regard the teleological explanation of human actions as scientifically deficient. Teleological explanations did not admit of analysis into components in the same way that causal explanations admit of the distinction between different causes and thus seemed to be less clear and scientific. Furthermore, teleological explanations did not allow of measurability, quantification, and calculation as causal explanations in

8  Gunnar Schumann the natural sciences, foremost physics, did. Teleological explanations also did not seem to allow for the same degree of application to the future like causal explanations in applied science and technology. Prediction based on teleological explanations is often far less far-reaching, calculable, precise, and certain than prediction based on causal explanations in the sciences (esp. in Astrophysics). No wonder that teleological explanations fell more and more into disrespect. What seemed legitimate, even called for, for explanations of individual actions, also seemed so for explanations of social and collective aspects of human affairs. The idea of a causal determination of collective human action is fueled from the apparent analogy between the explainability of complex physical phenomena on the basis of the properties and interactions of the elements in which they can be analyzed on the one hand, and the explainability of complex social and/or historical phenomena on the basis of the properties and interactions of the elements in which they can be analyzed on the other. The idea was formed that just as the behavior of a complex physical system, e.g. an ideal gas or a liquid, can be explained by the kinetic properties and forces of elementary particles, the behavior of a social complex, e.g. the state or the market, can be explained in principle by the volitional (and doxastic) properties of individuals. This seems to be the ultimate reasoning e.g. behind Thomas Hobbes’ analysis of the state. Usually, the analogue to physical force behind human behavior was taken to be man’s self-interest. In the age of enlightenment, the idea of a necessitating mechanism driving the course of events was transferred to the study of the historical past. In the material (speculative) philosophies of history by Giambattista Vico, Anne Robert Jacques Turgot, Nicolas de Condorcet, Jean-Jacques Rousseau, Immanuel Kant, and others, the idea of a (more or less) necessitation of the course of history by the sum of unintended consequences of intentional actions played a key role.7 In economics, Bernard Mandeville, Adam Smith, and Karl Marx were prominent proponents of this idea. The idea of a course of events driven by man’s self-interest was spelled out in many different ways, according to the different assumed initial situations and factors that humans would face. The systematic debate about the nature and methods of historiography, the humanities and social sciences reaches back to the rise and formation of the humanities as academic disciplines in the 19th century. Empirical investigations seem to suggest that even laws or at least more or less loose regularities could be found in human behavior, which might serve as a basis for explanation for large-scale social phenomena (rate of birth, marriage, suicide). The idea of regularity and law-likeness of human behavior came to a first peak in the 19th-century positivist theory of science of Auguste Comte, John Stuart Mill (2002: book VI, ch. 1), and Henry Thomas Buckle. Nineteenth-century Positivism took sociology and economics as social physics respectively physics of man. The fact

Introduction  9 that there are hardly any experiments in historiography and only few in social sciences marks no decisive difference between them and the sciences and does not render a scientification of human sciences impossible (cf. Friedman 1958: 10): (1) The historiographer or social scientist would just have to wait for real (uncontrolled) “experiments”, i.e. investigate social events that have factually occurred (examples of how an increase in the amount of money in a country affects the inflation rate are abundant). (2) Also some sciences cannot conduct experiments (astronomy, geology), whereas on the side of social sciences the idea of social experiments in sociology and economics get more and more popular in recent years. (3) The distinction between controlled and uncontrolled experiments is only a gradual one anyway. Even in scientific experiments never all parameters of the observed situation are controlled by the scientist. Another idea connected to the idea of human science as a social physics was the idea of social engineering (cf. Little 2010: 407f.). According to Karl Popper, Marx intended his Capital as a general theory of capitalist society, as a basis for predicting its future and its specific internal transformations. Marx’s theory of historical materialism (“History is the history of class struggles”, “History is the unfolding of the contradictions between the forces and conditions of production”) is Popper’s central example of a holistic theory of history and Marx’s theory of revolution is Popper’s example of “social engineering” (cf. Little 2010: 407f.). Neoclassical economics further pushed the idea of a world in which atomistic, individual persons act just analogously to particles according to certain laws in a mechanical system. The logical consequence was the introduction of mathematics to the realm of men, making economic human affairs calculable. The introduction of mathematical methods was not undertaken in historiography but the same form of scientific grounding was attempted with the introduction of the idea of historical laws. The goal of this transfer of the methods of physics to the human realm was to elevate the sciences of man (historiography, sociology, economics) to the rank of sciences in the strict sense, or even to secure their scientific character in the first place. This idea has repeatedly cropped up in intellectual history since the modern scientific age until today. This idea of “scientification” of the humanities has three features (von Wright 1971: ch. I.2): (1) methodological monism, i.e. the idea that the methods of explanation in the sciences is the same as in the humanities (the contrast position is called “methodological pluralism”); (2) the exact sciences, above all, mathematicised physics set the ideal for all other sciences; (3) scientific explanation is causal, i.e. explanations consist in the subsumption of individual cases under hypothetical general laws (including those of “human nature”). In the second half of the 19th and in the early 20th century, some theorists opposed the idea of a scientification of the humanities. The debate emerged mainly on the field of historiography, which is often taken as

10  Gunnar Schumann the paradigm case of the humanities. Some historians and philosophers believed that historiography was not apt to adopt the methods of physics. Johann Gustav Droysen, Wilhelm Dilthey, Wilhelm Windelband, Heinrich Rickert, Georg Simmel, Max Weber,8 Benedetto Croce, Robin G. Collingwood—all these thinkers reject the appropriateness of scientific methods for the humanities. However, this anti-positivist movement was very diverse: Droysen (1977: 425) and Dilthey (1973: 85ff, 131, 146f.) marked the dichotomy between the methods of the sciences and the humanities as the one between “Erklären und Verstehen” (“Explanation and Understanding”). This title has become the prominent acronym of the debate between causalists and anti-causalists to the present day. Dilthey invoked the notion of empathetic understanding: The genuine method of the historian would be to put himself mentally into the position of historical agents in order to explain their actions. This drew some justified criticism (e.g. by Hempel 1942: 50): It is hard enough to put ourselves in the position of our contemporaries—it seems next to impossible to put oneself in the position of e.g. a medieval peasant or monk or paranoid historical rulers. Furthermore, it does not seem required for the reliability of a historical explanation that the historian puts himself imaginarily in the position of historical agents. Dilthey also took the so-called hermeneutic circle to be the characteristic feature of explanations in the humanities, a thought that became influential in 20th-century phenomenology. The suitability of this criterion has also been disputed (Stegmüller 1979). In any case, Dilthey’s distinction between “explaining” and “understanding” is not in accordance with our ordinary use of these words, for it is perfectly legitimate to speak of the understanding of natural events (e.g. processes in the center of the sun) as well as of the explanation of human actions (e.g. the French Revolution). Windelband marked the difference as one between law-giving disciplines and those that describe individual events or objects for which he coined the terms “nomothetic” and “idiographic” (Windelband 1894). Rickert developed this idea further and drew a distinction between two types of concept formation in the natural and the “cultural sciences” (Rickert 1986). For Collingwood, the task of the historian was the “re-enactment of past experience”, i.e. to discover the thought “behind” past actions, i.e. the agent’s intentions and beliefs (Collingwood 1946: ch. V, § 1; Collingwood 1998: Pt. IIIc, ch. XXX; see Thomas Keutner’s paper in this volume). II.2  Logical Positivism of the “Vienna Circle” The wave of positivism in the first half of the 19th century was followed by an anti-positivist wave in the second half of the 19th century and around 1900, which in turn was countered by a new strong positivist movement in 20th-century empirically oriented analytic philosophy after

Introduction  11 WWI: The Logical Empiricism of the Vienna Circle. Today’s analytical philosophy and contemporary scientific theory feeds to a large extent on this tradition of Logical Empiricism with its ultimate goal of a scientification of philosophy itself. That the heritage of Logical Empiricism is so dominant today is also related to the negligence of insights of Ordinary Language Philosophy in contemporary analytical philosophy, as will be discussed shortly (von Wright 1971: ch. I.5). Teleological explanations of individual human action involve mental concepts such as “purpose”, “aim”, “goal”, “intention”, “desire”, “want”, and “belief” or “knowledge” which seem not to be objectively observable and thus became suspicious to modern empiricists and empirically minded scientists. Furthermore, the idea of mental causation as a cross-substantial interaction between entities of the res cogitans and the res extensa seemed to violate the causal closure of the physical world. Each voluntary action would have to be seen as a miracle. Modern Empiricism tried to get rid of the idea of the metaphysical res cogitans that did not fit into the scientific worldview and took mental concepts to be mere verbal placeholders for physically describable behavior (“Logical Behaviorism”; c.f. Carnap 1932/33) or physical entities (mostly of a neurological sort, as in the physicalist variant of the “Identity Theory of Mind”; e.g. Smart 1959)9 to account for their alleged causal efficacy. The dualist picture of the mind as an independent res cogitans was abandoned in favor of a materialist conception of it, according to which the role of the Cartesian soul was taken over by the brain and the nervous system. The empirical sciences were entrusted with the task of investigating and determining the true efficient causes of human behavior, especially physiology as a branch of physics in the wide sense. The general thought that the reasons which agents give of their actions are unreliable or at least irrelevant for a proper explanation of them was prepared by causalist interpretations of Marxism and Psychoanalysis and shared by economists like Vilfredo Pareto. It seemed that in both the reasons with which the agents explained their actions played no major role, instead were judged to be unreliable and marked as mere “ideology” (Marxism) or “rationalization” (Psychoanalysis) that would have to be replaced by the true reason for which the agents acted, i.e. the unconscious causes underlying their actions (Neurath 1931: 419; cf. Winch 1958: 97f). In Logical Empiricism, monism reached its peak in the aspiration of the “Einheitswissenschaft” (unity-of-science). The Logical Positivists pursued the program of cleansing modern humanities and social sciences from metaphysical residues. This would be achieved by showing how all statements of the humanities and social sciences, including those of psychology, could in principle be reduced to statements of physics, namely physical descriptions of agents’ behavior (Neurath 1931; Carnap 1932/33). Consequently, part of their agenda was to show that also the methods of explanation of the historical and the social sciences did not differ substantially.

12  Gunnar Schumann Karl Popper and Carl Gustav Hempel famously claimed that science and historiography employ explanations which have the same logical form: that of the deduction of the explanandum from laws (Popper 1935; Hempel 1942). The full explanation of a historical event consists of a set of true general laws of the form “Whenever Y happens, X happens” on the one hand, and a set of true statements that state the occurrence of particular events (antecedent conditions) on the other, as premises of a deductively valid syllogism from which the explanandum follows. To explain an action (taken as an event) is to regard it as an instance of a strict law. This “deductive-nomological” (DN) or “covering law” model was taken to be the standard form of explanation in general. Historical explanations should be causal explanations of the DN-form. Likewise, predictions are possible in historiography—they have the same logical form as explanations, except that the laws and antecedent conditions are known in advance and the predicted event is derived from them. That the practice of explanation in historiography is deviating from this model was clear to Hempel from the start, for historiography and sociology do not employ strict or even ceteris paribus laws like the sciences. Hempel offered two explanations for this: (1) Many laws are those of individual or social psychology, which are familiar from everyday life and self-evident, such that they would not be stated explicitly. (2) It is often very difficult to formulate the underlying assumptions with sufficient precision and in accordance with all empirical evidence. An example of such a historical explanation is: “The Dust Bowl Farmers moved to California because continued drought made their existence precarious and California promised better prospects”. The explanation is based on general laws, such as that populations have a tendency to migrate to regions that offer better living conditions. However, it would be difficult to cast this assumption in the form of a general law that is well supported by all relevant empirical evidence. Many historical explanations would have to be taken as conclusions derived from probabilistic laws, the probability values of which will only be roughly known. Nonetheless, Hempel insisted that unless a historical explanation would state a general law as its major premise, the explanation would be a mere explanation sketch that rests on a more or less vague provisional regularity statement which will have to be replaced by a fully explicit strict law as historiography progresses (Hempel 1942: 48). II.3  Critical Reaction to Logical Positivism Subsequently, Popper’s and Hempel’s approach has been widely criticized as insufficient, in part even for explanations in the natural sciences.10 Hempel’s DN-Model was taken to be especially inapt for historiography: William Dray objected against Hempel that the historian would have no use for historical (sociological, economical, psychological) laws: Either

Introduction  13 the laws would be too general and counter-instances to them could easily be found or they would become so specific that they only match one particular case and lose their generality, i.e. law-character. When the task is to explain e.g. why Louis XIV of France died so unpopular and the explanation would rest on the law-like premise “Rulers who wage many wars and put a heavy burden on their people become unpopular” then this law could easily be falsified by counter instances from history. In order to make the law fit the historical facts, the covering law theorist would have to qualify the law with further amendments, as e.g. “European rulers around 1700 who waged the Franco-Dutch War, the Nine Years’ War, and the War of the Spanish Succession and put a heavy burden on their people become unpopular”. But then the “law” would cover fewer and fewer cases, ultimately only one case, such that it would lose its law character and could no longer explain the unpopularity of Louis XIV of France (Dray 1957: ch. II.3). Concerning an important subdomain of explanations in historiography, namely the explanation of past actions, Dray (being inspired by Collingwood) proposed his model of rational explanations. Their task is to show that an intentional action of an (historical) agent was rational or appropriate from the agent’s perspective. Explanation thus would consist in the reconstruction of a calculation of a historical agent and the action would be explained by presenting it as the agent’s means to achieve his purposes. Dray stressed that there is an element of appraisal in action explanations (which is not, one could add, found in causal explanations). If the reasons for a given action are to be rationally explanatory, then they must be “good reasons” in the sense that the action done is the one the agent would have considered to be one that is to be done (in the situation as perceived by the agent, i.e. given his means-ends-beliefs) (Dray 1957: ch. V.2). Hempel and others replied that the task of explanations is to account for why the action in question occurred in fact and not only why it was appropriate to do (from whatever perspective) (Hempel 1965: ch. 10.3.1f.). To explain a human action by reasons and rationality is to show that the action was a manifestation of the agent’s general disposition to act. And, dispositional explanations of actions still are causal explanations, for they involve a general law and an occasion and the occasion may then be called the “cause” of the action. At the same time as Dray, Elizabeth Anscombe and Abraham Melden also argued that causal explanations are not the appropriate form of actions explanation (Anscombe 1957; Melden 1961). Both authors elaborated ideas about actions and reasons of late Ludwig Wittgenstein. Already in the 1930s and 1940s, Wittgenstein developed a position that fundamentally opposed reductionism both in philosophy of mind and action theoretical causalism of the Vienna Circle and their followers. In the Blue Book, Wittgenstein argued that reasons are not causes—mainly

14  Gunnar Schumann because causal explanations are hypotheses, but stating one’s reason is not. (Wittgenstein 1958: 15). In the Philosophical Investigations Wittgenstein stressed the point (among others) that we can say what we do— without having to observe our own behavior and that this differs from explanations that state the causes of an event (Wittgenstein 1984: §§ 631f.). In her book Intention, Anscombe argued, on the one hand, that actions explanations that make use of backward-looking motives like revenge or gratefulness involve an element of evaluation which is alien to causal explanations. There is something considered to be good or bad by the agent when an action is explained by such motives (Anscombe 1957: § 14). On the other hand, she argued that from the first-person perspective, intentional actions are those the reasons for which we know without using observation—neither of our environment nor of ourselves. This distinguishes intentional action like waving one’s arm from e.g. the wincing of one’s leg when we fall asleep. The latter are explained by causes whereas the former are not explained by reasons which we know independently of observation (“in order to greet the people over there”). She called this form of knowledge of what our action aims at and what the action therefore is “practical knowledge”. According to her (Anscombe 1957, § 33ff.), what is known in practical knowledge is represented in a practical syllogism, which consists in its simplest form of a premise of something characterized as desirable by the agent and another premise stating a sufficient means to realize the goal, whereas the conclusion is the action itself. The function of a practical syllogism is not to prove the truth of a statement, as in a theoretical syllogism, but to represent an order that is always there when something is done intentionally. Anscombe’s Intention can be seen as the starting point of the modern analytical debate on action theory. Melden, too, argued that teleological explanations that state the intentions of the agent could not be regarded as causal explanations. In his argument, he makes use of a consideration originally brought up by Hume. According to Hume, a requirement for causality is that cause and effect must be logically independent from each other (Hume 2008: sec. IV, part I). But, since intentions of agents can only be identified and specified by the very action they are intentions of, there is a logical connection between them which violates Hume’s requirement that an effect should be logically independent of its cause (Melden 1961: ch. V; Melden speaks of “volitions”). In short, teleological explanations cannot not be regarded as causal explanations for there is a logical connection between intentions and actions, not a contingent one as in causal relations. This argument has become known as the “Logical Connection Argument (LCA)”. Ordinary Language anti-causalists like Alan White and Georg von Wright also proposed forms of it. Dray gave no precise description of the logical structure of his rational explanation, just that explaining an action rationally is to see it as the conclusion of a practical argument whose premises are the agent’s

Introduction  15 motives and beliefs. Von Wright also proposed the practical syllogism as the standard model of explanation of historical actions. In von Wright, the practical syllogism contains in reverse order all the steps required to explain actions, including two premises that state the agent’s intention to do x and the agent’s belief that y is a necessary means for x. The conclusion has the form “Agent a sets himself to do y” (Von Wright 1971: ch. III.4). According to von Wright, a practical syllogism thus understood is valid only “ex post actu” (the necessity of the syllogism does not imply determinism). Von Wright also pointed out (as Anscombe and Winch did) that there is a conceptual distinction between bodily behavior and intentional action, such that there is also a distinction in their mode of explanation. Only when the item of behavior is understood as intentional, the appropriate explanation is teleological, when it is not so understood, the appropriate explanation is causal (von Wright 1971: ch. IV.1  & 166). This is an anti-causalist argument that stresses the point that actions are not mere physical events, which is a central premise of all recent causalist theories of action. To conceive of an item of behavior of an agent as an action is to give a rudimentary explanation of the action (Anscombe 1957: §§ 43, 46, 47; von Wright 1971: ch. III.9; von Wright 1985: II.6). Explanation of the explanandum and understanding what it was belong together. This explanation by re-description is also important for historical explanations, e.g. something which used to be thought of as a religious reformatory movement may, after closer historical examination, turn out as a class struggle for land reform (von Wright 1971: 134; Little 2007: 1). To sum these developments up: In opposition to action theoretical causalism, some philosophers (especially proponents of the so-called Ordinary Language Philosophy of the 1950s and 60s), argued that human actions are to be explained teleologically by reference to reasons (e.g. Robin G. Collingwood, Ludwig Wittgenstein, Gilbert Ryle, Elizabeth Anscombe, Peter Winch, William Dray, Abraham Melden, Charles Taylor, Alan Donagan, Alan White, Anthony Kenny, Georg Henrik von Wright). This form of explanation includes reference e.g. to the intentions, purposes, goals, and desires of the agent in a way that lets us see how the action was meant. The agreement among these philosophers was that reasons explain actions by rationalizing them, whereas causes could not possibly do so. Anti-causalists like Melden, White, and von Wright also proposed forms of the “Logical Connection Argument” (LCA), according to which there is a logical connection between intentions and actions, not a contingent one as in causal relations. II.4  Davidson’s Advance The anti-causal movement in action theory and theory of historical explanation was clearly quite strong in the heydays of Ordinary Language Philosophy in the 1950s and early 1960s (with von Wright 1971

16  Gunnar Schumann being a late representative). But from the early 1960s on, causalism became the mainstream position in philosophy again. The causalist reaction was mainly spearheaded by Donald Davidson and his 1963 paper “Actions, Reasons, Causes”. Davidson granted the anti-causalists that it is reasons that explain and rationalize human actions (and also that rationalizations are expressed by practical syllogisms), but they could be nonetheless taken as causes, for goals, purposes, and intentions (Davidson speaks of “pro-attitudes”, especially “desires”) and beliefs of agents can be seen as mental states (which, according to Davidson, also have a physical [neuronal] description) that cause the action. So, Davidson admitted that human actions would be explained in terms of desires and beliefs of the agents, but maintained that this is still a causal explanation. Davidson’s argument for causalism was that agents can have multiple reasons for a certain action, but not all these reasons explain why the agent acted in that way. The reason which explains the action (a socalled “primary reason”) is the one that effected the action, i.e. which is also its cause. A “primary reason”, according to Davidson, consists of a “pro-attitude” and an appropriate belief, and is a rationalizing cause of the action. The issue between anti-causalists like Anscombe and Melden and causalists like Davidson can thus also be expressed as a disagreement about the kind of relation between the premises and the conclusion of a practical syllogism: for the anti-causalists the relation is conceptual, for the causalists the relation is contingent, causal. Davidson defended the causalist explanation of actions against objections of anti-causalists and proposed a form of causalism which does not require the causal laws to be known explicitly. Against the Logical Connection Argument of Melden he replied amongst other things that causal statements can also be analytic, namely when we describe an effect in light of its cause, e.g. when we say “The cause of B caused B”. From Davidson’s point of view, whether a causal statement is analytic or synthetic thus depends solely on the way how we describe cause and effect in it. Also, contra Melden, Davidson held that the agent’s desires can very well be described independently of their respective actions, e.g. in feelings and actions that they do not rationalize. So, according to Davidson, there is no need to deny a causal understanding of the relation of pro-attitude and action. It was a distinct contribution of Davidson to the debate that explanations of action can be causal without general laws playing a direct role in them. This helps Davidson to immunize causalism against the objection that action explanations (in historiography and in everyday life) do not require strict laws (Stoutland 1976: 283). That a primary reason causes an action is presented as a singular causal statement, not a law-like one. In this sense, Davidson’s theory is not a DN-theory of action explanation. Still, Davidson does not simply abandon the nomological notion of causality. He holds that when events are in a relation of cause and effect, they only need to have some descriptions that instantiate a law.

Introduction  17 Therefore, if reasons cause actions, then that does not mean there must be a law that generally links a reason and an action, but only that there is some description of the reason that correlates with a description of the action and instantiates a general law (the description of the reasons may be neurological, chemical or physical) (Stoutland 1976: 283). It is not attitudes of a particular kind that are linked to actions of a certain kind, rather a singular causal statement links certain volitional and cognitive attitudes on a particular occasion with the agent’s behavior on that occasion. In order to be able to say that there is a causal connection between a reason and an action we do not have to state or know the law explicitly, although it is implied that there is one. Recent followers of Davidson abandoned the monist philosophy of mind and the reductionist claims of the Identity Theory. The Identity Theory of mind and materialist positions, although still around, came under fire in analytical philosophy in the 1970s. Today, non-reductivist, even neo-dualist theories of mind have returned and the anti-metaphysical program of the Vienna Circle is not universally advanced. Causalists in action theory nowadays are not necessarily materialist reductionists anymore. They sometimes explicitly leave that difficult question of philosophy of mind open and concentrate on arguing for the view that actions are explained by causes. Given that most causalists nowadays refer to Davidson, it is important to emphasize that he only provided a defense of causalism and showed at best how causalism might be still a tenable position (according to his superficial understanding of the arguments of the Wittgensteinians). His only argument for causalism (namely that we determine an agent’s real reasons for acting as the ones that are causally efficacious for his action) can be regarded as presupposing causalism—he certainly did not provide an elaborate and independent rationale for causalism. (It should also be noted that this celebrated argument is everything but new: The idea that the reasons the agent states might be unreliable such that the agent’s real reasons for action have to be taken as underlying causes is, as mentioned, already to be found in Marxism and Psychoanalysis. Winch critically discusses this idea in Pareto and Weber; Winch 1958: 98, 105f., cf. also 44f.) This is surprising given the fact that most causalists of today rely on Davidson’s paper. In any case, Davidson’s view evolved into what is sometimes called the “standard view” in action explanation. In addition, Frederick Stoutland objected against the “Logical Connection Argument” that an alleged logical connection between desire resp. intention and action does not rule out a causal connection between both relata (Stoutland 1970). (This was just an attack on Melden’s specific formulation of the LCA, Stoutland himself was an anti-causalist.) In analytic action theory, the causal theory was widely accepted up to the present day, not so much because it had responded successfully and once and for all to the objections and

18  Gunnar Schumann difficulties raised by anti-causalists, but because (1) there was no fully developed and clear alternative by the anti-causalists, maybe apart from von Wright 1971. But, von Wright weakened his anti-causalist argumentation by bringing in the case of an agent who can be said to have an intention without acting accordingly (the tyrant murder case) (for a critique, see Schumann 2017). (2) Causalism dominates because it seems to capture the spirit of the scientific age, today more so than ever before. Many Philosophers believe that it has to be right, as Stoutland put it (Stoutland 1976: 271).11 Causalism is also presupposed or undisputed in recent theories of collective agency (e.g. Searle 1990; Bratman 1999). The arguments and positions of Ordinary Language anti-causalism were mostly neglected (following the decline of Ordinary Language Philosophy in general) and causalism became the default position in contemporary action theory.12 II.5  Some Objections From Anti-Causalism However, it is not that causalism is without problems. Davidson faced the question of how to reconcile his causalist theory of action with the nomological character of causality and the anomaly of the mental (i.e. the thesis that there are no strict psycho-physical laws), a question he tried to answer with his “Anomalous Monism” (which over and above is a form of a token-token Identity Theory of the mental with the physical). That project (quite a subtle and baroque theory) was arguably rebutted by authors like Jaegwon Kim. Kim argued that Davidson’s theory assigns to mental properties only an epiphenomenal role, rather than the needed causal one (Kim 1993a). Furthermore, there emerged the problem of deviant causal chains. When causalists define intentional actions as caused by intentions or pro-attitudes, then this leads to an obvious problem: A pro-attitude may cause an action in an indirect, deviant way. Davidson himself gave an example: A mountaineer, holding another mountaineer on a rope at a cliff, might want to kill his partner by letting the rope go. He thus forms the intention—and this intention upsets him so much that he loses control over his hands and fingers, and lets go of the rope. Thus, ultimately, the intention to kill the partner by letting the rope go, caused the action—but we would not be inclined to say that this was an intentional action, for the action was only caused accidentally, and not “in the right way”. Davidson gave up hope of specifying this “right way” within the framework of a causal theory of action, but that did not persuade him to give up his causalism (Davidson 2001: 80f.). Davidson has challenged philosophers who do not think that reasons are causes to account for what could be meant by “because” in sentences of the form “He did it because .  .  .” followed by a reason (Davidson 1963: 693). Let’s suppose some agent performs an action having several reasons, but in fact only acts on one of them. The causalist’s challenge to

Introduction  19 the anti-causalist is to ask them how they can, within an anti-causalist framework, account for the distinction between the reason on which an agent acted and other reasons which could be invoked to rationalize or justify that same action, but which were not the reason on which the agent acted. This has become known as “Davidson’s Challenge”. The causalist claims to have a good answer: The reason on which the agent acted is also the cause of the action. This challenge was taken up by some anticausalists, such as George Wilson (1989) and Scott Sehon (1994). Wilson and Sehon proposed anti-causalist accounts, which in turn were objected to by Alfred Mele (2000) who tries to defend Davidson’s causal view from the objections raised by anti-causalists and problems like the one of deviant causal chains. George F. Schueler has also participated in that debate13 (see this subsection). Mele’s most recent reply to Sehon in that debate can be found in this volume, as well as Schueler’s most recent anticausalist account.) Apart from this discussion, one may ask why one should take Davidson’s Challenge seriously at all? Why should anti-causalists be asked to justify their stance when, in everyday life, it is clear that there is a deep difference between teleological and causal explanations? To impose the burden of justification unilaterally on the anti-causalists is no argument for causalism. Also, Davidson’s Challenge suggests that only the causal interpretation of “because” is unsuspicious and intelligible; but this rather seems to be an expression of scientism and to ignore the fact that in everyday language the “because” in statements about reasons for actions and the “why” in why-questions are not used in a causal sense. In general, causalism is often simply assumed to be right and not argued for explicitly. And it is striking that many of the causalists’ arguments come down to claiming that the causal theory can be right regarding certain anticausalist objections, not that it is. This holds for causalists as Davidson and Mele. The latter even confesses that he has always only defended causalism against objections, and not developed a positive argument for it (Mele, in this volume: 46). Causalism is often more an expression of an “it has to be so”-attitude, a certain worldview, which is characterized mainly by the belief in the overall applicability of the natural sciences. This does not mean that doubt about the appropriateness of causal explanations of actions is an expression of supernatural, esoteric, magical, or metaphysical thinking. It is in fact the doubt that explanations of actions are given by scientific means. Causalists like Mele would complain that desires or intentions must play a relevant role in producing the bodily movements that make up an action, but maybe it is wrong to think of actions as kinds of bodily movements in the first place (cf. Schumann in this volume). Philosophers in the tradition of late Wittgenstein and Ordinary Language Philosophy, such as Peter Hacker (1996), Bede Rundle (1997), and Severin Schroeder (2001: 156f.) pointed out that Ordinary Language

20  Gunnar Schumann Philosophers like Gilbert Ryle (1949) and Abraham Melden (1961) already gave answers to the question of how the reasons for which agents acted are to be determined. An answer to the causalist’s challenge was already available before Davidson raised it. In a 2001 paper, Severin Schroeder formulated some Wittgensteinian objections to Davidson’s causalism. Davidson’s main argument (that the real reason for which the agent acted should be taken as the one which effected the action) does not speak in favor of the causal theory. In fact, we normally do not conduct any empirical investigations into the causes of an action when we seek for an explanation. Instead, we determine the reasons “that lead to the action” by looking at the context of the action. Why did Peter go outside? (1) The answer depends on what we know about the circumstances of the action. We often recognize known patterns of human action in the agent’s behavior.14 (2) It depends on what we know about the person. (If we know that the person never cares about her health then she will not have walked out for health reasons. Still, of course, such background knowledge will not always be sufficient to settle the matter). And (3) we may ask the person. Of course, the agent’s reply may be insincere but to decide if it is or not, we consult (1) and (2) (Schroeder 2001: 153f.). Schroeder then stresses Wittgenstein’s and Anscombe’s considerations against causalism: A causal explanation is a hypothesis, while sincerely giving one’s reasons for an action is not the statement of a hypothesis. Except for some special cases, the agent has the absolute authority to indicate what his reasons are—but this absolute authority does not exist with regard to causes. Furthermore, (as Schueler in his contribution to this volume also notes) Davidson’s move to speak of desires instead of goals, aims, and intentions helped facilitate his causalist view, for in one sense of the term, “desire” can refer to a mental state, such as a craving for chocolate or a preference that it would not rain on Sunday. Desires in this sense can indeed be described independently of their respective actions, as Davidson argued, for one can desire to eat a cookie without actually doing so (and vice versa). But it may very well be doubted that this holds for other pro-attitudes like intentions, goals, or decisions. Davidson takes a primary reason to consist of a pro-attitude towards actions of a certain kind and a belief that the action is of that kind. But, reasons for actions are usually states of affairs, not mental states like beliefs and desires (or intentions) (Schroeder 2001: 150f.). The reason why I  switched on the light was to see what’s going on, not the fact that I wanted/intended/desired to see what is going on. The reason why I squatted was that my shoe laces are untied, not the fact that I believed or knew that my shoe laces are untied. It is not the pro-attitude or the belief qua mental state that counts as a reason, but the content of the pro-attitude or the belief.15 Through the subtle but illegitimate shift from reasons as propositional contents to psychological facts about the agent, Davidson’s causal construal of human action explanation gets off the ground. (See Glock’s and Schmidt’s paper in this volume for a defense of

Introduction  21 causalism about action explanations from this anti-causalist objection.) In general, to think of intentions, desires, or beliefs as mental objects, states, or processes suggests thinking of them as a kind of entity (mental or neuronal)—which fuels the causal construal of action explanations (Schroeder 2001: 151). There are still other objections to causalism. So, Davidson’s position seems to be everything but unobjectionable. Accordingly, the causalist mainstream seems to have come under fire (again) in recent years by an anti-causalist “fin de siecle movement” (Wilson 1989; Hacker 1996; Rundle 1997; Schroeder 2001; Schueler 2005; Sehon 2005; et al.).16 Still, it is just a minority of contemporary philosophers of action who maintain opposition to causal explanations of intentional human conduct. II.6  Recent Causalism in Philosophy of Historiography The situation is similar in philosophy of historiography and theory of social sciences. Although Hempel’s DN-Model is no longer seriously debated as a paradigm of historical explanation (there are exceptions— cf. below), the view that historical and social explanations are causal explanations, because human actions are to be explained by causes, is predominant in recent philosophy of historiography and theory of social sciences (e.g. Hedstrom and Swedberg 1998; Tucker 2008; Macdonald and Macdonald 2008; Kincaid 2009; Gerber 2012; Risjord 2014: 89ff). Theorists of socials science like Hedström and Swedberg (1998: 23f.) invoke James Coleman’s model of explanations of collective actions, the “macro-micro-macro” model. According to that model, regularities (statistical correlations) between social macro phenomena have to be explained by “mechanisms” that involve individual agents and their actions. A proper social explanation has to show how phenomena on the macro level have an impact on individual agents and their social relevant behavior on the micro level and how the latter in sum has consequences for the macro level again. So, there are three types of “mechanisms”: (1) individual agents face certain social conditions that have an influence on them (“macro-micro mechanism”); (2) the agent’s desires, beliefs, and occasions generate certain actions (“micro-micro mechanism”); (3) these individual actions have intended or unintended collective consequences (“micro-macro mechanisms”). According to Hedström/Swedberg the first two are psychological or socio-psychological mechanisms, which have to be spelled out causally (though Hedström/Swedberg are silent on which account of causality). This model is the modern version of the physicalist picture of the workings of society which is around since early modern times, and it rests explicitly on a causal theory of action (Hedström and Swedberg 1998: 12; cf. Hacker 1996: 543–547). The once strong opposition to causalism in historiography and the alternative proposal of historical explanations as rational explanations of past actions (cf. Collingwood 1946; Dray 1957; Winch 1958; Passmore 1962;

22  Gunnar Schumann Donagan 1966; von Wright 1971) have largely disappeared in recent discussions of the topic. It is fair to say that their positions and arguments have not received the attention they deserve. Concerning philosophy of historiography, the concept of narratives dominated the issue of explanation in the last decades (Arthur Danto, Hayden White, Frank Ankersmit). Although narrativism has been invoked to draw a distinction between sciences and the humanities (Mink 1998), the concept of narrative explanation was not introduced so as to oppose causal approaches of historiographical explanation, but was understood as compatible with it (Danto 1965: ch. XI). Narrative explanations were introduced against the claim that historiographical explanations are nomological in character; nonetheless narratives were still given the task of delivering a causal story about how a historical event came to be. Thus, narrativism in historiography does not normally differentiate between causal and teleological explanations and risks collapsing into a form of causalism (Stueber 2012: 24). Others have claimed that regularity accounts of causation are inappropriately deployed in historiography and social science. They have suggested that the DN-model of historical explanation should be substituted not by teleological explanations but by a conditional (Mackie 1965) or counterfactual (e.g. Gerber 2012; in this volume) account of causation. The counterfactual account of causation in particular is often deemed to be capturing the form of explanation that operates in historiographic accounts of the past. But these accounts face objections against the respective theory of causation (cf. Tucker’s contribution to this volume) like (1) overdetermination: There might be cases, in which effect E would have been produced by another cause C2 regardless of whether C1 would have existed or not. In these cases, it could not be said that if C1 had not been, E would not have come into existence; (2) epiphenomenon: “If the iron bar did not glow white, the bar would not be liquid” might be true but we still would not say that the white glowing of the bar was the cause of its being liquid. (3) Insufficient differentiation between conceptual and causal relations: “If yesterday had not been Monday, today would not be Tuesday” or “If my sister had not given birth to a child, I would not be an uncle”. But these are conceptual relations; my sister’s giving birth is not the cause of me being an uncle. (4) The counterfactual slogan by itself does not give any criterion for cause-selection. Of all conditions that have a counterfactual relation towards effect E, we do not say all are a cause of the effect and do not state them in a causal explanation of E. The big bang or the successful prevalence of vertebrae at a certain point of evolution of species were counterfactual conditions for the occurrence of homo sapiens and thus of the battle of Waterloo. But it would be absurd to demand that the big bang or the successful prevalence of vertebrae be included in any explanation of the battle of Waterloo. In any case, theoreticians that want to invoke counterfactuals will have to add important

Introduction  23 qualifications to their counterfactual theory. Apart from these objections there is the issue that counterfactuals are neutral with respect to teleological and causal explanations. Counterfactuals can be used, in the context of rationalizations, to identify the reason why an agent acted by ruling out other goals through the use of counterfactuals. For example: “Had Mary really intended to kill John she would not have done it in front of her children”. For example, Sehon (2005) treats counterfactuals in the context of teleological explanations as the means to provide an answer to Davidson’s Challenge. Therefore, while philosophy of action has witnessed a teleologist backlash against causalism in the last two decades, causal explanation talk is alive and well in philosophy of history. The focus of the debate there has shifted from the quarrel between causalists and teleologists to a discussion concerning the role of narratives in historiography. Such a discussion tends, more often than not, to assume the adequacy of a causal, even if non-nomological, form of historical explanation. Although there have been some dissenting voices (e.g. Castellani 2007; Rosenberg 2008: ch. 2), causalism is flourishing in recent analytic philosophy of historiography and of social sciences. Even neo-Hempelian approaches (Scholz 2008, 2014; Schurz 2011) crop up again that attempt to improve the DN-Model with so-called “normic laws” (non-strict general statements of the form “If A, then normally B”). In general, one could say that postHempelian theories of historiographical explanation attempt to defend causalist models of explanation—only without the assumption of strict laws between cause and effect: the narrative approach, the causal mechanism approach, counterfactual approaches, the normic law approach (in a way, even Davidson’s anomalous monism attempts to have causal explanation without explicit reference to laws being necessary). Taking stock, the history of the debate between causalists and teleologists resembles the motion of a pendulum: First, there was a causalist mainstream in early 19th-century positivism, which was countered by a strong movement of “Verstehen” of the late 19th-century hermeneuticists. This in turn was countered by the unity-of-science movement of the logical positivist of the 20th century, followed by an anti-causalist counter by Collingwood, Wittgenstein, and their followers. From the early 1960s on, causalism has become mainstream in philosophy of action and human sciences again. The causal view of explanations of human action seeks to integrate human action in the scientific worldview. The temptation to explain the course of human history or any of its subsystems just in the same way as the behavior of a macroscopic physical phenomenon by analyzing it into its parts and their movements is present until today.17 According to the modern scientific worldview, the world is stratified and each higher layer is based on an underlying one and can be reduced to it, at least in principle (Oppenheim and Putnam 1958). The same applies to the

24  Gunnar Schumann (assumed) causal relations governing each layer: Those between entities of the higher layer depend on those of the lower ones. Physics explores the lowest layer: the elementary particles and most general laws of the material universe; chemistry explores the compounds of these elementary particles and their reactive behavior; biology has as its subject even more complex entities such as e.g. proteins, cells, organs, and organisms to which we ascribe life and their behavior. Behavioral science deals with higher living beings, to whom we attribute psychological predicates and abilities (like consciousness, emotions, and will), and psychology is concerned with humans. People form groups and social communities of various sizes and degrees of cohesion, which are the subjects of sociology and social sciences like law, economics, political sciences, historiography. There are similarities in the intellectual dealing with (some) physical and (some) social systems (as analyzability, quantifiability, calculability, predictability, sometimes experimentation). Due to its use of large mathematized theory apparatus, Economics appears to be most advanced in adopting the methods of sciences (as was approved by members of the Vienna Circle). Frequently, Economics is taken as an ideal for sociology, which again is sometimes taken as an ideal for historiography. The most prestigious prize for scientific achievements is awarded in Physics, Chemistry, Medicine, and, since 1969, in Economics. Nowadays, the Cartesian model of the bifurcated world has been replaced by the model of a stratified world of hierarchically ordered levels of entities and their characteristic properties and interrelations. But the model of a stratified world is not very different from that of a bifurcated world (Kim 1993b: 189ff.): The general problem of how the mind relates to the body is still the prevalent problem of philosophy of mind. At this point, causal theories of human action take on an important strategic role: If it were possible to grasp mental phenomena such as intentions, decisions, goals, and purposes (which notoriously seem to escape the physicalistic-scientific access) as causes, then the apparatus of scientific explanation could be applied (at least “in principle”) to human actions, both individual and collective, such that all layers from persons upwards would be accessible to the physicalistic-scientific methods. (The gap between the level of individual action explanations and the explanation of macroscopic social phenomena like parties, companies, organizations, states, markets, wars, and revolutions would be bridged by specific theories of collective action. Their task is to provide explanations of how individual mental states and/or actions can make up collective actions and not a mere aggregate of individual actions.) So, if reasons for actions could be understood as mental or rather neurophysiological causes, then there would be no need to move away from the program of scientific-physicalistic explainability of human actions. (That the hopes in this respect are still very high is shown by the fact that nowadays there are disciplines like “neuro-economics” which attempt to investigate

Introduction  25 human behavior in certain economic decision situations with the help of neuroscientific methods.) Causal theories of action therefore take on an important hinge function—if they are correct. If it can be shown that human actions in general require a causal explanation, then the central premise of a causalist theory of historical/social explanation can be taken for granted. Or conversely: If it can be shown that human actions cannot be appropriately explained causally then the causalism about historiographical explanation collapses and the notion of a “historical cause” (“social cause”) cannot be taken to have the same meaning that the term “cause” has in the natural sciences. The debate between causalists and anti-causalists in philosophy of historiography is thus an application of the debate between causalists and anti-causalists in action theory. This is why both fields of debate, those concerning the appropriate manner of historical explanation and individual human actions, are connected and are brought together in this volume. Are human actions, individually or collectively, ultimately determined by causes—be it by historical or sociological regularities (strict or loose), by “institutional structures”, by mental or neurological states or can the agents’ reasons for their actions be taken at face value in general? Despite the contemporary dominance of causalism, the dispute over the appropriate form of action explanation, with its implications for historical explanation, is not settled once and for all. In theory of action and philosophy of historiography, as well as in academic historiography, sociology, and economics, there have been fierce methodological discussions about appropriate forms of explanation. The quarrel between causalism and anti-causalism is thus of ongoing interest. The issue is not only relevant for philosophers of action, or for philosophers of historiography, or for the philosophy of the human sciences. The dispute touches on our picture of human beings and their ability to act freely and voluntarily, and how far they can be considered responsible for their deeds. As it has been rightly noted, the quarrel between causalists and anti-causalists is not only of academic relevance but, in the end, of existential importance (von Wright 1971: 32; Kim 1973: 388).

III.  About the Contributions Given the aforementioned debates, the chapters in this volume address issues which remain controversial to the present day: What is involved in the concept of explanation? Is the explanation of human actions analogous to the explanation of physical events? How are human actions to be explained—by reference to causes such as inner states or brain states, or maybe external (social) causes? Do causal explanations involve generalizations or do causal relations have to be conceived as counterfactual relations, especially in historiography? Are there causal relations between mental and physical phenomena and, if so, how are they to be

26  Gunnar Schumann understood? Or are human actions rather to be explained by reference to reasons—understood as distinct from causes? According to the latter position, reasons explain actions by rationalizing them and agents are conceived of as rational agents. What are reasons, and what is the relation between a reason and an action? Are intentions, desires, and purposes mental states in a strict sense? Are they maybe dispositions? Reasons seem to have an internal connection to actions and normative force, whereas causes do not. Also, Davidson’s psychological construal of reasons seems implausible—are there alternatives to his “psychocausalism” available? Or are teleological and causal explanations of action not mutually exclusive at all? Should we maybe adopt a pluralist account according to which reasons have to be understood non-causally whereas explanations of actions still have to be understood causally? What about the explanation of past human actions? Few would be willing to deny that history consists of human actions, be they individual or collective. What then is the right way to explain historical actions— especially, when the agents are long gone and a good deal of the context of past actions is lost? The central aim of this volume is to help revive the debate between causalists and anti-causalists in philosophy of history and philosophy of action. Therefore, one part of the contributors represent the causalist side (Mele, Gerber, Glock/Schmidt), and the other one the anti-causalist side (Schueler, Löhrer/Sehon, D’Oro, Schroeder, Keutner, Schumann) while further contributions may be taken as reconciling or neutral positions on the issue (Sandis, Little, Kincaid, Tucker). The volume comprises contributions from action theorists (Mele, Schueler, Sehon/Löhrer, Sandis, Schroeder, Glock/Schmidt) and philosophers whose research-focus lies in philosophy of historiography and social science (Kincaid, Little, Tucker, Gerber) or both (D’Oro, Keutner) with the aim to make clearer how the problem of historical explanation is embedded in action theory and how they are interrelated. The first chapter is by Alfred R. Mele. In the beginning, Mele confirms that he has always defended causalism against objections, not argued for it. Mele then continues a debate on an attempt of George Wilson to answer Davidson’s Challenge. Here is some background information to the reader unfamiliar with recent developments in causalism-anti-causalism debate in action theory: Mele (2000, 2003) came up with a variation of an example originally from Wilson: Norm has three reasons to climb up a ladder since he has three things he wants to retrieve from the roof of his house: his hat, his tools, and some bricks. In the example, it is further supposed that Norm actually only climbs up in order to fetch his hat. The causalist’s challenge to the anti-causalist now is, how he would account for the fact that Norm only climbs up to get his hat—without invoking that the desire to get his hat caused Norm’s action of climbing the ladder. The causalist claims to have a good answer to that challenge, for the desire

Introduction  27 to get his hat was the one that caused Norm’s action whereas the other reasons were not effecting it. Wilson proposed an anti-causalist answer to Davidson’s Challenge. Wilson said that Norm’s action of climbing up the ladder is explained by his desire to get his hat. This was the purpose and Norm’s bodily movements were aimed at fulfilling this desire. Wilson also presents some contextual conditions that have to be met in order to say that bodily movements aim at the fulfillment of a desire. It is part of these conditions that Norm knows both what he is doing and what he will do next, and, as he performs his movements, that he is capable of adjusting his behavior if he believes his hat had changed its position. Mele objected that Wilson did not succeed in giving a satisfactory response to Davidson’s Challenge, for an example could be thought of in which all of Wilson’s conditions for intentional acting are met without there being an intentional action. Suppose that halfway on the ladder, Q-signals from scientifically far advanced Martians hit just the right input on Norm’s muscles, such that Norm continues to climb the ladder as if everything were natural. But then, according to Mele, it is wrong that Norm was the agent of his movements. Norm’s bodily movements would be in accordance with his desires respectively intentions, without these movements being explained by his desires respectively intentions. To meet Wilson’s contextual conditions of intentional acting, Mele adds the assumptions that the Martians aim to make Norm believe he is acting, whereas in reality they read his thoughts and then control him for a while—but stop when they realize that Norm is changing his plans and then give him back full control over his movements. Mele’s thought experiment is to show that all the conditions Wilson demands could be fulfilled without Norm acting intentionally. Norm’s desire respectively intention played no direct role in the production of his bodily movements (According to Mele, it could not even be said that Norm tries to move his legs for the Martians suppress that, too). Only the assumption of a causal connection between Norm’s volitions and his movements could account for that. Mele uses his Martian example to defuse another reply to Davidson’s Challenge by Sehon (1994). Sehon proposes to invoke counterfactuals in order to distinguish between reasons an agent had and the reason that effected the action. Does Heidi put a heavy book on the top shelf because it belongs there, or because she wants to train her arm? The answer, says Sehon, depends on a series of counterfactual conditionals. For example: Would she have put the book on the bottom shelf, if it had belonged there or rather on the upper one? And would she have put something on the upper shelf if something more suitable than a book had been available to train her biceps? Mele objects that the truth of these counterfactuals is dependent on the truth of whether Heidi was able to aim her behavior at all. The story with the Martians and their Q-signals could be brought up again here: Heidi, when hit by the Q-signals as Norm before, would then not be in control of her arm movements, for her desires and intentions

28  Gunnar Schumann had no causal role to play in the production of them. In his latest book (2016), Sehon again tries to defend Wilson’s contextual conditions as sufficient for acting. Even in Mele’s example with the Martians and their Q-signals, Norm would perform an intentional action (let’s assume now that Norm knows about the interference of the Martians): It is then as if Norm no longer performs his action via a normal physiological way with efferent nerves and muscle-stimulation, but via the freaky one including the Martians and the Q-signals. Just like god in occasionalism, the omnipotent Martians are the executive performers of Norm’s movements. Mele replies to that line of argument in his contribution to this volume. Also, in his article, Mele rebutts Sehon’s critique of earlier work of him on deviant causal chains. Mele sticks to the view that primary deviant causal chains are not a knock-down objection to causalism.18 George F. Schueler’s chapter discusses Mele’s defense of Davidson’s Challenge after some introductory remarks on the general dispute between causalists and anti-causalists. Schueler recapitulates the quarrel between Wilson and Mele and also brings the criticism that Mele’s enhanced example of the sensitivity of the Martians to Norm’s intentions brings this case close to an occasionalist scenario in which Norm’s action would still count as intentional. Schueler argues that the general relevance of the right construal of this example is of minor importance. What the Q-signal example shows is that it is a necessary condition of intentional action that it must not be caused by anything beyond the agent’s control as in cases of drug-induced movements. But all of that leaves open whether the teleological structure of action explanations is only apparent and depends on underlying causal links. Schueler goes back to the core of the issue and points out that Davidson’s move to speak of desires instead of goals, aims, and intentions facilitates the causal view, for in one sense of the term, “desire” can refer to a mental state, such as a craving for chocolate or a preference that it would not rain on Sunday. But “desire” in the sense relevant for explanations of intentional actions just means that the agent has a certain goal or purpose in doing an action. What also speaks against causalism is that the purpose of an intentional action is a logically constitutive feature of that action, just as the propositional content of a belief is a constitutive feature of that belief. So the connection between an agent’s intentional action and the goal she was pursuing in performing it is not of the right sort for causation, any more than the connection between a belief and the content of that belief is. Schueler further argues against the view that bodily movements produced by desires be intentional actions. Therefore, Causalism is wrong. Causality only plays a role in agency in the fact that means and purposes are connected causally and in the fact that the movements instantiating an intentional action must be causally explainable. In their chapter, Guido Löhrer and Scott R. Sehon also defend the view that the nature of reason explanations of human action is irreducibly

Introduction  29 teleological rather than causal. This has a bearing on historical explanations, because historiography is concerned with the explanation of significant actions and events. These will typically involve a complex of many distinct actions (individually or jointly executed, cooperative or antagonistic) and their consequences. Long explanatory stories will involve both statements about the purposes of individual agents as well as causal effects of those actions, and then further purposes of other agents. It concerns a big mixture of teleological and causal explanation. Löhrer and Sehon go on to argue that the causal relations between events are typically marked by regularity that are investigated by science whereas in teleological explanations we employ rationalizing principles. We attempt to make the observed behavior intelligible by ascribing to the agent sensible goals and take the agent’s action to be thought to help achieve the goal. Normally we would ascribe the goal(s) that make(s) the most sense of her behavior. Such rationalizing principles are used in historical explanations. When it comes to finding the real reasons for which a historical agent acted, the context often has to be included in the interpretive project. This often requires the interpretations of other actions of the involved agent(s). This also holds for seemingly irrational actions which on closer examination turn out as intentional actions with unusual goals. In arguing against causalism, Löhrer and Sehon stress that the causalist need not deny that we use rationalizing principles in the process of determining the right reasons. According to the causalist, they might just play the role of a heuristical device to find the causes underlying the reasons explanation. But this raises a tricky question for the causalist: Why think that rationalizing principles will work as a heuristic in the case of finding the causes of human action and not in cases of causal explanation of physical phenomena (momentum transfer, supernovae)? In the physical realm we would not dream of explaining in terms of purposes and intentions. In the following, Löhrer and Sehon discuss a possible explanation for why rational principles are a heuristic for causes, namely that natural selection has us “designed” to be rational beings. They reject this suggestion. That humans act rationally does not license the view that we can therefore know that the behavior of the humans is caused by physical or mental states. Saying that agents can believe propositions and desire states of affairs is not tantamount to the claim that the physical causes of behavior are identifiable with those beliefs and desires. Also Giuseppina D’Oro argues for an anti-causalist view of human action explanations. In her chapter, which is to a large extent written as a dialogue between a causalist (loosely modeled on Mele’s position) and an anti-causalist, she also is concerned with Davidson’s Challenge. To repeat, Davidson argued that the anti-causalist could not distinguish between reasons for acting that may justify the action but are just made up by the agent and the real reasons for which the agent acted. The causalist could account for this: The real reason was the one which effected

30  Gunnar Schumann the action. D’Oro disagrees: The real reason provides an answer to the question: “What reasons for acting makes the most sense of the action in the light of all I know about the agent?” but not “What reason caused the agent to do the action?” Thus, the real reasons are not different in kind from the make-believe reasons: They are as justificatory as all other reasons and not preceding Humean causes. The more is known about the agent, the better the anti-causalist may decide between different rationalizations. Still, there might be cases in which one, even the agent himself cannot decide for which reason in particular the agent performed an action. Here the causalist claims to have an advantage over the anticausalist, for he can still draw a distinction between the real and the mere rationalizing reasons. The anti-causalist stresses that the causalist himself could not identify the real reasons in such a case. The causalist replies, that his point is not an epistemological one, but that, according to him, there is a fact of the matter whereas the anti-causalist would be committed to admit that there is none—just like there would be no fact of the matter in the explanation of actions of fictional characters. But real persons have real psychological processes with causal powers and so the explanation of actions must be based on them, not on mere rationalizing narratives. The causalist thus put his point as this: “In virtue of what it is true that S acted for R?”. The anti-causalist replies to this that still we understand persons and their reasons for acting in the same way—no matter if they are real or fictional characters. Also, in the type of case at issue, one might say that an agent had several reasons for acting. In any case, the real psychological process does not do the explanatory work: Even granted that it did occur before the action, its occurrence is not relevant to the correctness of the explanation. D’Oro ends her chapter with this alternative: If Davidson’s Challenge is to be understood epistemologically, causalist and anti-causlist draw even. If Davidson’s Challenge is to be understood on the level of truth makers, then in cases of doubt about the real reason for an action, a criterion would be imported which is alien to the spirit of the question “Why did agent A do X?”, for that question is about the agent’s motive for the action, not about the start of a causal chain. Severin Schroeder, in his contribution to this volume, formulates some objections to a recent attempt by John Hyman to reconcile Davidson’s causalist view with Wittgenstein’s plea for reason explanations of human actions. According to Hyman, action theoretical causalism is a tenable position if one takes causal relations not to be nomological in character but as directly observable and takes desires to be dispositions. Hyman also tries to defuse the deviant causal chain objection against causalism by pointing to the general problem of distinguishing between dispositions’ manifestations and their side effects: Desires, taken as dispositions, explain actions causally in either case—no matter whether the action was effected in a deviant or non-deviant way. Schroeder wants to show that

Introduction  31 Hyman’s account is not convincing for three reasons: (1) The problem of deviant causal chains cannot be so easily defused. Admittedly, deviant causal chains do not rule out that intentions might be causes of actions, but that is not the issue. The issue is whether the word “because” in action explanations has a full causal meaning or not. (2) Intentions cannot be taken to be desires. If I hand over the salt to the person next to me after he asked me, then my action is usually not accompanied by something that could properly be called a desire. And, dispositions are not causes. The point of a dispositional explanation is to present the explanandum as an instance of a general pattern. If someone’s politeness surprises us and we wonder whether he is ironical or wants to flatter us, then it makes sense to say that that is his regular behavior: There is no hidden agenda, he is just really a polite person. And, if desires are taken to be ad hoc dispositions they cannot be taken to be causes of an action but only as a misleading way of reiterating the fact that the triggering event did indeed cause the intentional action. In general, dispositional explanations are just about general tendencies, not about causes.19 Also, we have no idea about what the material bases of psychological dispositions are. (3) Against Wittgenstein’s point that we have a first-person authority in stating what our reasons are, whereas we do not have this type of authority about causes of events, Hyman objects that we can also be immediately certain of causes for actions. Schroeder admits this, but points out that this again is not the issue, because the crucial difference between reasons and causes is one between an authority conferred by our concepts (not a matter of knowledge at all) and any kind of empirical finding, be it immediate perception or repeated observation. The firstperson authority does not consist in truth-apt reports of a state of affairs but marks certain kinds of sincere avowals of a person’s reason as constitutive of the concept of a reason. An agent’s reason simply is defined by his sincere avowal of his reason. Thus, Wittgenstein has found a criterion for what makes a reason the reason for which the agent acted. The chapter by Hans-Johann Glock and Eva Schmidt explores whether a version of causalism about reasons for action can be saved by giving up Davidsonian psychologism (i.e. the position that it is reasons qua mental states that explains the action) and endorsing objectivism (i.e. the position that reasons are a states of affairs), so that the reasons for which we act are the normative reasons that cause our corresponding actions. They address two problems for “objecto-causalism”, actions for merely apparent normative reasons and actions performed in response to future normative reasons—in neither of these cases can the reason for which the agent acts cause her action. To resolve these problems, Glock and Schmidt move from objecto-causalism to “objecto-capacitism”, which appeals to agential competences manifested in acting for a reason. They briefly apply this view of reasons for action to historical action explanations. As is not untypical for causalism in action theory on the strategical

32  Gunnar Schumann level, the paper tries to develop a version of causalism that is still tenable in light of certain anti-causalist objections, such as the normativity of and objectivism about reasons. Constantine Sandis’ chapter considers various constraints placed by philosophers on what they take to be a shared concept of a reason for action. Sandis aims at showing that these constraints are incompatible with one another, proceeding to argue that we would therefore do better to embrace a conceptual pluralism. On such a pluralism, there is no such thing as the concept of a reason for action. Interminable debates about the nature of reasons for action arise precisely because no single thing called a “reason for action” can perform all the varying functions that philosophers require of it. As with products such as three-in-one shampoos, each individual function is performed at the expense of others. Unlike such stuffs, however, some of the desired functions of reasons cannot be combined at all. Sandis concludes that neither disjunctivist nor anti-disjunctivist accounts of reasons for action are capable of providing a unifying account of them. In the first part of her chapter, Doris Gerber wants to give a very general argument against the assumption that causality can be reduced to (strict) regularity; this argument stresses the point that causality has essentially something to do with change and transformation and that the concept of change is more fundamental than the concept of regularity. The counterfactual analysis of causation is more accurate to meet this crucial feature of causation. In the second part she introduces very briefly the concept of action she is relying on and deals with the more specific problem of mental causation. And in the third and last part Gerber turns to specific aspects of historical explanations, namely the fact that historians are typically interested in rather far-reaching consequences of human actions and the fact that the so-called historical meaning of an event can change over the course of time. These aspects seem to challenge the assumption that historical explanations have a causal structure, however, one can embrace these aspects by modifying the general counterfactual analysis in a minor way. Harold Kincaid sketches a naturalist approach to the causalism/anticausalism debate. His chapter consists of two parts. In the first, he argues for several points: that the debate over reasons as causes cannot be resolved by analyzing concepts and testing them against intuitions and imaginary cases; that the behavior/action distinction is not some fundamental conceptual truth but a distinction of value only where and when cognitive science provides good empirical evidence supporting it; that appeals to reasons as causes are best justified if they fit into empirically well-supported results; that results across the cognitive sciences suggest that much human behavior is best explained by processes that do not look anything like reasons, creating problems for causalists and acausalists alike; that traditional acausal accounts border on mystical metaphysics,

Introduction  33 and, nonetheless, there is an approach based on work by Dennett and on revealed preference theory in microeconomics that can make sense of a certain kind of acausal story about human action. In section 2, Kincaid elaborates this story by looking at causalism/acausalism approaches to social entities. The social sciences standardly treat macrolevel entities— states, bureaucracies, and interest groups for example—as agents in the same sense that humans are studied. Here, a causal story—that these social entities have reasons that causally explain their actions—seems highly implausible. In the end this approach be a large stretch, for it is hard to see what would count as a group reason that explained the actions of collective entities. Kincaid claims that his Dennettian revealed preference approach, however, makes good sense here. According to Kincaid, we need to ask: Do the choices of such groups make for a consistent utility functions? Do these functions allow us to find explanations—real patterns—describing their behavior? The chapter by Thomas Keutner reexamines the discussion about reasons and causes before Davidson. Keutner first gives a sketch of Robin Collingwood’s approach. He categorizes Collingwood’s approach as monistic, for Collingwood argues that only historiography is the appropriate place for causal explanations, while their use in the sciences is only a metaphoric derivation from this primary use. According to Collingwood, the idea of causality is in its core an anthropomorphic one (agential causation), such that even natural phenomena are in the end explained as if they were human phenomena. Hempel was a monist of the opposite sort when he demanded to assimiliate historical explanations to scientific ones by making use of general laws. Keutner then discusses some pluralist approaches (Gardiner and Dray), according to which rational explanations and explanations based on regularities are both appropriate in historiography, although they should be used complementary. Third, another type of relationship of both methods is characterized: that of a language game of mutual (pragmatic) contradiction: Explanations invoking efficient causes may be used to contradict a given rational explanation and vice versa. Gunnar Schumann aims to show that explanations of human actions (be they past or present) are not causal explanations at all. He argues that the appropriate form of explanation of human actions takes the agent’s intentions and beliefs as premises of a practical syllogism, from which the explanandum, the action, follows logically, not causally. Schumann argues for his anti-causalist position with the help of an improved version of the so called “Logical Connection Argument” (among others). In general, to explain a human action requires an investigation of the action’s context to determine the goals and intentions of the agent. This is applied to past human actions: Historians comb through the bequeathed source data, to get a picture as precise as possible concerning the context of past actions and to reconstruct from it the intentions, goals, purposes,

34  Gunnar Schumann and beliefs of the historical agents. So, historical explanations work like everyday explanations of contemporary’s actions: As e.g. the crossing of a street by a contemporary agent under certain contextual conditions is the expression of his intention to go to work, so e.g. Caesar’s crossing of the Rubicon is the expression of his intention to seize power in Rome under the historically transmitted circumstances of his action. Thus, according to Schumann, to explain an action by a reason is not to refer to another event which is a Humean cause of it, but to embed the action in a context such that it can be understood, what the agent went after and thereby it will be understood, what the action was. Daniel Little attempts to reconcile the view that there is causation in history and society and the view that human agents are self-conscious, intentional beings, by arguing for three ideas. (1) Historical and social events may very well be causally explained, there is something like social causation. (2) The best way of understanding social and historical causation is through discovery of underlying causal mechanisms and powers: real concrete social arrangements through which one social event or condition produces another event or condition. (3) Most important, Little argues for a particular view of the substrate of social causation: socially embedded actors who act out of an expression of their values, goals, emotions, beliefs, and mental frameworks in the settings of a range of institutional and structural circumstances. According to Little, it can be shown that the same causal structure can be discovered among events and conditions in the social world and the natural world. We can observe and validate relations of necessity, sufficiency, and causal relevance among social events and processes, and we can identify some of the mechanisms that underlie these relations. Causal order in the social world can be observed, investigated, and documented. There is a substrate for any particular domain of causation, the features of which constitute the “necessity” or “push-force” that exists between cause and effect. Whereas in the physical world, the necessity associated with causation is natural necessity, in the case of the social world the idea might be captured along the lines of “agentic necessity”: “Given a social environment populated with actors something like this and embodying rules and institutions something like  that, change A  brings about outcome B through the actions of these ordinary actors” (Little, in this volume: 237). Social causes are the result of motivated actions of concrete social agents, and these agents are not subject to anything analogous to the laws of nature. Also, it would be pointless to look for general laws of bureaucracy, the military, or colonialism. The substrate to causal order in the social world is the fact of a population of socially constituted, socially situated actors whose beliefs and actions constitute social entities. Human action is meaningful action, and we cannot make sense of it without attributing meanings, intentions, and frameworks of understanding and desire to the individuals who constitute a social encounter. It is an important

Introduction  35 component of social science and ordinary life to attempt to reconstruct the nature of the beliefs and representations that are held by individuals in various settings. This was clear to the hermeneutic approach to social life (e.g. Dilthey), and it is the core of the philosophy of history expressed by Collingwood and Walsh. But the classical hermeneutic approach tended to underestimate the importance of causation and mechanisms in the social world. Mechanisms remain crucial for historical explanation because we want to discover the causally relevant events that produced a given historical outcome, e.g. the success of the Chinese Communist Revolution or the meltdown of the Chernobyl nuclear reactor. According to Little, this view of historical explanation results in a resolution of the key riddle within the philosophy of history, concerning whether historical knowledge is causal or interpretive. With an actor-centered understanding of the nature of social causation, we can maintain the validity of causal explanations of social and historical outcomes while at the same time recognizing the meaningfulness and intentionality of human action. In order to explain historical outcomes we must both interpret actors and identify causal connections among institutions, value systems, and individual actors. Aviezer Tucker thinks that much of the debate about causation in philosophy of historiography, i.e. about the covering law model and its alternatives, is “trapped in 1940s amber”. He argues for that in historiography the concept of cause is less important than that of what he calls origins. Not inference of causes, but “the inference of origins are the primary inferential pursuit of historians” (Tucker, in this volume: 251). Since past times are irretrievably lost and over, knowledge of it is gained by inference, not observation or intuition. Everything we know about the past, including causal connections between past events, is derived from information in the present that was transmitted from past events, processes and, causes and effects. Tucker calls the entities in the past that send out information origins. Origins transmit information to receivers, i.e. to what historians call historical sources. In this way, for example, cosmic background radiation transmits information from the origin of the universe to scientific instruments or species transmit information about their properties and ancestry via DNA to descendant species or texts transmit information from the authors thoughts to hard disk drives and readers. In the course of his chapter, Tucker explicates his concept of origin by distinguishing it from the concept of cause. He discusses the most prominent theories of causation and demonstrates why origins are to be conceptually distinguished from causes. But Tucker is no skeptic about causation in history. Quite contrary, causal sequences in history (like the radicalization of the French Revolution) can be the origins of information and can be inferred from historical documents. But, according to Tucker, the concept of origins is primary compared to that of cause, because the extent to which historians can offer causal explanations of past events is entirely dependent on whether information from the past

36  Gunnar Schumann reaches receivers in the present and the origins of which can be inferred. Conceptions of historical explanation like causation or understanding of past minds are derived from origins. With respect to the question if there is a principal difference between causal and rational explanation of action, Tucker holds that there is no difference in principle and that the link between motive and action is one between cause and effect. Furthermore, that Tucker takes the concepts of origin, receiver, and information transmission to be the same for the investigation of the past of nature and of history in the narrower sense shows that Tucker sees no principal difference in the epistemology of sciences and historiography. So, despite his reservations against the concept of cause in historiography, Tucker’s approach nonetheless seems to be quite scientifically in spirit.

Notes 1 In German language this difference is represented by different interrogative pronouns: “Wozu?” and “Wofür?” on the one hand and “Warum?” on the other, whereby the former can be substituted by the latter, but not vice versa: One may ask “Wozu / Warum hast du das getan?” (“Why did you do this?”) but it would make no sense to ask: “Wozu hat die Billardkugel angefangen zu rollen?” if there is not an agent present capable of intentional action (whereas it would make perfect sense to say: “Warum hat die Billardkugel angefangen zu rollen?” [“Why did the billiard ball start rolling?”]). 2 Collingwood would say that history is made up from human actions altogether. For the sake of argument I  only make here the weaker claim that human actions explanations are a subset of historical explanations. 3 It is perhaps worth noting that the “philosophy of historiography” is that part of philosophy of history that deals with the aspects of description, knowledge, explanation, and narration of the past, and how historians choose their subject (analytic philosophy of history), while “speculative” or “substantial philosophy of history” deals directly with history (e.g., whether history is contingent or necessary, whether it is directed towards a goal or cyclical). 4 It is worth mentioning for it is still sometimes confused, that this “teleologism” in philosophy of historiography is not the position of “teleologism” in (substantial, material) philosophy of history, i.e. the view that history itself has a goal or aim towards it progresses. 5 Aristotle’s view about whether human actions are to be explained causally or teleologically is not totally clear. In NE 1139a32ff he seems to favor a causalist view, but he always regards teleological explanations as appropriate for human actions (e.g. Met 1013a33ff). Hacker (2010: 183) argues that Aristotle’s distinction between four ways of the use of αἰτίov is in fact a distinction between four different ways of explanation (cf. also Glock 2014: 16). 6 Bacon, Novum Organon (2000: book II, aph. 2); Descartes Meditationes (2017: IV, 6) and Principia Philosophiae (1983: 1. I 28); Hobbes, De corpore (2012: ch. IX, X); Spinoza, Ethica (1996: p. I, Prop XXXVI, Append). 7 E.g. Vico (2000: § 1108) or Kant (1986: 22, 26f.). The idea of “autonomous processes” is still sometimes prominent in modern historiography (cf. Meier 1978). Sometimes autonomous historical processes are thought of as consisting of feedback mechanisms (cf. Hoyningen-Huene 1983). I do not wish to imply that all these authors were adherents of causalism—e.g. Vico

Introduction  37 is arguably the first philosopher who sees a difference in principle between our knowledge of natural phenomena and knowledge of human affairs (Vico 2000: § 330). 8 Weber’s position seems to be less clear though: On the one hand he emphasized the character of sociology as “verstehende Soziologie”, on the other he speaks as if understanding needs supplementation by statistic-causal analysis. Cf. Winch 1958: ch. 4.3f.; Rosenberg 2008: 36f. 9 Materialism in philosophy of mind was not invented by logical positivists though, but already present as a minor position in historical thinkers like Julien Offray de La Mettrie. 10 Serious objections of this kind are that explanations in general are always context dependent, such that there is not simply one logical form of whyexplanations (Passmore 1962). Also, not all DN-relations are causal relations, for Hempel’s model does not differentiate between evidence for event E and causes for the occurrence of E (“problem of explanatory asymmetries”). There is also the “problem of explanatory irrelevancies”, according to which there simply might be laws or regularities which simply are not relevant for the occurrence of the explanandum in question. 11 Causalist Theories of action were proposed by e.g. Alston (1967), Brandt and Kim (1963), MacIntyre (1966), Armstrong (1968), Goldman (1970: ch. III.5III.7), Mele (2003), Smith (1994). 12 Glock (2014: 18) holds that Wittgensteinians owe an explanation of why causalism rose again like phoenix from the ashes. He argues that it is not so much that important lessons from Wittgenstein have been forgotten in recent analytic philosophy, as Wittgensteinians “lament”, but due to the good plausibility of causalism. Davidson even is not that far from Wittgenstein in many respects (24, 27). But especially when Davidson later stresses that “beliefs and intentions are not little entities lodged in the brain” (Davidson 1999: 654), it becomes questionable how this is reconcilable with his conviction that beliefs, desires, and intentions are states that are causally effective. According to Glock, of the battery of anti-causal objections against causalism only one seems valid or irreconcilable with Davidson. However, I think that Davidson cannot be reconciled with the teleologist’s position. Davidson was a staunch causalist, because he even stuck to causalism when faced with the objection of deviant causal chains [cf. Mele, in this volume]. Glock seems to me to reach his result only because he does not consider further one of von Wright’s points, that (1) actions are not events, (2) leaves the objection of deviant causal chains out entirely and (3) by taking Davidson’s replies to anti-causalist arguments like the logical connection argument as compelling. 13 This debate can be found translated into German in Horn and Löhrer (2010). 14 “You may observe how someone in a morose mood, gently massaging his forehead and moaning, finally leaves his paperwork to go outside, where he takes deep breaths and moves about gymnastically. Someone else goes for a walk with a copy of Trees and Flowers of Britain under his arm, and stops frequently to scrutinize plants and compare his observations with entries in the book. In both cases the reason why the person went for a walk is evident” (Schroeder 2001: 153). 15 For a detailed discussion of that argument cf. Glock &Schmidt in this volume. 16 cf. Sandis 2009: 2; Horn and Löhrer 2010: 7; D’oro and Sandis 2013: 28–33. 17 The analogy between particles and a gas on the one hand and individual agents and the market is still to be found in more recent writings on sociological theory (Stinchcombe) (cf. Hedström and Swedberg 1998: 7). 18 To indicate some potential problems of this debate: It should be noted that bodily movements cannot be said to be in accordance with an aim

38  Gunnar Schumann or intention of an agent at all. It seems to be a common misunderstanding between Wilson, Mele, and Sehon that bodily movements and actions are identical. Schumann’s paper attempts (among other things) to cast some doubts on that assumption. Actions and their physical realizations are conceptually different and it might be that therefore the explanation of an action is not identical to the explanation of a bodily motion as a physical happening. An action’s explanation needs not even have the role to explain how the bodily movements of the agent were produced when performing the action, as Mele would have it. Only when one assumes that the bodily motions of an acting agent need to be explained by his volitional states, Mele’s causalist view seems attractive. Also, Mele presupposes that intentional actions are preceded by something that could properly be called “trying to act”. Admittedly, when an agent fails to perform an action when he was hindered or could not complete the action it is often appropriate to say that he (at least) tried. But in normal cases it would not make sense to speak of trying when no difficulties in performing the action occur. The ascription of trying also is not based on an inner experience or mental happening, but only makes sense when based on a publicly observable behavior of an agent—and it seems to be precisely this misconception to which the participants of this quarrel stick. At last, when Mele adapts his example such that the intentions of Norm are reliably executed by “freak chains” (such as by Martians) it could still be said, that Norm acts intentionally—just like it would have been Norm’s move in a chess game if he would have asked another person to put a piece for him. (Of course there are further misunderstandings that come with this, namely that the Martians can simply “read off” the aims and intentions from Norm’s mind by scientific devices. But reading the mind of others is neither a (neuro-) scientific nor metaphysical process. Mele implicitly thinks of aims and intentions as neurophysiological or mental entities or events with semantical contents—but this is misconceived.) 19 It is interesting to note here, that Hempel’s (1965) reply to Dray’s account of historical explanations consisted in pointing out that rational explanations should in fact be understood as “broadly dispositional” (in contrast to Dray) and that Churchland, Patzig, and the normic law approach (Schurz, Scholz) more or less agree to that.

References Alston, W. (1967). Wants, Actions, and Causal Explanations. In: H. Castaneda, ed., Minds, Intentionality, and Perception, 1st ed. Detroit: Wayne State University Press, pp. 301–341. Anscombe, G. (1957). Intention, 2nd ed. Oxford: Blackwell, 1963. Armstrong, D. (1968). A Materialist Theory of Mind. London: Routledge. Bacon, F. (2000). The New Organon: Or True Directions Concerning the Interpretation of Nature. Cambridge: Cambridge University Press. Brandt, R. and Kim, J. (1963). Wants as Explanations of Actions. The Journal of Philosophy, 60(15), pp. 425–435. Bratman, M. (1999). Shared Intention. In: M. Bratman, ed., Faces of Intention: Selected Essays on Intention and Agency, 1st ed. Cambridge: Cambridge University Press, pp. 109–129. Carnap, R. (1932/33). Psychologie in Physikalischer Sprache. Erkenntnis, 3(1), pp. 107–142.

Introduction  39 Castellani, F. (2007). Causation and Teleological Explanation of Action. In: F. Castellani and J. Quitterer, eds., Agency and Causation in the Human Sciences, 1st ed. Paderborn: Mentis, pp. 187–196. Collingwood, R. G. (1946). In: Jan van der Dussen, ed., The Idea of History, rev. ed.. Oxford: Oxford University Press, 1993. Collingwood, R. G. (1998). An Essay on Metaphysics, 2nd ed. Oxford: Oxford University Press. D’oro, G. and Sandis, C., eds. (2013). Reasons and Causes: Causalism and Anti-Causalism in the Philosophy of Action, 1st ed. Basingstoke: Palgrave Macmillan. Danto, A. (1965). Analytical Philosophy of History. Cambridge: Cambridge University Press. Davidson, D. (1963). Actions, Reasons and Causes. Journal of Philosophy, 60(23), pp. 685–700. Davidson, D. (1999). Replies to Critics. In: L. E. Hahn, ed., The Philosophy of Donald Davidson, Library of Living Philosophers, Vol. XXVII. Chicago: Open Court. Davidson, D. (2001). Freedom to Act. In: D. Davidson, ed., Essays on Actions and Events, 2nd ed. Oxford: Clarendon Press, pp. 63–81. Descartes, R. (1983). Principia philosophiae (Principles of Philosophy). Dordrecht: Reidel. Descartes, R. (2017). Meditations on First Philosophy. Cambridge: Cambridge University Press. Dilthey, W. (1973). Der Aufbau der geschichtlichen Welt in den Geisteswissenschaften. In: W. Dilthey, ed., Gesammelte Schriften, Vol. VII, 6th ed. Leipzig: Teubner, pp. 79–188. Donagan, A. (1966). The Popper-Hempel Theory reconsidered. In: W. Dray, ed., Philosophical Analysis and History, New York: Greenwood Press, pp. 127–159. Dray, W. (1957). Laws and Explanation in History. Oxford: Oxford University Press. Droysen, J. (1977). Grundriss der Historik. Die letzte Druckfassung (1882). In: P. Leyh, ed., Historik. Historisch-kritische Ausgabe, 1st ed. Stuttgart-Bad Cannstatt: frommann-holzboog, pp. 413–488. Gerber, D. (2012). Analytische Metaphysik der Geschichte: Handlungen, Geschichten und ihre Erklärung. Frankfurt: Suhrkamp. Glock, H-J. (2014). Reasons for Action: Wittgensteinian and Davidsonian Perspectives in Historical and Meta-Philosophical Context. Nordic Wittgenstein Review, 3(1), pp. 7–46. Goldman, A. (1970). A Theory of Human Action. Princeton: Princeton University Press. Hacker, P. M. S. (1996). Wittgenstein: Mind and Will. Analytical Commentary on the Philosophical Investigations. Vol. 4, Part 1. Oxford: Wiley-Blackwell. Hacker, P. M. S. (2010). Human Nature: The Categorical Framework. Oxford: Wiley-Blackwell. Hedström, P. and Swedberg, R. (1998). Social Mechanisms. An Introductory Essay. In: P. Hedström and R. Swedberg, eds., Social Mechanisms. An Analytical Approach to Social Theory, 1st ed. Cambridge: Cambridge University Press, pp. 1–31.

40  Gunnar Schumann Hempel, C. (1942). The Function of General Laws in History. In: M. Martin and L. McIntyre, eds., Readings in the Philosophy of Social Science, 1st ed. Cambridge: MIT Press, 1994, pp. 43–53. Hempel, C. (1965). Aspects of Scientific Explanation. In: C. Hempel, ed., Aspects of Scientific Explanation and Other Essays in the Philosophy of Science, 1st ed. New York: The Free Press, pp. 331–496. Hobbes, T. (2012). The Metaphysical System of Hobbes in Twelve Chapters from Elements of Philosophy Concerning Body. London: Forgotten Books. Horn, G. and Löhrer, C., eds. (2010). Gründe und Zwecke. Texte zur aktuellen Handlungstheorie. Frankfurt: Suhrkamp. Hoyningen-Huene, P. (1983). Autonome historische Prozesse: kybernetisch betrachtet. Geschichte und Gesellschaft, 9(1), pp. 119–123. Hume, D. (2008). Enquiry into Human Understanding. Oxford: Oxford University Press. Kant, I. (1986). Schriften zur Geschichtsphilosophie. Ditzingen: Reclam. Kim, J. (1973). Review of ‘Explanation and Understanding’. The Philosophical Review, 82(3), pp. 380–388. Kim, J. (1993a). Can supervenience and ‘non-strict laws’ save anomalous monism? In: J. Heil and A. Mele, eds., Mental Causation, 1st ed. Oxford: Oxford University Press, pp. 19–26. Kim, J. (1993b). The Non-Reductivist’s Troubles with Mental Causation. In: J. Heil and A. Mele, eds., Mental Causation, 1st ed. Oxford: Oxford University Press, pp. 189–210. Kincaid, H. (2009). Causation in Social Sciences. In: H. Beebee, C. Hitchcock and P. Menzies, eds., The Oxford Handbook of Causation, 1st ed. Oxford: Oxford University Press, pp. 726–743. Kocka, J. (1986). Sozialgeschichte. Begriff—Entwicklung—Probleme. Göttingen: Vandenhoeck & Ruprecht. Little, D. (2007). Philosophy of History. In: The Stanford Encyclopedia of Philosophy [online] Summer, 2017 ed. Available at: https://plato.stanford.edu/ archives/sum2017/entries/history/ [Accessed Feb 28, 2019]. Little, D. (2010). Action in History and Social Science. In: T. O’Connor and C. Sandis, eds., A Companion to the Philosophy of Action, 1st ed. Oxford: WileyBlackwell, pp. 401–409. Macdonald, G. and Macdonald, C. (2008). Explanation in Historiography. In: A. Tucker, ed., A Companion to the Philosophy of History and Historiography, 1st ed. Oxford: Wiley-Blackwell, pp. 131–141. MacIntyre, A. (1966). The Antecedents of Action. In: B. Williams and A. Montefiore, eds., British Analytic Philosophy, 1st ed. London: Routledge and Kegan Paul, pp. 205–225. Mackie, J. L. (1965). Causes and Conditions. American Philosophical Quarterly, 2(4), pp. 245–264. Meier, C. (1978). Fragen und Thesen zu einer Theorie historischer Prozesse. In: C. Meier and‎ K-G. Faber, eds., Theorie der Geschichte II. Historische Prozesse. Beiträge zur Historik, 1st ed. München: dtv, pp. 11–68. Melden, A. (1961). Free Action. London: Routledge and Kegan Paul. Mele, A. (2000). Goal-Directed Action: Teleological Explanations, Causal Theories, and Deviance. Philosophical Perspectives, 14, pp. 279–300. Mele, A. (2003). Motivation and Agency. Oxford: Oxford University Press.

Introduction  41 Mill, J. (2002). A System of Logic: Ratiocinative and Inductive. Honolulu: University Press of the Pacific. Mink, L. (1998). History and Fiction as Modes of Comprehension. New Literary History, 1(3), pp. 541–558. Neurath, O. (1931). Soziologie im Physikalismus. Erkenntnis, 2(1), pp. 393–431. Oppenheim, P. and Putnam, H. (1958). The Unity of Science as a Working Hypothesis. Minnesota Studies in the Philosophy of Science, 2, pp. 3–36. Passmore, J. (1962). Explanation in Everyday Life, in Science, and in History. History and Theory, 2(2), pp. 105–123. Popper, K. (1935). Logik der Forschung. Vienna: Springer. Rickert, H. (1986). Kulturwissenschaft und Naturwissenschaft. Stuttgart: Reclam. Risjord, M. (2014). Philosophy of Social Science: A Contemporary Introduction. London: Routledge. Rosenberg, A. (2008). Philosophy of Social Science, 3rd ed. Boulder: Westview Press. Rundle, B. (1997). Mind in Action. Oxford: Oxford University Press. Sandis, C., ed. (2009). New Essays on the Explanation of Action. Basingstoke: Palgrave Macmillan. Scholz, O. (2008). Erkenntnis der Geschichte—eine Skizze. In: A. Frings and J. Marx, eds., Erzählen, Erklären, Verstehen. Beiträge zur Wissenschaftstheorie und Methodologie der Historischen Kulturwissenschaften, 1st ed. Berlin: de Gruyter, pp. 111–128. Scholz, O. (2014). Philosophy of History: Metaphysics and Epistemology. In: M. Kaiser, O. Scholz, D. Plenge and A. Hüttemann, eds., Explanations in the Special Sciences: The Case of Biology and History, 1st ed. Dordrecht: Springer, pp. 245–253. Schroeder, S. (2001). Are Reasons Causes? A Wittgensteinian Response to Davidson. In: S. Schroeder, ed., Wittgenstein and Contemporary Philosophy of Mind, 1st ed. Basingstoke: Palgrave Macmillan, pp. 150–170. Schueler, G. (2005). Reasons and Purposes: Human Rationality and The Teleological Explanation of Action. Oxford: Clarendon Press/Oxford University Press. Schumann, G. (2017). Von Wright’s Theory of Action and the ‘Logical Connection Argument’. In: I. Niiniluoto and T. Wallgren, eds., On the Human Condition: Philosophical Essays in Honour of the Centennial Anniversary of Georg Henrik von Wright, 1st ed. Helsinki: Philosophical Society of Finland, pp. 325–338. Schurz, G. (2011). Erklären und Verstehen: Tradition, Transformation und Aktualität einer klassischen Kontroverse. In: F. Jaeger and J. Straub, eds., Handbuch der Kulturwissenschaften, 1st ed. Stuttgart: J. B. Metzler, pp. 156–174. Searle, J. (1990). Collective Intentions and Actions. In: P. Cohen, J. Morgan and M. Pollack, eds., Intentions in Communication, Cambridge: MIT Press, pp. 401–415. Sehon, S. (1994). Teleology and the Nature of Mental States. American Philosophical Quarterly, 31(1), pp. 63–72. Sehon, S. (2005). Teleological Realism: Mind, Agency, and Explanation. Cambridge: MIT Press. Sehon, S. (2016). Free Will and Action Explanation: A Non-Causal, Compatibilist Account. Oxford: Oxford University Press. Smart, J. (1959). Sensations and Brain Processes. Philosophical Review, 68(2), pp. 141–156.

42  Gunnar Schumann Smith, M. (1994). The Moral Problem. Oxford: Blackwell Publishers. Spinoza, B. (1996). Ethics. London: Penguin Classics. Stegmüller, W. (1979). Walther von der Vogelweides Lied von der Traumliebe und Quasar 3 C 273. In: W. Stegmüller, ed., Rationale Rekonstruktion von Wissenschaft und ihrem Wandel. Stuttgart: Reclam, pp. 27–86. Stoutland, F. (1970). The Logical Connection Argument. American Philosophical Quarterly. Monograph Series, 4, pp. 117–129. Stoutland, F. (1976). The Causal Theory of Action. In: J. Manninen and R. Tuomela, eds., Essays on Explanation and Understanding. Studies in the Formation of Humanities and Social Sciences, 1st ed. Dordrecht: Reidel, pp. 271–304. Stueber, K. (2012). Understanding versus Explanation: How to Think about the Distinction between the Human and the Natural Sciences. Inquiry, 55(1), pp. 17–32. Tucker, A. (2008). Causation in Historiography. In: A. Tucker, ed., A Companion to the Philosophy of History and Historiography, 1st ed. Chichester: WileyBlackwell, pp. 98–108. Vico, G. (2000). New Science: Principles of the New Science Concerning the Common Nature of Nations. London: Penguin Classics. Von Wright, G. (1971). Explanation and Understanding. London: Ithaca. Von Wright, G. (1985). Of Human Freedom. In: S. McMurrin, ed., The Tanner Lectures on Human Values. Vol. VI, 1st ed. Salt Lake City: University of Utah Press, pp. 107–170. Wilson, G. (1989). The Intentionality of Human Action. Stanford: Stanford University Press. Winch, P. (1958). The Idea of a Social Science and Its Relation to Philosophy, 3rd ed. London and New York: Routledge. Windelband, W. (1894). Geschichte und Naturwissenschaft. Rede zum Antritt des Rectorats der Kaiser-Wilhelms-Universität Strassburg, geh. am 1. Mai 1894. Strassburg: Heitz. Wittgenstein, L. (1958). The Blue and Brown Books. Oxford: Blackwell. Wittgenstein, L. (1984). Tractatus logico-philosophicus. Tagebücher 1914–1916. Philosophische Untersuchungen. Werkausgabe. Vol. 1. Frankfurt: Suhrkamp. Wittgenstein, L. (2009). Philosophical Investigations, 4th ed. Oxford: WileyBlackwell.

Part I

Causal vs. Teleological Explanation of Action

2 Causalism On Action Explanation and Causal Deviance Alfred R. Mele

In the philosophy of action, there are causal theories about what it is for something to be an action, about what it is to explain why someone did something, A, A being an intentional action, and about what it is for an action to be intentional. Here I discuss the first two issues, focusing on some threads in Scott R. Sehon’s discussion of them in his 2016 book.

1.  Action Explanation: Some Background Causalism about action explanation is usefully situated within a broader context. David Lewis contends that “to explain an event is to provide some information about its causal history” (1986: 217). The following more modest thesis is entailed by Lewis’s thesis: (C1) Nothing is an explanation of an event unless it provides some information about the event’s causal history. If C1 is true, then if all intentional actions are events, the following thesis also is true: (C2) Nothing is an explanation of an intentional action unless it provides some information about that action’s causal history. A  theorist who holds that there are both adequate and inadequate explanations may, in principle, assert both that C2 is false and that C2 would be true if “explanation” were modified by “adequate”. To address this issue, I reformulate C2 as follows: (C3) Nothing is an adequate explanation of an intentional action unless it provides some information about that action’s causal history. C3 is a statement of causalism about action explanation, and it is a consequence of Lewis’s thesis about event explanation in general, on the assumption that all actions are events. Al asks Ann why Norm climbed a ladder just now, and Ann replies, “to get his hat”. (Norm left his hat on his roof while making some repairs up there.) Suppose for a moment that causalism is true. If Ann’s reply is an adequate explanation of Norm’s climbing the ladder, what information does it provide about that action’s causal history? A  variety of different answers are open to causalists, depending on their views about such things as mental causation and the nature of reasons for action.

46  Alfred R. Mele 1. Some causalists say that if Ann’s reply is an adequate explanation, a proper account of the explanation attributes to Norm a reason to climb the ladder that is constituted by a desire to get his hat and a belief to the effect that climbing the ladder is conducive to getting it and this desire-belief pair is among the causes of his climbing the ladder. 2. What about causalists who reject the thesis that reasons are states of mind? They may say that if Ann’s reply is an adequate explanation, a proper account of it attributes the desire-belief pair just mentioned to Norm, which pair is among the causes of the action at issue. 3. Some causalists worry that mental states have no causal clout. They may say that if Ann’s reply is an adequate explanation, a proper account of it attributes the desire-belief pair just mentioned to Norm and the neural realizers of that pair are among the causes of the action at issue. 4. What about causalists who are skeptical about the existence of states of mind? They may say that if Ann’s reply is an adequate explanation, a proper account of the explanation asserts that Norm wanted to get his hat and believed that climbing the ladder is conducive to getting it and the fact that he wanted to do this and believed what he did about climbing the ladder is among the causes of the action at issue. Options for causalists of the kinds just described are discussed in Mele (2013: 164–168). Here I simply articulate these options and forge ahead. Readers who would like more details may consult Mele (2013). The main point to notice now is causalism’s flexibility. Although I  have advocated causalism about action explanation in various books and articles, I have not offered a direct argument for it. My strategy has been to review leading attempts to answer a challenge to anti-causalists first posed by Donald Davidson in 1963 (reprinted in Davidson 1980), to argue that these attempts fail, and to draw some morals from their failure.1 The challenge, in effect, is to produce nontrivial conceptually sufficient conditions for the truth of a sentence like the following that have no causal component: Norm climbed the ladder to get his hat.

2.  Norm and the Martians In his 1989 book, George Wilson offers a detailed anti-causalist response to Davidson’s challenge. I have argued that the conditions he offers fail to be sufficient for the truth of a sentence like “Norm climbed the ladder to get his hat” because they are not even sufficient for acting (Mele 2003: 45–50). That is, if I am right, the conditions can be satisfied by someone who is not acting. Scott R. Sehon disagrees (2016: 56–63). In his

Causalism  47 latest installment on this issue, Sehon contends that the main character in an attempted counterexample of mine to Wilson’s conditions is acting (2016: 57). So I  can forgo summarizing Wilson’s proposed conditions (for summaries and discussion, see Mele 1992, 2003, 2010, 2017) and focus on the disagreement between Sehon and me. Why should this disagreement matter to anyone but Sehon and me? Well, if Sehon is right, Wilson has met Davidson’s challenge and has falsified causalism about action explanation. And if I am right about Wilson’s conditions, anti-causalists who find what I say about this persuasive may be motivated to try to succeed where Wilson failed. Consider the following story (from Mele 2010: 185–186; also see Mele 1992: 248–249, 2003: 45–46). Norm has learned that, on rare occasions, after he embarks on a routine activity (for example, tying his shoes, climbing a ladder), Martians take control of his body and initiate and sustain the next several movements in the chain while making it seem to him that he is acting normally. He is unsure how they do this, but he has excellent reason to believe that they are even more skilled at this than he is at moving his own body, as, in fact, they are. (The Martians have given Norm numerous demonstrations with other people.) The Martians have made a thorough study of Norm’s patterns of peripheral bodily motion when he engages in various routine activities. Their aim was to make it seem to him that he is acting while preventing him from even trying to act by selectively shutting down portions of his brain. To move his body, they zap him in the belly with M-rays that control the relevant muscles and joints. When they intervene, they wait for Norm to begin a routine activity, read his mind to make sure that he plans to do what they think he is doing (for example, tie his shoes or climb to the top of a ladder), and then zap him for a while—unless the mind-reading team sees him abandon or modify his plan. When the team notices something of this sort, the Martians stop interfering and control immediately reverts to Norm. A while ago, Norm started climbing a ladder to fetch his hat. After he climbed a few rungs, the Martians took over. Although they controlled Norm’s next several movements while preventing him from trying to do anything, they would have relinquished control to him if his plan had changed (for example, in light of a belief that the location of his hat had changed). That is where this story ends. I went on to call attention to Wilson’s claim that sentiently directing a bodily movement that one performs entails exercising one’s “mechanisms of [. . .] bodily control” in performing that movement (1989: 146). And I said the following: Norm did not exercise these mechanisms in his performance of the movements at issue. Indeed, he did not make even a minimal effort to perform these movements; owing to the Martian intervention, he made no effort at all—that is, did not try—to do anything at the time.

48  Alfred R. Mele And it is a platitude that one who did not try to do anything at all during a time t did not sentiently direct his bodily motions during t. (2010: 186) Sehon lodges the following complaint: When Mele stipulates that the Martians prevent Norm from even trying to act, Mele thereby attempts to stipulate, as part of the example, that Norm is not acting. Since the nature of action is the very point at issue, this attempted stipulation begs the question. We can say that the Martians control Norm’s body, but, unless we simply presuppose the falsity of the teleological view from the outset, Mele cannot simply stipulate that the Martians keep Norm from trying to act. (2016: 58) When I wrote Mele (2003) I naively supposed that it would be granted on all sides that Norm is not acting in my story (that is, after the Martians take over), even though he satisfies Wilson’s proposed conditions. I naively supposed that all sides would see my story as a counterexample to Wilson’s proposed sufficient conditions. I  certainly did not suppose that the story would convert all anti-causalist readers to causalism, and I predicted that some anti-causalist readers would try to succeed where Wilson failed—that is, that they would try to produce a non-trivial, informative, anti-causalist set of sufficient conditions for A-ing in order to B that is not falsified by my story. Wilson’s “teleological view” is only one such anti-causalist view; so I did not view myself as closing the door on all anti-causalist teleological views of action explanation with this one example. And I did not see the story as begging the question against Wilson—any more than Edmund Gettier’s attempted counterexamples (1963) beg the question against previous proponents of the thesis that having a justified true belief that p is conceptually sufficient for knowing that p. While I am at it, I note that the nature of knowledge is, in Sehon’s words, “the very point at issue” in Gettier’s article. Indeed, when philosophers are in the business of presenting attempted counterexamples to proposed conceptually sufficient conditions for X, one expects the nature of X to be at issue. Sehon reports that as long as the Martians are completely reliable, he is willing to assert that “Norm is acting when he climbs the ladder under the causal influence of the M-rays” (2016: 57). So he is in a position to contend that my story is self-contradictory. He can contend that some of the details of the story are conceptually sufficient for Norm’s continuing to climb the ladder after the Martians take over and hence for his continuing to try to climb it; and, of course, Norm’s continuing to try to climb it is inconsistent with another detail of the story, namely, that the Martians prevented him “from trying to do anything” (Mele 2010: 185).

Causalism  49 How does Sehon motivate the surprising claim that Norm is acting in my story? He returns (2016: 58) to an idea that he presented in an earlier book (2005: 168). In the earlier book, Sehon attempts to support the idea that it “is not obvious that Norm’s behavior fails to be an action” (2005: 168) by sketching a version of my story in which Norm is about to shoot someone when “the Martians take over his body and make it carry out the dirty deed”. Sehon asks, “Does this completely absolve Norm of responsibility for shooting his professor?” He reports that it is not obvious to him how to answer this question. “Accordingly”, he writes, “even in the routine case of going up the ladder, I take it not to be obvious that Norm failed to act”. Here is part of my reply in Mele 2010: Things are not nearly as murky as Sehon thinks. I start with an obvious point. Norm is responsible for shooting his professor only if Norm shoots his professor. This is an instance of the truism that a person P is responsible for doing A only if P did A. But notice that P may have some responsibility for the shooting of X even if, because P did not shoot X, P has no responsibility for shooting X. For example, if P hires—or forces—someone to shoot X, then, other things being equal, P has some responsibility for the shooting of X. In Sehon’s story, the Martians “make it seem to Norm as if he is acting, and if Norm had changed his mind and decided to put the gun down [they] would have immediately relinquished control” to Norm (2005: 168). (Here Sehon is following my lead, of course.) So readers may be strongly inclined to see Norm as responsible for more than, in Sehon’s words, merely “having a plan to commit murder.” In light of the distinction I have drawn between P’s having some responsibility for shooting X and P’s having some responsibility for the shooting of X, those who, like me, are convinced that Norm did not shoot the professor can maintain that, even so, he has some responsibility for the shooting of the professor. (2010: 188) Sehon attempts to “strengthen the case” for his judgment that Norm is acting in this case—and, specifically, his judgment that Norm shoots the professor—“by considering related examples” (2016: 59), including an example featuring occasionalism (59–60). I comment on occasionalism shortly. I want to call attention now to an example of mine from Mele (2010). It appears immediately after the passage last quoted from that article and reads as follows: Imagine that Sehon’s story remains basically the same, except that a Martian makes himself invisible and then, with his slim but powerful tentacle, pulls Norm’s paralyzed finger down on the trigger. (The Martian also makes it seem to Norm as though Norm is pulling the

50  Alfred R. Mele trigger, and “if Norm had changed his mind and decided to put the gun down, the [Martian] would have immediately relinquished control” to Norm.) Obviously, Norm did not pull the trigger. So Norm did not shoot the professor. That entails that Norm is not responsible for shooting the professor. Even so, he may have some responsibility for the shooting of the professor. After all, the Martian would not have pulled the trigger with Norm’s finger if Norm had not been bent upon shooting the professor, and the professor’s life would have been spared if Norm had changed his mind. Sehon’s uncertainty about whether Norm shot the professor seems to derive from his failure to distinguish having some responsibility for shooting the professor from having some responsibility for the shooting of the professor. (2010: 188) In this passage, I claim that it is obvious that Norm did not pull the trigger and therefore did not shoot the professor. But I thought that these things were obvious in the earlier story about Norm and the professor, and Sehon disagrees. Sehon does not respond to this “tentacle” version of the story in his 2016 book. Regarding the earlier story, Sehon writes: “I submit that the Martians’ intervention makes no difference at all in how we treat Norm. In this case, the alleged distinction—Norm being responsible for shooting the victim versus Norm being responsible for the shooting of the victim—is a distinction without a difference” (2016: 59). Would he say the same in the present case? For the reasons I identified, Norm may, as I put it, “have some responsibility for” the shooting of the professor. But if, as certainly seems to be the case in the present story, he did not shoot the professor, he is not responsible for shooting him. We definitely have what looks like a distinction with a difference here. We philosophers would like to persuade all of our philosophical opponents with our attempted counterexamples. But even if we believe that doing so is not in the cards, our attempted counterexamples can still do significant work. For one thing, they can persuade (almost) everyone else who reflects on them; and our (remaining) opponents may be a relatively tiny group. In Mele (2010), I  explain why my story about Norm and the Martians in Mele (2003) portrays the Martians as using M-rays rather than moving Norm with their invisible tentacles (2010: 188–189). That story, as I have mentioned, was designed to be a counterexample to Wilson’s proposed set of sufficient conditions for goal-directed action, and that set of conditions includes the person’s “performing movements”. Although Wilson is happy to say that a wire clutcher whose body is being jerked about by the wire’s electrical discharge is performing movements in his thin sense (1989: 49), I doubted (and still do) that he would count an isolated case of a man whose limb motions are caused—in the way Norm’s are in my “tentacle” story—by Martians pulling and pushing on his limbs as a case of a man’s performing movements with those limbs.

Causalism  51 But what about Sehon? Would he see a significant difference between Martians secretly pushing and pulling Norm from the outside and their secretly pushing and pulling Norm from the inside, as it were? I will not venture a guess. In yet another story about Norm and the Martians—a new one—the Martians move Norm by pulling on invisible cords that are attached to Norm with a marvelous Martian invention that enables them to mimic ordinary overt human actions with human bodies that they control. As in the original story, if Norm were to abandon or modify his plan, the Martians would immediately stop interfering. These mischievous Martians proudly view themselves as master puppeteers as they move Norm’s body up the ladder with their invisible cords, and they quietly chuckle at the people below who are impressed by how smoothly Norm moves up a ladder for a man his age. Would Sehon say that Norm is climbing the ladder—acting—in this story too? I mentioned that Sehon discusses occasionalism in an effort to “strengthen the case” for his judgment that Norm shoots the professor in his story. He writes: Occasionalism amounts to Norm and the Martians writ large: Instead of godlike Martians sometimes in control of one person’s body via magical M-rays, we would have an actual God always in control of everything via his omnipotent will. So if we were to conclude, with Mele, that Norm does not perform an action when the Martians are the causal power, then we should conclude that nobody has ever performed any action whatsoever if occasionalism is true. (2016: 60) Matters are not so simple. Suppose that we learn tomorrow that the things we have been calling cats are robots and that these things always have been robots (Putnam 1962: 660). According to one well-known view, these things are cats (Putnam 1962: 660). Suppose instead that we learn tomorrow that a few of the things we have been calling cats are robots—and that the rest are animals, just as we thought. According to the same well-known view, these few things are not cats (Putnam 1962: 660). Similarly, someone who judges, with me, that Norm does not act in my stories may also—without contradiction—judge that if we were to discover that occasionalism is true (or that our motions are always produced by M-rays), we would not have discovered that human beings do not act. Human action would be preserved in this person’s view, just as the existence of cats is preserved, in Putnam’s view, in the thought experiment in which we learn that the things we have been calling cats are—and always have been—robots. Sehon offers no argument for his final sentence in the passage last quoted. In that sentence, he simply ignores the popular view just described. He also ignores another reason to be skeptical of the claim

52  Alfred R. Mele at issue—the claim that “if we were to conclude, with Mele, that Norm does not perform an action when the Martians are the causal power, then we should conclude that nobody has ever performed any action whatsoever if occasionalism is true” (2016: 60). As I observed in Mele (2003), “a deviant causal connection between an x and a y is deviant relative to ‘normal’ causal routes from x-s to y-s” (2003: 53; more on this shortly). A causalist may argue that Norm is not acting owing to the deviant way in which his pertinent bodily motions are produced—a way that is highly abnormal in his world. But in an occasionalist world, there is nothing abnormal about God’s being centrally involved in the production of bodily motions. Thus, causalists have deviance-featuring grounds for concluding that Norm is not acting (after the Martians take over) that are absent in the case of an occasionalist world.2 I continue to see no good reason to believe that Norm is acting (after the Martians take over) in the stories rehearsed here. And I continue to regard my “ladder” case as a counterexample to Wilson’s proposed sufficient conditions for acting in pursuit of a goal.

3. Deviance I mentioned that, in addition to there being causal theories about action explanation, there are causal theories about what it is for something to be an action and about what it is for an action to be intentional. Paul Moser and I (Mele and Moser 1994) have offered an analysis of what it is for an action to be intentional. (We leave action itself unanalyzed in that article.) I have never offered an analysis of action nor of acting in pursuit of a particular goal. But I have argued (in the indirect way mentioned in the section 1) for causalism about action explanation and I have taken up some issues that a causalist in the business of constructing an analysis of action would need to address—for example, deviant causal chains (Mele 1992: 197–210, 243–247, 2003: 51–63). In his 2016 book, Sehon reacts to some ideas of mine about causal deviance. After some stage setting, I reply to his reaction. Deviant causal chains raise apparent difficulties for the project of providing causal analyses of action and intentional action. The alleged problem is that whatever causes are deemed necessary for a resultant event’s being an action or for an action’s being intentional, and whatever larger set of conditions is deemed sufficient for the event’s being an action or for the action’s being intentional, cases can be described in which, owing to a deviant causal connection between the favored causes (for example, events of intention acquisition) and a resultant event, that event is not an action or a pertinent resultant action is not done intentionally. The most common examples of deviance divide into two types.3 Cases of primary deviance raise a problem about a relatively direct connection between mental antecedents and resultant bodily motion. Cases of

Causalism  53 secondary deviance highlight behavioral consequences of intentional actions and the connection between these actions and their consequences. The following are representative instances of the two types of case: Primary: A climber might want to rid himself of the weight and danger of holding another man on a rope, and he might know that by loosening his hold on the rope he could rid himself of the weight and danger. This belief and want [or, one might suppose, his intention to let go of the rope] might so unnerve him as to cause him to loosen his hold [unintentionally]. (Davidson 1980: 79) Secondary: A man may try to kill someone by shooting at him. Suppose the killer misses his victim by a mile, but the shot stampedes a herd of wild pigs that trample the intended victim to death. (Davidson 1980: 78) The second case can be handled by some version of the popular suggestion that an action’s being intentional depends on its fitting the agent’s conception or representation of the manner in which it will be performed. Of course, how close the fit must be requires attention (see Mele and Moser 1994). Cases of primary deviance are more interesting. Some causalists who regard cases of primary deviance as putative counterexamples to a causal account of what it is for an action to be intentional dismiss them on the grounds that they are not cases of action at all (Brand 1984: 18; Thalberg 1984). If this diagnosis is correct, primary deviance raises a question for attempted causal analyses of action. Can causalists identify something of a causal nature in virtue of which it is false that the climber performed the action of loosening his grip on the rope? In section 2, I mentioned an observation about causal deviance from Mele (2003): “a deviant causal connection between an x and a y is deviant relative to ‘normal’ causal routes from x-s to y-s” (Mele 2003: 53). That observation immediately preceded the following one: “what counts as normal here is perspective-relative. From the point of view of physics, for example, there is nothing abnormal about [the climber’s case]. And, for beings of a particular kind, the normal route from intention to action may be best articulated partly in neurophysiological terms” (53). I pointed out that “one way around the problem posed by our neuroscientific ignorance is to design (in imagination, of course) an agent’s motor control system” (2003: 53). I wrote: Knowing the biological being’s design in that sphere, we have a partial basis for distinguishing causal chains associated with overt action from deviant motion-producing chains. If we can distinguish deviant

54  Alfred R. Mele from nondeviant causal chains in agents we design—that is, chains not appropriate to action from action-producing chains—then perhaps we would be able to do the same for normal human beings, if we were to know a lot more than we do about the human body (including the brain, of course). I proceeded to design beings of a certain kind—I called them “Promethean agents”, after the god to whom I assigned the designing task— and I  applied the neurophysiology and psychology that I  described to some cases of primary deviance (2003: 54–63). The discussion is too long and complicated to summarize here (see Mele 2005: 339–345 for a summary). Instead I will comment on some confusions in Sehon’s response to this material. Sehon asserts that it is “somewhat baffling” that my “analysis is officially limited to these mythical agents” (2016: 93). But, as I pointed out in my discussion of Promethean agents (2003: 53, 62), and as is clear in any case, I do not offer an analysis and instead offer a sufficient condition for acting in pursuit of a particular goal. The condition offered avoids the problem about primary deviance at issue there. Sehon proceeds to discuss what he refers to as “Mele’s analysis” (2016: 94). But if I had offered an analysis, it certainly would not have been what he saddles me with. For example, Sehon contends (2016: 96) that “on Mele’s view”, agents have discrete intentions for all of the tiny actions involved in typing a sentence: “What a huge number of executive attitudes flying by in the ten seconds or so it took me to type that short sentence!” However, I myself (in the same book) explicitly reject the claim that all actional parts of larger actions are associated with their own discrete intentions: “When I walk to work, for example, my individual steps are intentional actions, but there is no need to suppose that each step requires its own distinct intention. In a typical case, a single, more general intention—an intention to walk to work along my normal route—is intention enough” (Mele 2003: 205). I  appeal to “subsidiary” actions of this kind as one source of counterexamples to the Simple View—the thesis that, necessarily, an agent intentionally A-s only if he intends to A (see, e.g., Mele 1997: 242–243). Regarding an analysis that he invents for me, Sehon asserts that it “fails as a necessary condition for goal-direction” and reports his beliefs that “it also fails as a sufficient condition” and that “it is not even clear that the analysis will help with the case of Mele’s philosopher” (2016: 97). However, my treatment of primary deviance does not include an analysis of goal direction; nor does it put forward a necessary condition for goal direction. My aim is to provide non-trivial, informative sufficient conditions for acting in pursuit of a particular goal that deal with primary deviance. So Sehon’s assertion about sufficient conditions is still on the table.

Causalism  55 The “case of Mele’s philosopher” that Sehon mentions is from Mele (1992). Here it is: “A  philosopher intends to knock over his glass of water in order to distract his commentator. However, his intention so upsets him that his hand shakes uncontrollably, striking the glass and knocking it to the floor” (1992: 182). Can I explain why this philosopher does not perform a goal-directed action of knocking over his glass? Yes, and the explanation mirrors the explanation I offer in Mele (2003) of the climber’s not performing the goal-directed action of letting go of the rope. In Mele (2003), I distinguish among four versions of the climber’s story and offer an explanation of the pertinent event’s being a non-action in each case (59–60). Here, to save space and to avoid excessive repetition, I discuss a “spilled water” analogue only of the most challenging version of the climber’s story. I give my previously unnamed philosopher a name, Phil, and I assume that he is a Promethean agent. Phil’s acquiring the intention to distract his commentator by knocking over the water glass (intention N) results in appropriate motor signals being sent. Those signals reach their targeted muscles and joints, and Phil begins to move his hand toward the glass. This movement unnerves him, with the result that his hand shakes uncontrollably. And the continued presence of intention N causally sustains the nervousness that results in Phil’s temporary loss of control over the motions of his arm and hand. Does this Promethean agent perform the goal-directed action of knocking over the glass? The following answer parallels what I  had to say in Mele (2003: 60) about the analogous climber. I enclose overlapping material in quotation marks. “Although there was a time at which” Phil was moving his hand toward the glass—and doing this in pursuit of the goal of distracting his commentator—“he does not perform the action” of knocking over the glass. “Nor, a fortiori, does he perform [it] in pursuit of a goal. Given his construction as a Promethean agent, [Phil] is not capable of contributing to the continuation of an overt action of his once his relevant bodily motions are not being guided by an intention of his. This itself does not entail that, after becoming unnerved, [Phil] is no longer performing the action” of moving his hand or moving it toward the glass; “for, in some cases, direct ballistic continuations of motions that had been guided may be parts of actions, as in the finger-snapping example [that I reproduce shortly]. However, the nervousness-produced motions” of Phil’s arm and hand “are not direct ballistic continuations of his previous motions. After all, they are produced by the nervousness that his beginning to [move his arm] causes. Even if they are, in some sense, continuations of earlier motions, they are not direct ballistic continuations. So, in short, [Phil’s] performance of the action of [moving his hand toward the glass] is terminated too soon for him to have performed” the goal-directed action of knocking over the glass.

56  Alfred R. Mele A few pages earlier in Mele (2003), I  commented on direct ballistic continuations of motions. Attention to that comment will prove useful: Consider an ordinary case of snapping one’s fingers. An agent embarking on a conventional right-handed finger-snapping with his palm facing up presses his thumb and middle finger firmly together and then, maintaining the pressure, simultaneously slides the thumb to the right and the finger downward. After a certain point, the agent is no longer in control of the process. The finger whips downward, thumping into his palm next to his thumb, and the thumb slides off to the right. Here there is direct ballistic continuation of motions that had been guided. (2003: 57) For expository purposes, I had a god design the Promethean agents. The passage just quoted was followed by my report that this god, Prometheus, regards the ballistic motions I described as part of the agent’s action of snapping his fingers. The agent’s active contribution to his action in cases like this ends before the action is completed on this view of things. Direct ballistic continuations of motions in such cases are the currently uncontrolled unfolding of immediately preceding segments of the action; their directness precludes their having events extrinsic to the process as proximal initiators. This is why the motions that are proximally initiated by the nervousness that Phil’s initially moving his arm causes are not direct ballistic continuations of previous motions. Here is another example of direct ballistic continuation. When a person kicks a ball, the leg is “flung” by the quadriceps muscle, which ceases its activity before the leg stops moving and before the foot comes into contact with the ball (Sheridan 1984: 54). Once the leg is flung, its movement in the kicking is no longer guided. If, just after the process goes ballistic, someone were to move the leg with a jolt of electricity, the resulting motion would not be a direct ballistic continuation of the guided motions. My discussion of primary deviance in this section is motivated by Sehon’s remarks, quoted earlier, on sufficiency and on the “spilled water” example—that is, his claims that an analysis that he invents for me “fails as a sufficient condition” and that “it is not even clear that the analysis will help with the case of Mele’s philosopher” (2016: 97). What he says in support of these claims is puzzling. Regarding the “spilled water” example, he writes: “we might say that the hand twitch was a direct ballistic motion, so even if the philosopher’s intention dissolves (or simply fails to give further causal input), this would not falsify (ii)”, where (ii), a plank in an analysis Sehon constructs for me, reads as follows: the intention “continues to send causal signals without which the relevant bodily motion would cease, unless the bodily motion is a ‘direct ballistic

Causalism  57 continuation’ of a bodily motion initiated by the intention” (Sehon 2016: 95). However, the uncontrolled, nervousness-prompted twitch is not a “direct ballistic continuation” of guided motions, as I  use the quoted expression. Sehon is free to use the expression differently; but when he does so, he is not addressing my actual response to primary deviance. I find in Sehon’s discussion of my proposal about primary deviance no cause for worry about my actual proposal. I continue to believe that, as I put it in Mele (2003), “I have shown that, in the case of at least one kind of agent, causalists can handle the problem allegedly posed by representative instances of primary causal deviance for the project of providing a causal analysis of acting in pursuit of a particular goal” (2003: 63). This is not to say that I myself have sought to provide such an analysis; I have not. I offered, in the case of Promethean agents, a certain “mix of causal initiation, sustaining, and guiding” as a sufficient condition for acting in pursuit of a particular goal (62). I do not claim that this mix is necessary for goal-directed action in Promethean agents (or in us). For example, some goal-directed A-ings of Promethean agents (these A-ings being actions) are not initiated by intentions to A. A case in point is a particular ordinary step that a Promethean agent takes during an ordinary walk to work. There is much of interest in Sehon’s new book. Here I  have focused on just two topics of mutual interest: our continuing disagreement about Wilson’s proposed set of sufficient conditions for action in pursuit of a goal, and a proposal of mine about primary causal deviance. I  have argued that Sehon’s latest response to my critique of Wilson’s conditions is unpersuasive and that his reaction to my proposal about primary deviance misfires.4

Notes 1 My latest effort is Mele (2017: ch. 3). There I collect replies of mine to D’Oro (2007), Ginet (1990), Sehon (1994, 2005), Wallace (1999), and Wilson (1989). 2 Stephen Kearns and Marcela Herdova encouraged me to make this point. 3 For a third type, see Mele (1992: 207f.). 4 I am grateful to Marcela Herdova, Stephen Kearns, Gunnar Schumann, and Scott Sehon for comments on a draft of this article.

References Brand, M. (1984). Intending and Acting. Cambridge: MIT Press. Davidson, D. (1980). Actions, Reasons, and Causes. In: D. Davidson, ed., Essays on Actions and Events, 1st ed. Oxford: Clarendon Press, pp. 3–19. D’Oro, G. (2007). Two Dogmas of Contemporary Philosophy of Action. Journal of the Philosophy of History, 1(1), pp. 10–24. Gettier, E. (1963). Is Justified True Belief Knowledge? Analysis, 23(6), pp. 121–123. Ginet, C. (1990). On Action. Cambridge: Cambridge University Press.

58  Alfred R. Mele Lewis, D. (1986). Causal Explanation. In: D. Lewis, ed., Philosophical Papers. Vol. 2, 1st ed. Oxford: Oxford University Press, pp. 214–240. Mele, A. (1992). Springs of Action. New York: Oxford University Press. Mele, A. (1997). Agency and Mental Action. Philosophical Perspectives, 11, pp. 231–249. Mele, A. (2003). Motivation and Agency. New York: Oxford University Press. Mele, A. (2005). Action. In: F. Jackson and M. Smith, eds., Oxford Handbook of Contemporary Philosophy, 1st ed. Oxford: Oxford University Press, pp. 334–357. Mele, A. (2010). Teleological Explanations of Actions: Anticausalism vs. Causalism. In: J. Aguilar and A. Buckareff, eds., Causing Human Actions: New Perspectives on the Causal Theory of Action, 1st ed. Cambridge: MIT Press, pp. 183–198. Mele, A. (2013). Actions, Explanations, and Causes. In: G. D’Oro and C. Sandis, eds., Reasons and Causes: Causalism and Anti-Causalism in the Philosophy of Action, 1st ed. London: Palgrave Macmillan, pp. 160–174. Mele, A. (2017). Aspects of Agency. New York: Oxford University Press. Mele, A. and Moser, P. (1994). Intentional Action. Noûs, 28(1), pp. 39–68. Putnam, H. (1962). It Ain’t Necessarily So. Journal of Philosophy, 59(22), pp. 658–671. Sehon, S. (1994). Teleology and the Nature of Mental States. American Philosophical Quarterly, 31(1), pp. 63–72. Sehon, S. (2005). Teleological Realism: Mind, Agency, and Explanation. Cambridge: MIT Press. Sehon, S. (2016). Free Will and Action Explanation. Oxford: Oxford University Press. Sheridan, M. (1984). Planning and Controlling Simple Movements. In: M. Smyth and A. Wing, eds., The Psychology of Human Movement, 1st ed. London: Academic Press, pp. 47–82. Thalberg, I. (1984). Do our Intentions Cause our Intentional Actions? American Philosophical Quarterly, 21(3), pp. 249–260. Wallace, R. (1999). Three Conceptions of Rational Agency. Ethical Theory and Moral Practice, 2(3), pp. 217–242. Wilson, G. (1989). The Intentionality of Human Action. Stanford: Stanford University Press.

3 Why and How? Teleological and Causal Concepts in Action Explanation George F. Schueler

The idea that explanations of actions in terms of the agent’s reasons are causal explanations of a certain variety still has wide support among philosophers. This chapter will reexamine that view.

1.  Intentional Actions Suppose you notice there is a nail in my office wall that wasn’t there previously. “Why is that there?” you ask. “Well”, I say, “the other day I brought a hammer and nail from home. I then held that nail at about a forty-five-degree angle against the wall and hit it several times with my hammer”. There is a question to which this answer might be the correct one, but probably that is not the question you were asking. I told you how I got that nail into my wall. Probably what you wanted to know was why I did it. You wanted to know the purpose or point of my action. Things done intentionally are done for reasons.1 So there is always a point or purpose to doing them. Describing that purpose is a common way of saying what the action is. So “It’s there to hang my new portrait of the Dean” would answer your question by saying why I put that nail in my wall. Whenever I  do something for a reason there will be some purpose in doing it and of course there will always be some way in which I pursue that purpose (or try to). To perform an intentional action is to pursue some goal or purpose in some specific way.2 So the ambiguity just noted in explaining my action will always be there. It is the difference between asking for the purpose or point of whatever I am doing and asking how I am pursuing that purpose. I will mark this difference by saying that the former answers a “Why?” question and the later answers a “How?” question. What is the relation between these questions? They seem logically independent in the sense that answering one question about some action doesn’t determine the answer to the other. Consider our example. Explaining in detail how I held the hammer, how the muscles in my arm contracted to produce the movement of the hammer against the nail,

60  George F. Schueler even how the neuronal interactions in my brain coordinated looking at the nail and swinging the hammer against it, still tells us nothing about why I am putting a nail in my office wall. Likewise, finding the purpose I am pursuing doesn’t tell us anything about how I am trying to achieve that purpose. Suppose you know only that I  am “hanging my picture of the Dean”, say because that is what I tell you on the phone. You still don’t know that I am pounding a nail into my wall. That might be the best way to do it but I might not know that. I might think that a blob of chewing gum would work fine. Still, there is an important relation between these two questions. Both can be about the same thing, some action someone is performing. That is because all actions done for reasons have a purpose and all are done “in some way”. But purposes are logically distinct from how they are pursued. Even if, unknown to me, my method of pursuing some goal has no chance of succeeding, that doesn’t mean I am not actually pursuing that goal, trying to achieve it. At the same time, when purposes are pursued, they must of course always be pursued in some way or other. So both sorts of question will be appropriate about any intentional action, just in virtue of the fact that it is an intentional action. This is not an accidental feature of intentional actions. Something to which both these questions were not applicable would not be an intentional action at all. This is clear if we change the nail example slightly. Suppose I  hammered that nail into my wall solely as a result of hypnotic suggestion. In this case it could be that I “did exactly the same thing” in the sense of the “How?” question. It could even be that every muscle contraction, every nerve impulse that controlled how I held the hammer and swung my arm, was exactly the same as in the original example. But if I was merely responding to a hypnotic suggestion, the “Why?” question does not apply, or at least there is no answer to it. I had no purpose in doing what I was doing. And if there was no purpose I was pursuing then of course there was no way I  was pursuing it either and so the “How?” question would have no application either. So in such a case I was not performing an intentional action at all. Something similar would be true if I have been given some neurotoxin that interferes with my ability to control by own bodily movements, so that the muscle contractions in the arm holding the hammer are caused by the drug. In such a case there is a sense in which I am still “doing something”, waiving the hammer around perhaps. But if I am unable to control even my own muscle contractions I am not performing an intentional action. And that would be true even if I wanted or intended to drive that nail into the wall and even if by chance my flailing resulted in my striking the nail and driving it into the wall. My muscle contractions were caused by that drug, not me, and so I was not performing an intentional action. These are cases where we might say I “was a participant in” a set of events which might even have been the same as the ones I would have set

Why and How?  61 in motion had I been pursuing a goal. They would have thus been “how” I would have pursued that goal, had I pursued it. My bodily movements might have been exactly the same as they would have been had I been pursuing the goal of hanging that picture, even though I was performing no intentional action at all and so neither the “Why?” nor the “How?” question apply. It is also easy to think of situations where I have a goal or purpose but do nothing to pursue it. That happens when I intend to do something in the future but so far have done nothing to further my intention (i.e. I have what are sometimes called a “prior intention”). I might intend to hang my picture of the Dean next Thursday, say, even if as yet I have done nothing whatever to further this plan. In that case I certainly seem to have a goal, hanging that picture next Thursday, but of course there is as yet no action about which we could ask why or how it was done. So on the view being proposed, for an agent to perform an intentional action just is for her to pursue some purpose in some specific way.

2.  Reasons and Causes—The Uncontroversial Part All this is relevant to the debate about “causalism” in contemporary philosophy of action. Causalism is the view that, as Mele puts it, “human actions are, essentially, events that are suitably caused by appropriate mental items, or neural realizations of those items” (Mele 2003: 38). So if to perform an intentional action is to pursue a goal in some way, the causalist claim will be that to be goal directed in this way just is for the relevant behavior to be caused (in the right way)3 by a desire or intention or other appropriate mental state. This would include Davidson’s claim that reasons for actions are causes of those actions assuming we understand reasons, as Davidson does, as desire/belief pairs (Davidson 1980: 3–19). But since, on one reading, it is uncontroversial that “reasons are causes” (as I will explain in this section), it would be better to say that causalism is the view that reasons explanations are causal explanations in a way that some other explanations are not. To be a “full fledged” causalist would be to hold that “actions” can even be defined as bits of behavior caused in the right way by reasons: e. g. that “[a]ny behavioral event A  of an agent S is an action if and only if S’s A-ing is caused in the right way and causally explained by some appropriate nonactional mental item(s) that mediate or constitute S’s reasons for A-ing” (Aguilar and Buckareff 2010: 2). Here however I am going to stick to the issue of whether reasons explanations of actions are causal, and leave aside the question of whether this is the defining feature of actions. It is important to notice though that there is a common sense of “cause” in which we don’t say anything controversial by saying “reasons are causes”, namely if we are thinking of a minimal understanding of “cause” along the lines of “whatever explains this event”. That is to

62  George F. Schueler understand “cause” as covering the same ground as “because” and so allows essentially any explanatory factor, including purposes, goals, even “absences”, to count as “causes”. For example, an old man who is getting a bit unstable on his feet needs a cane to help him walk. One day he doesn’t take his cane with him and stumbles while on his way to the grocery. If he had had his cane with him he wouldn’t have fallen. So he fell because he didn’t have his cane; the absence of his cane caused his fall. So we need to be careful here. The sense of “cause” in which the absence of his cane can cause someone to fall is so minimal that it covers absolutely any explanatory factor. So it is unsurprising that exactly the same formula works for reasons for acting. But this means it doesn’t raise any issues about explanations of actions over which anyone might disagree. Both causalists and anti-causalists in this debate think that reasons explanations are genuine explanations and in this minimal sense of “cause” that is all that is required for reasons to be “causes”. So if there is to be a difference between the two positions here we need to think of causalists as defending a stronger view in which causal explanations contrast with some other sort of explanations. And the obvious contrast is with teleological explanations. That is why Mele calls the contrasting view “anticausalist teleologism” (Mele 2003: 38). According to causalists, reasons explanations of actions, once we unpack them, are relevantly similar to serious causal explanations in sciences such as physics and chemistry while anti-causalists say they are not. They think reasons explanations are essentially and irreducibly teleological in form. Since both sides agree that reasons explanations really do explain, what causalists want, and anti-causalists resist, is to “reduce” or “explain away” the apparently teleological form of reasons explanations in something like the way evolutionary explanations explain away cases of apparent purposiveness in nature. Evolutionary explanations “preserved” the idea that there are purposes in nature by explaining them away, i.e. understanding them as disguised causal explanations that don’t really employ goals or purposes at all. Causalists would like to do something similar for reasons explanations of actions. Anti-causalists think this cannot be done. Reasons explanations are at least apparently teleological in form since they refer to the goal, aim, or purpose of the action in question. The issue is whether this teleological form of explanation can do the explanatory work as it stands or whether it is somehow covertly dependent on a reference to efficient causes, i.e. whether in order to explain the action in question these apparently teleological explanations must be, at bottom, causal in the way explanations in chemistry or biology are. One issue facing anyone hoping to show that an explanation of some action that refers to the agent’s aim or goal is at bottom somehow dependent for its explanatory power on reference to efficient causation is to say what the cause that triggers the action in question actually is. Aims

Why and How?  63 and goals seem to do the explanatory work in reasons explanations but, like absences, they don’t seem like the sorts of things that could be causes in the required sense. For that, we seem to need events or states, or in any case something that can serve as an efficient cause (see Mele 2013). Following Davidson (1980: 3–19), causalists have commonly dealt with this problem by noting that reasons explanations often make reference to what the agent wants. They then use this fact to justify speaking of the agent’s desires or pro attitudes, which are taken to be mental states, in particular propositional attitudes analogous to beliefs, which can serve as efficient causes.4 So the idea is that teleological explanations of actions only work because an appropriate mental state, perhaps a “pro attitude”, causes the relevant behavior (“in the right way”). E.g. explanations in terms of what the agent “wants” are understood to be about “desires”, which are then thought of as states which can cause actions. I will argue in section  5 that this apparently innocent terminological shift is really quite substantive and problematic. But first we should look at an influential argument in defense of causalism.

3.  Mele’s Defense of Davidson’s Challenge Davidson famously challenged philosophers who don’t think that reasons are causes of actions to say what else could support the “because” in sentences of the form “ ‘He did it because [. . .]’ where we go on to name a reason” (Davidson 1980: 11). Agents who do something for a reason can have, and know they have, other reasons for doing the same thing, reasons on which they do not act. So what justifies saying an agent acted “because” of one specific reason, and not for another, if it wasn’t that this reason caused her to do so? The teleological advocate’s response is to say that since reasons explanations are teleological, what supports the “because” is the actual purpose the agent had in doing what she did. As Löhrer and Sehon put it, “[T]he truthmaker of a correct teleological explanation is the underlying teleological fact, the fact that the behavior was directed by the agent towards that particular end” (Löhrer and Sehon 2013: 5). Consider Wilson’s example of Norm, who is climbing up to the roof of his house to retrieve the hat he left there when he has several other possible reasons for making that climb, having left some tools and some bricks on the roof as well (Wilson 1989: 290). To say he is climbing in order to retrieve his hat, that is that he is climbing because he had the goal of retrieving his hat, and not to retrieve the bricks or the tools, is at least to say that the fact that he had the goal of retrieving his hat explains his climbing. That would seem to give us the required “because” but does not seem to support the causalist claim that somehow this explanation involves efficient causation. It is a teleological explanation, since it refers to Norm’s purpose in climbing that ladder, and it seems to work fine as an explanation without further analysis.

64  George F. Schueler Mele has given a response on behalf of the causalist to this alleged teleological explanation, a response that is supposed to show that the teleological answer, or at least Wilson’s version of it, simply doesn’t work, doesn’t show Norm climbed because he had the goal of retrieving his hat. Suppose, Mele says, that just as Norm is about to begin his climb up the ladder, “random Q-signals from Mars” strike his brain so as to disable the usual neurological connections between his brain and the rest of his body while simultaneously causing him to move exactly the way he would have moved had his movements been under his own control. So it is the Q-signals that cause him to move as he does. We can even suppose that some actual Martians are controlling the rays. So if Norm had changed his mind mid-climb and decided to come back down, the Martians would have turned off the rays and returned control of his body back to Norm. And all this happens so smoothly that he doesn’t even notice that it is the Q-signals that are causing his movements.5 “In that event”, Mele says, “it is false that Norm climbed the rest of the way up the ladder. Although his body continued to move up the ladder as it had been, and although he intended of his movements that they ‘promote the satisfaction of [his] desire [to get his hat]’, Norm was no longer the agent of the movements”. Norm still had the same goal and indeed the same desires and intentions but since it is the Q-signals causing him to move as he does, he is not moving because he has this goal. “Instead, Norm went the rest of the way up the ladder because the Q-signals provided certain input to his muscles” (Mele 2003: 46). So it looks like, with the Q-signals acting as efficient causes of the movements that propel him up the ladder, Norm does not act because he has the goal of retrieving his hat, indeed doesn’t genuinely act at all, in the intentional sense, even though he has this goal and does indeed retrieve his hat. This example shows that merely having a goal and behaving in a way that satisfies that goal, which was Wilson’s version of the teleological account of action explanation, is not enough to justify saying that one acted because one had that goal. The Q-signals caused the behavior and so “Norm was no longer the agent of the movements”, as Mele says. But we need to be careful in describing this example. In the first version Mele gives, the Q-signals are brief and arise randomly. In a later version, he suggests that we can imagine actual Martians controlling Norm’s behavior and insuring that Norm remains unaware of their intervention by relinquishing control if he changes his mind about his purpose. But once we imagine the Martians to be adjusting Norm’s movements to correspond to his intentions, and especially if we imagine them doing this for an extended time, it looks as if Norm is performing intentional actions after all. The Martians are just helping him do so. “Occasionalists”, you may recall, held that God intervened in much the way Mele’s Martians do, connecting human (mental) intentions with their bodily movements every time anyone acted (cf. Lee 2014). And occasionalists

Why and How?  65 did not think this undermined human agency.6 Instead, on their account God was an integral part of the way all intentional actions were performed. If the imagined Martians with their Q-signals reliably adjust Norm’s movements to copy exactly what he would have done had he been in control of his behavior, it is not clear why we shouldn’t say, in the same way, that he is still performing intentional actions too, now with the help of the Martians. So I think we should just forget about the Martians and stay with Mele’s original Q-signals case where the Q-signals arise randomly and affect Norm only briefly, while he is climbing the ladder. Thus described, this example is not really different from the variant on the nail hammering example where it was the drug that caused my arm to move. In that case, because the drug caused my arms to flail about, no intentional action was performed whether or not the nail actually got driven into the wall. Both these cases are what we might call “marionette cases” because no matter what it looks like (or even what the agent herself thinks), at the moment of the action the agent’s behavior is not actually under her control. Her movements are being caused by “strings” pulled by something outside her control for the brief period of the action at issue. And I take it that this is what Mele means when he says of Norm that because it was the Q-signals that caused his behavior, he “was no longer the agent of the movements” here.

4.  What Does Mele’s Argument Show? But it is important to be clear about what this example does not show. Like the marionette briefly operated by invisible strings pulled by a puppet master, both Norm under the influence of the Q-signals, and me under the influence of the drug that caused my arm to flail about, fail to perform any intentional action, no matter what it looks like (even to Norm and me). That is because it is a necessary condition for performing an intentional action that the behavior of the agent not be caused by something outside the agent’s control.7 If my arm with the hammer just moves because of the drug I have been given, or if Norm’s movements are secretly caused by random Q-signals from Mars, then even if each of us retains his original purpose, neither of us is actually pursuing his purpose since our movements are for that period not under our control. So no intentional action was performed and neither the “Why?” nor the “How?” questions have answers. Merely having that purpose and engaging in the appropriate movements are not sufficient for performing an intentional action. But it does not follow from this8 that the explanatory force of reasons explanations must be or even can be explained in causal terms in the way causalism says. Those are further and much stronger claims. Failure to perform an intentional action because your bodily movements are caused by something outside your control does not show

66  George F. Schueler that you succeed in performing an intentional action only if your movements are caused by your mental states in the way causalism claims. True enough, the Q-signals example shows that in such circumstances “a putative teleological explanation of an action in terms of a goal, aim or purpose G does not explain the action”. That is because, as Mele says, no intentional action was performed in such a case. A necessary condition of something being an intentional action has not been met. So there is no action to explain. But it is a different and additional claim, not supported by the Q-signals example, to say, as causalists also want to, that reference to the agent’s goal does not explain the action “unless the agent’s wanting or intending to (try to) achieve G has a relevant effect on what he does” (Mele 2010: 194). That is to say not merely that no explanation is given in the Q-signals case but also that no explanation is given in that case because there was no causal connection between the agent’s wanting or intending and what was done. That is the “causalist” claim and if causalism is true then this will be the correct explanation of what has gone wrong in the Q-signals case. But in order to get that conclusion here we would have to add to the Q-signals case the assumption that causalism is correct about action explanations, which of course would simply beg the question here.9 Causalism is not itself supported by the mere fact that in the Q-signals case no intentional action was performed. No “marionette case” of this sort will be an example of an intentional action, no matter what the correct account of intentional action is. What the Q-signals case shows is that it is a necessary condition of an intentional action being performed that the agent’s movements not be caused by something outside her control. In this example that necessary condition is not met because the random Q-signals block Norm from control of his own bodily movements while themselves causing him to climb up the ladder. They operate on Norm just as the strings would on a marionette or that drug does on me when I am flailing about with my hammer. But this leaves open the issue of whether the teleological structure of reasons explanations of intentional actions is only apparent and somehow depends on underlying causal connections between the agent’s mental states and her bodily movements, i.e. the issue of whether causalism is true. One might not notice this if one thinks of answering the “Why?” question in terms of “what the agent wants” and then translates that into “what desires the agent has”. This was the way Davidson implicitly dealt with the problem that purposes and aims don’t seem like the right sort of things to act as efficient causes of anything. And it is true that in one sense of the term, “desire” can refer to a mental state, such as a craving for chocolate or a preference that it not rain on Saturday. So it seems easy to think that such a state, or its “neural realizer”, might explain the action in question by causing that action. The picture this then gives is that in Mele’s Q-signals story what happens is that instead of Norm’s

Why and How?  67 desire to retrieve his hat causing him to climb the ladder, the Q-signals both block the normal causal consequences of his desire and produce his movements themselves. But as I’ve argued, nothing in the Q-signals story by itself requires this interpretation. We only get this picture if we implicitly import causalism by moving from “Norm wants to retrieve his hat” to “Norm has a desire to retrieve his hat” and then understand “desire” as referring to a distinct state with causal powers. But we don’t want to beg the question here. And the fact that those Q-signals mean Norm did not perform an intentional action does not require that we accept this causal picture of the role of desires in action explanation. All the Q-signals example shows is that it is a necessary condition of action explanation that the agent’s movements not be caused by something outside her control, as if she were for that moment a mere marionette. And that could be true, so far as the Q-signals example is concerned, even if it is not true that causation by reasons is necessary for intentional action in the way causalists claim. The most we can say about Mele’s Q-signals example then is that it shows that Norm’s climb up that ladder was not an intentional action. The larger issue, of whether reasons explanations of actions are only explanatory because they rely on efficient causation, is still open. Mele’s Q-signals example (in the version I have described) shows that having a purpose and behaving so as to achieve that purpose (as Norm does) is not sufficient for the performing of an intentional action. But that is far from supporting the idea that one’s desires or other mental states causing the relevant behavior is sufficient for the performing of an intentional action.10 So we are left with the conclusion that the teleological account of agents’ reasons and the causalist account are in this respect “on all fours”. On neither account would Norm’s Q-signal induced movement up that latter count as an intentional action.

5.  Agents’ Reasons Are Not Causes So the Q-signals example leaves the teleological view no worse off than causalism itself. But I want to argue that causalism is false. We can start by reexamining how Davidson understands agents’ reasons as pairs of mental states, desires and the associated beliefs about how to satisfy them. As I have argued, this idea presupposes that since an intentional action always involves the agent trying to get what she wants, we are always justified in speaking of a mental state, a desire, which can then be regarded as having causal powers. But there is a gap in the reasoning here. The phrase “what the agent wants” might not refer to some inner state of the agent at all. It is just a mistake to think that because purposes or goals are commonly described as “what the agent wants” in doing whatever she did, we are automatically justified in thinking that there is some internal mental state, a “desire” or “pro attitude” that then might

68  George F. Schueler cause the action. Sentences of the form “X wants Y” cover two distinct phenomena: things the agent likes or is in favor of, and purposes or goals she has. The jargon term “pro attitude” covers up this distinction but these are different things.11 Here is an example. If I am anticipating tomorrow’s picnic, I want the weather to be nice. I’ll be glad if it is, unhappy if it isn’t. But the fact that I want this need not provide me with a goal that might lead me to do anything. I realize that nothing I can do can influence the weather. And so unless I am irrational I won’t take producing nice weather tomorrow as the goal of anything I do. But I still want nice weather. My desire for this consists of the fact that I am in favor of it. So here the term “desire” apparently refers to an internal mental state of favoring. I will call this a “proper desire”. The phenomenon involving purposes is exemplified by intentional actions.12 And the term “the agent’s reason” typically refers to the purpose of the action, what she wants. If I walk across campus to check my parking meter, my reason for doing this is the purpose of my action, i.e. to check my parking meter. I would describe this is by saying “I want to check my parking meter”. The purpose of the action is typically referred to when we speak of an agent who is performing some action as wanting something. But having a goal is not the same thing as having a positive or favorable attitude toward something. I might have a favorable attitude toward, say, having a chocolate éclair while realizing how unhealthy such things are and so refusing to adopt the goal of eating one. So I want an éclair in the “favoring” sense; I might even feel a strong craving for one. But I am not being dishonest when I turn down the offer of one by saying that I do not want an éclair. I don’t have the goal of eating one. I might even have it as one of my goals not to eat such things. It is perfectly possible to perform an action with some goal, such as avoiding sweets, for which one has no proper desire, a goal that one does not “like” or “favor”. Think of the last boring meeting you forced yourself to attend. So sentences of the form “X wants Y” cover two distinct phenomena: the agent liking or having a positive feeling toward something, and purposes or aims she has in doing what she does. And though these can of course overlap, they can also come apart, as we just saw. One can be strongly in favor of something the achieving of which is not, even could not be, one of one’s goals. And one can pursue a goal to which one has no “positive attitude”. Proper desires can plausibly be taken as including inner states such as cravings and urges, which presumably are the sort of things that might cause other things. But it is not plausible to take purposes in that way, though it is purposes that are required for performing intentional actions. Why couldn’t the purpose of some action (or its “neural realizer”) be the cause of that action? The answer is that the purpose of an intentional action is the content of the intention analogous to the intentional content

Why and How?  69 of a belief, though of course in the case of an action the content is not a proposition (perhaps a reference to a state of affairs?). But in any case it doesn’t make sense to think of this as “causing” the action any more than it makes sense to think of the content of a belief as causing it. The purpose of an intentional action is a logically constitutive feature of that action, just as the propositional content of a belief is a constitutive feature of that belief. So the connection between my intentional action and the goal I was pursuing in performing it is not of the right sort for causation, any more than the connection between a belief and the content of that belief is. If I  am heading across campus in order to check my parking meter then I am doing this because I want to check (have the goal of checking) that meter, even if I don’t like doing this at all. The fact that I have this purpose constitutes my reason for heading across campus and explains what I am doing. I am heading across campus because I want to check my parking meter.13 But my having this purpose doesn’t cause me to head across campus, in the sense of “cause” required by causalism. Causes have to be distinct from their effects.14 But that makes no sense for an intentional action and the purpose that explains it. To perform an intentional action is to pursue some purpose in some way. If I  am walking across campus in order to check my parking meter then my reason for walking across campus, my purpose in doing this is to check my parking meter. If my purpose here were the cause of my action, and so only contingently connected to it, then there could be some other cause of this same action. But if we try to imagine some other purpose, say returning a book to the library, causing this action, the result is incoherent. “I  walked across campus in order to check my parking meter because I wanted to return my book to the library” seems unintelligible.15 What could this mean? Of course I might walk across campus in order to return a book to the library but then the goal of returning a book to the library didn’t cause me to check my parking meter. It led to a different action, returning a book to the library. So to make sense of “a purpose that causes my action” it will have to be because some mental state, not itself the purpose of the action, causes the movements involved in what I  do and this whole becomes my “action-with-some-purpose”. Mere movements don’t constitute an intentional action. I might “walk across campus” while sleepwalking, for instance. So the causalist claim will have to be that the intentional action (i.e. including its purpose) consists of some behavior caused by a proper desire or other mental state(s). As Mele says, “human actions are, essentially, events that are suitably caused by appropriate mental items, or neural realizations of those items” (Mele 2003: 38). The “why” question is then presumably answered on this view by reference to the mental state (presumably a desire or similar “pro attitude”) that caused this behavior. This desire (or the desire plus the behavior it causes) is what provides

70  George F. Schueler the purpose of the desire-caused-behavior, which on this account constitutes the intentional action. Obviously, this will only work if we can make sense of a mental state that is, or somehow produces, “my purpose in performing this action” and which is itself distinct enough from my behavior that it could cause that behavior. There would seem to be two possible “mental states” that might serve as causes of behavior in this way: what I  have called “proper desires” where I am in favor of some state of affairs, such as my desire that I eat some ice cream,16 and intentions such as my intention to stop at the grocery on the way home this evening. But on examination neither of these is a plausible candidate to both cause and provide the purposes of actions. Intentions that occur prior to the action pretty obviously won’t work as the causes of actions causalism needs since most intentional actions are not preceded by such prior intentions. Think of the last clever witticism you made or the last casual conversion you engaged in. Your speech acts were intentional but probably not preceded by any prior intentions to say what you did. So if the causalist wants to use prior intentions as causes of actions, most intentional actions will not be covered. And in any case, causalism is supposed to explicate the view that “reasons are causes”, i.e. explanations of actions in terms of the agents’ reasons are said to be causal explanations. But even in those cases where an agent does something she had a prior intention to do, it is very rare that an agent’s reason for performing some action is that she had earlier formed the intention to do it. If I intend to stop at the grocery on the way home from campus and then subsequently do so, my reason for stopping is unlikely to be simply that I earlier formed the intention to stop. Presumably, I formed that earlier intention because I thought there was some good reason to stop and when I did stop, I stopped for that reason. So even when some action is preceded by a prior intention to perform it, that prior intention is very unlikely to be the agent’s reason for doing what she did. Proper desires, such as my desire that I eat some ice cream, might seem better candidates for causes of my behavior, at least when they have the appropriate content, but they are not. The most obvious problem is that it is just not the case that all intentional actions have as goals things the agent likes or has a positive attitude toward. Failure to distinguish wanting something in the sense of having it as a goal of one’s action and wanting something in the sense of liking it or being in favor of it might lead to confusion here. But once we make this distinction it should be clear that though it is always true that there is something I “want” when I perform some intentional action, this refers to the goal of my action, and is not necessarily something I have a pro attitude toward in the sense of liking or favoring it. Keeping a promise one wishes one had not made might be an example. Or for that matter, when I turn down that éclair in spite of the fact that I would really like to eat it, what I want (my goal) is to cut down on sweets. I might do that in spite of the fact that I am only cutting

Why and How?  71 down on sweets because my doctor told me to, not because I somehow like not eating sweets. The purposes of my intentional actions no doubt sometimes involve things I have a favorable attitude toward. But there would seem to be plenty of other things that make some goals worth pursuing, or lead people to pursue them, other than the fact that the agent likes what is being pursued. There is also a deeper problem for causalists in taking proper desires as causes of behavior. Even when I have a proper desire for something and act in order to satisfy that desire, the purpose of my action is not the same thing as the object of my proper desire. Suppose I have a craving for ice cream (a proper desire if anything is) and so I go to the fridge in order to get some. That might seem the clearest sort of case where my proper desire causes me to act. The goal of my action is that I get some ice cream. But even in this case the purpose of my action and the object of my proper desire are different. Something more is needed to go from the craving for ice cream to having the goal of getting some. I must decide to act on that craving. On the basis of the fact that I have a craving for some ice cream, I decide to act in such a way that I get some. I go to the fridge in order to get some ice cream. But my purpose here is the result of my decision, which in this case was indeed based on the fact that I had a craving for ice cream, but it is not the same thing as the craving. I could have decided to get some ice cream without having the craving and I could have had the craving without deciding to satisfy it. The craving by itself won’t get me to the fridge. There is a difference between my proper desire for some ice cream and the purpose of my trip to the fridge. I might go to the fridge in order to get some ice cream even though I don’t have any craving (or other proper desire) for ice cream. Someone else might have asked me to do that, for instance. And I might have a craving for ice cream and yet not act so as to get any. I might decide to try to extinguish that craving instead. So even when I do act on my craving, what happens is that I decide to get some ice cream (perform that intentional action) on the basis of the fact that I have a craving for ice cream (I have that proper desire). It is not that the craving itself somehow automatically generates movement toward the fridge. Once we distinguish proper desires, which are mental states with objects but don’t involve purposes, from the fact that actions have purposes, we can see the problem for causalism. My proper desire only appears to generate the goal of the action it is supposed to cause because in citing my desire for something we have, mistakenly, identified the object of my desire with the purpose of the action it is supposed to cause. Desires that are supposed to cause actions are thought of as providing the purpose of the actions caused, via their objects, and yet nothing is said about how the object of a desire is supposed to be transformed by causation into the purpose of the action it produces. In order to work as an explanation of this action the desire has to be thought of as providing the purpose of the action. But it is simply a

72  George F. Schueler mystery how a proper desire could be the same as, or somehow automatically generate, the purpose of the action to which it leads. That purpose arises only if one decides to satisfy some desire one has. And even then it is the decision to satisfy the desire that gives the purpose of the action, not the desire itself. After all, one might decide to squelch or just ignore a desire, rather than to satisfy it. The purpose of the action that some desire “leads to” comes from one’s decision as to which action to perform and the intentional action that results from that decision. That is obviously true if one decides to try to eradicate the desire. My desire for an éclair plus my belief that éclairs can be had in this bakery might lead me go in, but it might also lead me to turn away, if I decide not to satisfy this desire but to try to squelch it. The confusion arises from the fact that if one decides to satisfy some desire then the object of the desire and the purpose of the resulting action can often be described in very similar terms. If I crave an éclair then the object of my desire is that I have one. If I decide to try to eradicate this desire then of course the purpose of the resulting action will be to be rid of my desire for an éclair. But if I decide to try to satisfy this desire then the purpose of the resulting action will be to have an eclair, the same as the object of my desire. But this purpose is not somehow “automatic” given the object of my desire. My action only has this purpose because I now intend to satisfy my desire for an éclair. So the causalist picture contains a serious gap. On this picture an intentional action consists of some bodily movement caused (“in the right way”) by a proper desire. But an intentional action always has some purpose. And neither the proper desire, nor the bodily movement it causes, nor the causal relation between them, accounts for the purpose of that intentional action. We can fail to see this problem in causalism if we don’t notice the difference between the object of a proper desire and the purpose of an action. But once we do notice this it is hard to see how a proper desire causing a bodily movement could constitute an intentional action. It is hard to see how these elements could combine so as to produce the purpose of such an action. Even if the bodily movement should result in the satisfaction of the desire, that doesn’t mean the movement had this result as a goal, any more than any other causal consequence of this movement was its goal. And in any case of course sometimes a desire will lead the agent to perform an action the purpose of which is quite different than the object of the desire that led to it, as when someone acts so as to extinguish a proper desire, rather than to satisfy it. So the causalist picture simply leaves unexplained an essential feature of intentional action, its purpose. Have we overlooked a possibility? We have examined two mental states that might serve as causes of actions, prior intentions and proper desires. On examination neither is plausible. Proper desires don’t provide purposes for the actions they are supposed to cause and prior intentions, while they do involve purposes, don’t precede all actions and don’t

Why and How?  73 provide reasons for the actions they precede. But it might be held that there are intentions that are simultaneous with actions (or behavior) and which are intentions that this behavior achieve some purpose or goal. Could such “proximate intentions” be held to cause this behavior?17 No. This is not a genuine possibility. The purpose of an action is the intentional content of that action, analogously to way the proposition believed is the intentional content of a belief. So the suggestion that there is an independent intention simultaneous with each intentional action that provides this purpose by causing the behavior makes no sense. Even if we ignore the problem of causes generating purposes, the intentional content of the action is not a separate “state” which could cause anything. One way to see this is to notice that if we accept this idea that reasons are “proximate intentions” then Davidson’s Challenge, which is central to the argument for causalism, will no longer work. In the case of Norm climbing that ladder, the idea was that Norm has three reasons to climb since he has three things he wants to retrieve, his hat, his tools, and some bricks. Since he only climbs for one of these reasons, to retrieve his hat, “Davidson’s challenge” is to say why this is so without supposing that it was this desire for his hat that caused him to climb. The causalist claim was that he climbed for one of these three reasons, and not either of the other two, because that desire caused him to act while the other two, though he still had them, were not causally active. But if we try substituting proximate intentions for proper desires in this example the result is incoherent. The claim would then be that Norm’s proximate intention to retrieve his hat by climbing the ladder caused him to climb the ladder. At the same time though his proximate intention to retrieve his tools by climbing the ladder and his proximate intention to retrieve the bricks by climbing the ladder would both still have to be there, i.e. he still has them, since they are supposed to provide the reasons on which he does not act. They are just not causally active. But this makes no sense. We would then have to say both that Norm is climbing the ladder in order to retrieve his hat and, because he still has the other two proximate intentions, that he intends of his climb that he retrieve his tools and that he intends of his climb that he retrieve the bricks. The “challenge” will only be possible if he still has the other two supposedly “inactive” proximate intentions, even if he perfectly well knows it is not possible to do more than one of these three things.18 But it is not possible to intend to do something one knows is incompatible with something else one intends to do. A driver who, arriving at a street corner, intends to turn right, cannot also intend to turn left (at the same time at the same intersection). I might both want to turn right and want to turn left. And my longing to turn left may not go away even when I decide to turn right. But I cannot intend to do both these things, at least not if I realize that doing one precludes doing the other. Proper desires can conflict without one eliminating the other. But for intentions, as for

74  George F. Schueler beliefs, the intention to do something precludes intending to do anything known to be incompatible with it.19 It is true enough that intentional actions “contain” or “are informed by” purposes which can perfectly well be described as “what the agent intends to do” or the like. But such intentions are not distinct mental states, like proper desires or prior intentions. If they were such distinct states a version of Davidson’s Challenge would be possible. But it is not. So we were not wrong in thinking that the only two genuine candidates for the mental states that could serve as causes of the behavior involved in an action are proper desires and prior intentions. And as we have seen, neither of these is plausible. What this means is that the Davidson’s implicit transition from “S wants Y” to “S has a desire for Y” is deeply problematic if desires are then understood as states that can serve as efficient causes of actions. The fact that intentional actions are done in pursuit of purposes, that is, that they require that there be an answer to the “Why?” question, is commonly registered in ordinary speech by saying that intentional actions always involve something the agent wants. But that is not the same thing as saying that there is an inner mental state of favoring involved, a proper desire, or indeed any distinct mental state which might then serve as the efficient cause of something else. Saying that Norm wants to retrieve his hat from the roof is simply a way of describing the purpose of his action. The Q-signals example, while it does show that in the circumstances described Norm performed no intentional action, leaves the larger question of the truth of causalism open. Thinking about why this example gets no further in advancing causalism has suggested that there is a serious problem in causalism itself. Causalism needs two points if it is to be correct: The phrase “the agent’s reason” will have to refer to a (proper) desire-belief pair (or some similar mental states or their “realizers”) and the explanation provided by referring to these mental states will have to be causal. I  have argued that exposing the mistake involved in not noticing that the agent’s reason (what she wants) is just the purpose of her action blocks the first of these points and, as a result, also blocks the second. This suggests that if the phrase “the agent’s reason” refers to the purpose of her action then, given what actions are, agent’s reasons cannot be causes of actions. Causalists face the challenge of trying to explain the purposes or goals of intentional actions in completely causal, and hence non-purposive, terms. If the arguments here are correct, it appears that this challenge cannot be met. At the same time, even if causalism itself is not true, there is still an essential causal element to any action explanation. To perform an intentional action is to pursue a purpose in some way.20 That is, if there is to be an intentional action there must be some way in which the goal or aim of that action is being pursued. That certainly seems to entail that it is a necessary condition of something being an intentional

Why and How?  75 action that the movements that instantiate it be explainable causally. At the same time, intentional actions are always done in pursuit of some purpose or goal. So it is also a necessary condition of something being an intentional action that there be an answer to the question of why it was done. The difficulty of answering the challenge just described suggests that this is simply a distinct question.21

Notes 1 If such things as spontaneously clapping one’s hands at some good news, or whistling as one walks along are intentional actions not done for reasons then what I say here will not apply to them. 2 If so-called “basic actions” are intentional actions then this applies to basic actions as well as to any others. Sometimes one does something by doing something else, e.g. one might signal the waiter by raising one’s hand. But if, as it is sometimes claimed, I don’t raise my hand “by” doing anything else, I just raise it, then this is a “basic action”. But if I intentionally raise my hand then I do it for a reason, i.e. in pursuit of some goal or purpose. To perform an intentional action is to pursue some goal in some way whether that goal is pursued by performing a basic or a non-basic action. If I had no purpose in raising my arm then, on this view, whether or not we call this a “basic action”, it was not an intentional action. Of course if I  raise my hand “in order to show that I can do it” or “in order to demonstrate a basic action” then I had a purpose in raising my hand. 3 The phrase “in the right way” needs to be added because of the problem of “deviant causal chains” first identified by Davidson himself (Davidson 1980: 79). The problem is that it is easy to think of cases where the desire/belief pair that according to Davidson’s version of causalism explains some intentional action can, when the causal chain between the mental states and the behavior is different, cause behavior which is not an action at all. 4 Though this is contested by Brian McLaughin in 2013. 5 Since Mars is between approximately 3 and 22 light minutes from Earth (depending on the positions of the two planets in their orbits) changes in the Q-signals controlling Norm by the Martian controllers couldn’t actually reach Norm in anything like the time that would be needed to make his movements appear “seamless”, if he decided to go back down the ladder, for instance. So for purposes of this example maybe we should regard these Q-signals as emanating not from the planet Mars but from Mars, PA, which is a small town north of Pittsburgh and only a few light nanoseconds from any other place on Earth. 6 Thanks to Scott Sehon for reminding me of this. 7 I expect “outside the agent’s control” is inherently vague but I assume that this by itself is not a disabling feature of this condition. The existence of borderline cases is consistent with clear cases on each side of the “border”. The cases discussed here are cases where the agent’s movements are outside her control if anything is. 8 Nor does Mele claim it follows. 9 For the same reason, it would beg the question to describe the Q-signals in this example as “blocking Norm’s desires from causing his movements” or the like. 10 As I said above, Mele himself does not claim that his example shows that a causal connection is sufficient.

76  George F. Schueler 11 As philosophers have often recognized. This distinction is implicit for instance in Joseph Raz’s claim that “There are many things people want but achieving them or getting them is not among their ends” (Raz 2005: 9). 12 It may be what Anscombe meant when she said, “The primitive sign of wanting is trying to get” (Anscombe 1963: 68). 13 Davidson himself gives a perfectly good example of this: “I flip the switch, turn on the light, and illuminate the room. Unbeknownst to me I also alert a prowler to the fact that I am home. Here I need not have done four things, but only one, of which four descriptions have been given. I flipped the switch because I wanted to turn on the light and by saying I wanted to turn on the light I explain (give my reason for, rationalize) the flipping” (Davidson 1980: 4–5). So the purpose here was to illuminate the room. That was why this action was performed. The method of achieving that goal was by flipping the switch. That was how it was done. 14 This is not a point about how we describe causes and effects since we sometimes describe effects “in terms of their causes”. “Sunburn” is a condition of the skin caused by exposure to the sun. So how it was caused is part of the “concept” of this skin condition, if we describe it this way. But still, I can refer to this skin condition without describing it as caused by exposure to the sun, as a certain sort of damage to skin cells for instance. Whatever caused that condition, and however we describe it, the skin condition and its cause are different things. 15 I am assuming of course that I  didn’t think that for some reason I  needed to check my parking meter in order to return my book to the library. That would make checking my parking meter part of returning my book, and not a distinct goal. 16 Of course, I can have a proper desire for something utterly outside my control, such as my desire that the weather be nice for our picnic on Saturday. But presumably the proper desires that cause actions will have to have as objects things I at least believe I can affect. 17 This sort of view is discussed in Searle (1983), especially Chapter 3. 18 That is a requirement here, in order to parallel the original version of Davidson’s Challenge where Norm has three distinct desires. Of course Norm might intend to retrieve all three items on his roof, just as he might have a proper desire to do that. But those would just be different examples. In order to get this “challenge” going the three desires (and so the three proximate intentions) must be distinct. So probably, we should assume in each version of the “challenge” that Norm thinks that each item on the roof can only be retrieved separately. 19 On the difference between conflicts of desires and conflicts of beliefs, see Bratman (1987), esp. Chapter 3. 20 This is redundant of course since to pursue a goal one must pursue it in some specific way. 21 Earlier versions of this paper were read at a conference on explanation in action theory and historiography in Hagen, Germany, in March  2016 and at the University of Delaware in February  2016. I  have benefitted greatly from comments by those at both venues. Special thanks are due to Gunnar Schumann, Giuseppina D’Oro, Alfred Mele, Guido Löhrer, Scott Sehon, Julia Tanney, Jeremy Cushing, Kai Draper, Mark Greene, Richard Hanley, Jeff Jordan, Joel Pust, Tom Powers, and Seth Shabo.

References Aguilar, J. and Buckareff, A. (2010). The Causal Theory of Action: Origins and Issues. In: J. Aguilar and A. Buckareff, eds., Causing Human Actions: New

Why and How?  77 Perspectives on the Causal Theory of Action, 1st ed. Cambridge: MIT Press, pp. 1–26. Anscombe, E. (1963). Intention, 2nd ed. Ithaca: Cornell University Press. Bratman, M. (1987). Intentions, Plans, and Practical Reason. Cambridge: Harvard University Press. Davidson, D. (1980). Essays on Actions and Events. Oxford: Oxford University Press. Lee, S. (2014). Occasionalism. In: The Stanford Encyclopedia of Philosophy [online] Spring, 2014 ed. Available at: http://plato.stanford.edu/archives/ spr2014/entries/occasionalism/ [Accessed Dec 20, 2018]. Löhrer, G. and Sehon, S. (2013). The Davidsonian Challenge to the NonCausalist. San Francisco: APA Pacific Div. Meeting. McLaughlin, B. (2013). Why Rationalization Is not a Species of Causal Explanation. In: G. D’Oro and C. Sandis, eds., Reasons and Causes, 1st ed. New York: Palgrave Macmillan, pp. 99–107. Mele, A. (2003). Motivation and Agency. Oxford: Oxford University Press. Mele, A. (2010). Teleological Explanations of Actions: Anticausalism vs. Causalism. In: J. Aguilar and A. Buckareff, eds., Causing Human Actions: New Perspectives on the Causal Theory of Action, 1st ed. Cambridge: MIT Press, pp. 183–198. Mele, A. (2013). Reply to Löhrer and Sehon. San Francisco: APA Pacific Div. Meeting. Raz, J. (2005). The Myth of Instrumental Rationality. Journal of Ethics and Social Philosophy, 1(1), pp. 1–28. Searle, J. (1983). Intentionality: An Essay in the Philosophy of Mind. Cambridge: Cambridge University Press. Wilson, G. (1989). The Intentionality of Human Action. Stanford: Stanford University Press.

4 Rationalizing Principles and Causal Explanation Guido Löhrer and Scott R. Sehon

1. Introduction Historiography, as an academic discipline, concerns itself with explanations that are usually more complex and heterogeneous than explanations of individual behavior. Historically significant actions and events will typically involve a complex sequence of events, comprising many distinct actions (individually or jointly executed, cooperative or antagonistic) and their consequences. We suggest that historical explanations of these sorts are abbreviations for a long explanatory story involving actions by many people, effects of those actions, and further actions by people, etc. Sometimes an individual’s behavior makes the most sense when not viewed as self-contained but when regarded as a contribution to a joint action or coordinated follow-up to other people’s past activities. Even so, the fundamental elements of such explanations will comprise reason explanations of actions by individual agents, along with ordinary causal explanations of the effects of those actions. Sometimes it is a useful fiction to treat big events like a stock market crash as one thing that could be the explanans in a proper singular explanation. Thus, we might say that the Wall Street Crash of October 1929 caused the Great Depression. Even so, each event—both the explanans and the explanandum— comprise, in part, many actions by individual people. In previous writings, both individually and together, we have defended a view about the nature of reason explanations of human action: We take the position that such explanations are irreducibly teleological rather than causal. Thus, given the view we have proposed concerning action explanation, the long explanatory story will involve both questions about the purposes of individual agents as well as causal effects of those actions, and then further purposes of other agents. It concerns a big mixture of teleological and causal explanation. In this paper, we will say a bit about that view (section  2) and its application to historical explanation, which we will discuss via a couple of examples from World War II (section 3). The teleological account of action explanation is controversial. According to the more orthodox

Rationalizing Principles  79 causal theory of action, reason explanations of individual behaviors are a species of causal explanation, invoking the agent’s reasons or intentions or their neural correlates or even facts about what the agents intended as the cause of the behavior (cf. Davidson 1980: 3 and 9; see Mele 2013: 168, on the cautious causalist). In sections 4–5, we will provide an argument against the causal account.

2.  Rationalizing Principles and the Teleological Account Almost everyone (with the possible exception of Quine-style eliminativists) will admit that there is an important distinction to be made between two sorts of human bodily activities: (1) those that are actions and (2) those that are mere happenings. In the first category are behaviors that we do for reasons—things we do on purpose, the behaviors for which we can be held responsible. This will include momentous actions like voting a particular way in a national election but will also include routine affairs like getting a glass of water or typing a sentence. However, there are many bodily motions and happenings that fall into the second rather than the first category. For example: hair growing, digestive processes in the stomach, falling down when pushed with sufficient force and convulsive motions during an epileptic seizure. Both the causal theory of action and the teleological account give theories of what we are doing when we are explaining actions—the behaviors in the first category. On the teleological account, behaviors done on purpose are explained by citing the state of affairs towards which they were directed. The general form of the explanation will be something like this: Agent A B’d in order to C. For example, we might say that Vera went to the kitchen to get water, or that Della walked to the polling place in order to vote. The causalist will presumably agree that we give explanations of this form, but according to the causalist such explanations are shorthand for a different sort of explanation. When we say, for example, that Vera went to the kitchen to get water, we are really saying that some appropriate mental state (e.g., an intention to get water) or neural correlate of a mental state caused her motion toward the kitchen. How do we know which reason explanations are true? On the causal account, this should come down to our normal scientific methods for determining when one thing caused another. First and foremost, we will suspect that X caused Y if events of the first sort have been frequently correlated with events of the second. Of course, we cannot infer causation from such correlation, for there are various other possibilities: There might, for example, be some further type of event, W, that causes both instances of X and Y. We won’t go into details here,

80  Guido Löhrer and Scott R. Sehon but there are many scientific contexts in which we think, Hume notwithstanding, that we are on relatively firm ground in saying that one event caused another. Presumably, if reason explanations are causal explanations, then we would ultimately rely on such scientific methods for determining the truth of a particular reason explanation. More on this in sections 4 and 5. On the teleological account, we do not exclusively use the methods of the natural sciences in determining the truth of a reason explanation. Instead, we make teleological explanations by employing a strengthened variant of Davidson’s principle of charity as a rationalizing principle (see Davidson 1970: 221–222; also see Dennett 1987). We attempt to make sense of the observational behavioral data, ascribing to the agent goals of intelligible value where it is understandable that the action in question could be thought to help achieve the goal. With the help of this method we strive for an overall theory of the agent which makes her behavior both synchronically and diachronically coherent (see Löhrer 2006) and as rational as possible in actual and in nearby counterfactual situations (see Sehon 2016: 28). In terms of the abstract example given above, we will explain agent A’s action (doing B) by ascertaining which goal state (C) would make the most sense of A, given other facts we know about the situation and about A. In normal cases of action explanation, application of such rationalizing principles is automatic and virtually invisible. If we see Jo putting toothpaste on a toothbrush, we assume she is doing this in order to brush her teeth, for the behavior is well suited to preparing to brush her teeth, and brushing her teeth has understandable value given the circumstances. We don’t contemplate the possibility that Jo is putting toothpaste on her toothbrush in order to instigate a trade war with Europe; absent a very special story, there would not be any obvious reason that a trade war with Europe would be of value from Jo’s perspective, and, more importantly, Jo would have no reason at all to think that putting toothpaste on a toothbrush would be an effective way to start a trade war. While we would not be able to attach precise numerical values to different candidate explanations, an explanation can make more or less sense of the agent in at least a couple of different ways. First, if we postulate that the behavior was aimed at achieving goal G, the behavior might be more or less effective as a means to achieving G; second, G could have higher or lower degrees of comprehensible value. For example, we would not normally contemplate the possibility that Jo is putting the toothpaste on the toothbrush in order to prepare to scrub the toothpaste onto the bathroom mirror. Her action would be a good means to that end, but it would be puzzling that she would have the goal of putting toothpaste on the mirror. Still this candidate explanation would make more sense of her behavior than would the claim that she was depositing the toothpaste in order to start a trade war; for in that case, the value of the alleged goal is

Rationalizing Principles  81 questionable at best, and it is ridiculous to think that she could achieve the goal by putting toothpaste on a toothbrush. Of course, the mere fact that we have found one possible goal state that seems to make sense of Jo’s behavior does not mean that we have found the correct explanation. For one thing, Jo might have had more than one reason—more than one goal—for her action. In addition to facilitating the further aim of brushing her teeth, she may have been aiming to stop her mother’s nagging. For another, further data might give us reason to rethink our original conclusion. If, after putting the toothpaste on the toothbrush, we see Jo hand the toothbrush to her little brother who then begins to brush his teeth, we change our mind about initial explanation of the behavior. But this is not a case of abandoning rationalizing principles; on the contrary, we now have new data (Jo handing the toothbrush to her brother), and our initial hypothesis no longer makes as much sense of her overall set of behavior. Our use of rationalizing principles in some ways comes out most clearly in cases where no rationalizing explanation immediately jumps to mind—i.e., when the behavior does not appear to be very rational. United States President Donald Trump provides a number of examples of such behavior. Here is one: his tendency to make public declarations that are simply and verifiably false. One notable example came early in his presidency when he and his spokesperson made claims about the size of the crowd at the 2017 inauguration, claims that were easily falsified by photographic evidence from numerous sources. Another oft-repeated lie is Trump’s claim that the United States is the “highest taxed nation in the world” (see Bloom 2017). The problem for Trump is that there are sources and data that clearly show that, by any reasonable way one would measure tax rates, the United States has relatively low taxes. Trump’s claim is demonstrably false, and the press is not shy about noting this. One would think that there would be something embarrassing and politically damaging about repeating a claim known to be false, and thus the behavior is initially baffling. Various theories have been put forward to explain the President’s behavior. First and foremost, commentators suggest that Trump tells falsehoods for political advantage in getting his agenda through. On this view, he is aware that some people will believe things if told them often enough, particularly in a climate (that Trump very much helped to create) in which many people on one side of the spectrum will simply believe what they want to hear and reject data given by mainstream news sources (see Konnikova 2017, DePaulo 2017). On an even stronger version of this theory, Trump is gaslighting us: boldly repeating lies often enough to make the American public doubt its ability to perceive things correctly, hoping thereby to gain political advantage (see Ghitis 2017). But there is a second theory that is frequently mentioned, namely that Trump believes his own lies (see Eichenwald 2017 and McLaughlin 2017). There

82  Guido Löhrer and Scott R. Sehon could also be other explanations for different examples of untruths told by Trump. Either way, precisely because it seems initially puzzling that someone would tell lies that one can easily unmask, we become more self-conscious in doing what we do in ordinary explanation all the time: We try to find the most rational story we can tell about the behavior; we try to make as good sense of it as we can. Moreover, to draw on another Davidsonian point, this use of rationalizing principles is not just some extra heuristic, something we bring in as a tie-breaker when other evidence is insufficient. Apart from using rationalizing principles, interpretation of agential behavior in terms of reasons would be wildly underdetermined. If we did not feel constrained to make rational sense of a person’s behavior, we could attribute almost any set of beliefs and desires we wanted as explanatory factors. To return to Jo and the toothpaste, apart from rationalizing constraints, it is perfectly consistent with all behavioral evidence to conclude that she is putting toothpaste on the toothbrush in order to instigate a trade war with Europe. All we need to do is to attribute to Jo the desire to start a trade war with Europe and the belief that she can do so by putting toothpaste on her toothbrush. One might protest that we have no evidence for these attributions; Jo might have even told us that she is going to brush her teeth and that she has no interest in starting trade wars. But that’s no problem, for we can further postulate that Jo likes to deceive people about her trade-related beliefs and desires. Moreover, even that move is a further attempt to rationalize her behavior, insofar as it postulates a desire (deceiving people) that would make sense of the statements and actions. Entirely unconstrained by any such rationalizing goal, we could simply hypothesize that, in Jo, the desire to cause a trade war with Europe causes her to put toothpaste on toothbrushes and causes her to deny that she has any such desire. Such behaviors may make no sense, but if we give up rationalizability, we give up making sense as a desideratum on explanations of behavior.

3. The Use of Rationalizing Principles in Historical Explanations Rationalizing principles are similarly at work in explanations of historical events. To take one example, at the end of May 1940, German troops had trapped the British and French Allies at Dunkirk, but the German armored forces were then commanded to halt. This order had the effect of allowing the Royal Navy to complete a massive evacuation effort that saved approximately 338,000 British and French soldiers from almost certain defeat and capture. Naturally, this is puzzling: In the context of a war, one does not normally allow the enemy to escape. So there is the question: Why did the German commanders give the order to halt, and why did Hitler sanction the order?

Rationalizing Principles  83 This is one of the most debated decisions of the Second World War, and historians have offered several explanations. One theory says that Hitler wanted the order to be interpreted as an offer to Churchill to strike up a compromise peace. Another theory says that Hitler and the generals halted the advance on Dunkirk in order to consolidate and to avoid an Allied counter-attack. A  further theory states that the halt would allow the armored forces to regroup and that the Luftwaffe would, as Göring promised, be able to annihilate the allied forces single-handedly (see Bond 1990: 104–105, and Kershaw 2008: 27). There are even some fringe claims that Hitler aimed to prolong the war because it was in the interest of British bankers, and Hitler was really an undercover British agent (see Rothschild 2013). Note that each such explanation, even the fringe ones, cite something that at least might be taken as a justifying aim of the action. We don’t consider the possibility that the halt order was issued because Hitler preferred the sight of parked tanks when visiting German headquarters in May 1940, nor do we entertain the hypothesis that Hitler thought that the order would lead to favorable intervention by inter-galactic aliens. Apart from our reliance on rationalizing principles, we have absolutely no reason not to make such crazy attributions. It is not just that the behavioral evidence underdetermines the conclusion, for we must live with underdetermination in all sorts of contexts. Rather, the point is that we cannot even begin to justify attribution of one reason rather than another (or one mental state rather than another) without some sort of background assumption to the effect that we are attempting to make sense of the agent. Of course, it is not enough to find some aim that is of conceivable value in an action. We do seek to find the correct answer—the aim or aims towards which the behavior was in fact directed. But there is an intrinsic link between the question and the rationalizing method, and we seek the goal state that makes the most sense of the actors at the time, given everything we know about them. This is why, for example, historians ultimately rule out the idea that the halt was aimed at the prospect of peace negotiations, for this does not fit with other actions in the surrounding timeframe, not least of which is the lack of any offer of peace in the ensuing weeks. (Hitler did apparently once claim in 1945 that this was the goal of the action; we see this to be untrue precisely because it doesn’t make sense of his other actions in the intervening periods.) The whole interpretive project ends up having a recursive flavor: The other data that we have about the actors at the time is itself largely a matter of interpretation done on rationalizing grounds. Rationalizing principles are used pervasively throughout the project of ascertaining historical explanations. Action explanation is even more complicated when it comes not to a single person’s decision or consent, but to the result of lengthy disputes

84  Guido Löhrer and Scott R. Sehon fought out during committee meetings. In such a case, it may happen that a joint resolution must be made comprehensible which at first glance does not correspond to any single contribution of any of the decision-makers involved. To give another historical example, in autumn 1941, a dense series of deliberations by Japanese military and political leaders resulted in an imperial consent to go to war against the United States. Studying the protocols of the so-called Liaison Conference and the Imperial Conference, historians follow-up in detail the process of decision-making and the resulting decision that led to the attack on Pearl Harbor (see Ike 1967 and Iriye 1999). The question they raise: “How was this momentous decision arrived at?” (Kershaw 2008: 334). Depending on which words one emphasizes, this question asks either for the prehistory or for the reason for this decision. In the first case, historians will cite Japan’s wars in China and French Indochina and the impact of Allied trade embargoes on the prospects for Japan’s hegemonic aspirations in the Pacific. Such an answer involves a series of causal explanations dealing with causal consequences of previous decisions which then narrow down the scope of reasonable future decisions. In the second case, we want to ascertain the purpose of the momentous decision: what the various members of the Japanese government hoped to achieve by making this move. As with the German decision regarding Dunkirk, the Japanese decision can seem puzzling. Four years after the decision, Japan found itself utterly devastated and completely dependent on the United States, and the sources indicate that many in the political and military leadership of Japan understood that chances of success for a war with the United States were low. On the other hand, there was also apparent consensus that Japan’s hegemony could not be achieved without access to oil and steel. Previous actions by Japan and the United States meant that Japan would soon run out of these supplies. Options for action and thus reason explanations are obviously restricted not only by further reasons but also by causes and states, for the shortage of steel made it not only unreasonable, but physically impossible to build, for example, aircraft carriers. However, despite unanimity over economic and military circumstances there were disagreements about the conclusion that should be drawn. The uncertainty of the war made some hesitant while others felt encouraged to trust to luck and to plan with best-case scenarios (see Kershaw 2008: 334 et passim). These differences led to numerous shake-ups in important positions. The hawks remained, and yet, as a result, those responsible for the crucial decision eventually came to a joint decision, a decision that each of them, including the hawks, had already declared unpromising or even hopeless (see Bix 2001: 420, Ike 1967: 238). On the face of it, this might seem to be a problem for the teleological account. On the teleological account, insofar as a behavior eludes rationalization altogether, it would not count as an action at all. And Japan’s

Rationalizing Principles  85 decision to go to war might appear entirely illogical: In order to protect Japan from the economic grip of the United States, they chose a means by which Japan was likely to perish altogether. That sounds pretty irrational, and the US Ambassador to Tokyo, Joseph Grew, even reported to his government in November  1941 that “Japanese sanity cannot be measured by American standards of logic. [. . .] Japan may go all-out in a do-or-die effort to render herself invulnerable to foreign economic pressure, even to the extent of committing national hara-kiri [.  .  .] Japan’s standards of logic cannot be gauged by any Western measuring rod” (Grew 1944: 469–470). But it would be hasty to take Grew’s statement literally. He does not really attribute a different logic to the Japanese; he does not, for example, claim that they did not accept the modus ponens (and such a suggestion might have been in the air after speculations sparked off by EvansPritchard’s work on the Azande [1937]). Had Grew not been able to explain to his government those Japanese considerations to which he had access, he would not have been the right person to fill his position. The difference between Japanese and American reasoning is not formal, and good historians will seek to make as much sense of it as they can. For example, Kershaw emphasizes that the decision to go to war was about “honour and national pride, the prestige and standing of a great power. The alternatives were seen as not just poverty, but defeat, humiliation, ignominy and an end to great-power status in permanent subordination to the United States” (2008: 380). Japanese policymakers might have put greater weight on the preservation of national honor than we would, but it is sufficiently rational to be taken seriously as a candidate action explanation. If the preservation of national honor is the primary goal Japanese policymakers aimed at before Pearl Harbor, then there is no tension between the belligerence, the historian Herbert Bix (2001: 417–419 et passim) attributes to the emperor, and the fact that Hirohito considered an armed conflict with the United States as “nothing less than a self-destructive war”, as the emperor was quoted as saying on July 31, 1941 (see Fackler 2014), just at the moment when the United States had imposed a total oil and gasoline embargo. And, moreover, the Japanese were by no means certain that they would lose the war, and there can be an understandable tendency to overestimate the probability of a desired outcome. Finally, in thinking about such complicated historical decisions, we should note that at no point does the historian investigate the causal properties of the brains of Japanese and Americans in order to see whether their mental states are realized differently or have different causal impacts. Throughout the process, one relies on rationalizing principles, thereby seeking to give the best sense to the behavior of the Japanese leadership without drawing on the more remote presumption of another logic.

86  Guido Löhrer and Scott R. Sehon

4. An Argument Against Causalist Accounts of Action Explanation The causalist about action explanation might not directly dispute that we use rationalizing principles in the process of determining which reason explanations are correct. After all, if the examples discussed here are at all illustrative, then the use of such principles is, in actual practice, both pervasive and indispensable. What the causalist will, and must, insist upon is that rationalizing principles are used as a sort of heuristic to find the physical causes of the behaviors in question. One might stipulate, as does Davidson himself, that in the explanation of behavior we seek causes that also rationalize the behavior; one might, with Marcus (2012), call these rational causes. But these rational causes are not some sort of cause in addition to the physical causes at work; we are still, according to the causalist, using rationalizing principles to identify the neural states (or whatever other sort of physical state the causalist might postulate) that physically caused the behavior in question. But all of this raises a very tricky question for the causalist: Why think that using rationalizing principles would work as a heuristic for finding the causes of a physical event? After all, in paradigm, uncontentious cases of causal explanation, we would never appeal to rationalizing principles. If we want to explain the motion of a billiard ball after a collision, we may refer to the mass and velocity of the incoming billiard ball, the elasticity of the collision, the conservation of momentum, etc.; at no point would we contemplate why it might be of value to the billiard ball to move in a certain direction. We certainly would not view it as a constraint on our explanation that it construes the billiard balls as rationally as possible. Moreover, it is not just that we don’t employ rationalizing principles as a constraint on normal causal explanations; if someone did attempt to use rationalizing principles to explain, for example, why a star went supernova, this would either be mere metaphor or a sign of great ignorance. In uncontentious cases of causal explanation, there is simply no reason at all to think that we could use rationalizing principles as means of identifying the correct of the events in question. Since even the causalist will likely accept the pervasive role of rationalizing principles in the case of the explanation of human action, it is incumbent on the causalist to explain why the situation is different here. If, in other causal realms, the application of such principles would at best be colorful metaphor and at worst be manifestly silly, why think that rationalizing principles will work as a heuristic in the case of finding the causes of human action, let alone a heuristic that is both pervasive and indispensable? There might be occasional cases in which a scientist has arrived at a good insight or hypothesis by thinking something of the form, “why would a black hole want to do that?” But no reasonable scientist would rely on this method as a way of ascertaining true causal

Rationalizing Principles  87 explanations in typical cases. So if the causalist indeed admits that we use rationalizing principles, even just as a heuristic, she has some explaining to do. Why expect this heuristic will have a greater than random chance of working to identify causes in this one particular realm? Without a good answer to this question, the causalist should, it seems, admit that the explanation of action answers to different principles and is fundamentally distinct from the causal explanation used in the sciences. The causalist might think that the answer to this challenge is very quick: We can be confident that using rationalizing principles will work because it is based on our empirical success with using such methods in the past. Of course, we would agree that we have used rationalizing principles in the past, and it has indeed successfully served our purposes, for we have thereby attained satisfying explanations of human actions, and we have been able to make predictions as well. But the causalist cannot merely appeal to this success; the causalist offering this explanation would need to give evidence that rationalizing principles have in the past correctly led us to the causes of actions. The non-causalist agrees that rationalizing principles have led us to correct explanations, but disputes that we have thereby found the causes of the actions, and we have noted that other examples of causal explanation give us no reason to think that rationalizing principles lead to correct causes. If the causalist is to appeal to the past success of rationalizing principles, then unless she is simply begging the question and assuming that causalism must be true, she must provide some independent way of confirming that rationalizing principles have led to causes. Such evidence will be hard to come by. We think it is fair to say that we don’t have any cases at all in which we have, without using rationalizing principles, confirmed a causal explanation of a human action in terms of a mental state or neural realizer thereof. Typically, our brains stay safely locked inside our skulls and are not subject to much direct observation, so we are rarely in the position to know what brain state causally produced a particular bodily motion. These days there are circumstances where scientists can, to a certain limited extent, see what is happening inside the brain. But such methods are of little use in confirming that rationalizing principles lead us to the right mental states as causes, for the brain states that we see do not come with labels (“desire to brush teeth” or “desire to start a trade war”). Perhaps the causalist’s idea is that agents themselves can simply see, by introspection, which mental state caused their behavior. Then we would be in a position to confirm that the rationalizing principles got the right answer, for they might correspond to what the agent introspected. We don’t doubt that agents can quite often have good introspective access to their reasons for action; indeed, if they lacked a reasonably high degree of transparency in this way, they would become less rationalizable, for they would either be clueless or self-deceived about their actions. But to

88  Guido Löhrer and Scott R. Sehon say we can introspect the causes of our behavior is blatantly questionbegging against the non-causalist. If we could simply see that our actions were caused by our mental states, then non-causalism of any form would be simply preposterous; there would be no debate, and books like this volume would not exist. In the next section, we turn to a more serious effort to explain why, on a causalist account, rationalizing principles would be expected to work as a way of finding causes.

5. A Possible Causalist Explanation: Evolution and Selectional Explanation Here is a possible explanation for why rationalizing principles can be expected to work as a heuristic for finding causal explanations: The process of natural selection has “designed” us to be basically rational beings. Evolution works in a way that is analogous to a computer programmer who programs a computer to play chess: If the programmer has done his or her job, then you will do reasonably well in predicting and explaining the computer’s moves by acting as if it is a good chess player who wants to beat you. Evolution, of course, did not literally design anything (hence the scare quotes in the first sentence of this paragraph), but one might argue that the appearance of design that results from evolution might license similar sorts of heuristics when applied to human action. Since the causalist needs to explain why rationalizing principles should be expected to work as a heuristic for finding correct causal explanations of human actions, any appeal to evolutionary theory must show two things. First is this: 1. Natural selection would explain the applicability of rationalizing principles in explaining human behavior. But that would not be enough. After all, the non-causalist fully agrees that rationalizing principles are applicable in explaining human behavior. The evolutionary causalist would have to go further and say: 2. Natural selection would justify using rationalizing principles as a heuristic for finding causal explanations of behavior. Let’s take these in the reverse order. We will first assume, for the sake of argument, that (1) holds. Granted that assumption, would evolutionary theory justify (2)? To explore this claim, it is worth addressing in more detail the analogy to the chess playing computer. To begin, it is worth noting something about the narrowness of the example. It is true that, for most of us, playing against a chess computer would pass an extremely limited version of the Turing Test: From the moves being made, we would not be able to tell whether we are playing a human being or a machine.

Rationalizing Principles  89 (This is evidently not true among expert chess players: Magnus Carlsen has been reported as saying that playing a computer is like playing an idiot who beats you every time.) Even so, this is still not a case of treating the chess playing program as if it were a rational agent in general; you don’t expect the computer either to feel pride at beating you or to feel sorry for you after you lose for the twelfth time in a row; you don’t expect the computer to have any plans for tomorrow afternoon or to have any beliefs about the weather, arithmetic, or the existence of God. So you treat the computer as a rational agent in a very limited sense: Bracketing all other things that would make for a life of a rational being, you simply act as if the computer knows how to play chess and wants to win. Within the narrow realm of playing a chess game, one might explain the computer’s last move by saying that it wants to protect its bishop or is hoping to catch you in a knight fork. However, the chess-playing algorithm, precisely because of its limitations noted in the previous paragraph, does not literally want anything and has no hopes. This is not to say that it would be impossible for any computer to have genuine desires or beliefs, just that we would not attribute genuine beliefs and desires to a mere chess-playing algorithm. Since the computer does not have actual beliefs and desires, the heuristic of treating the computer as if it were a rational chess player would not license causal claims about its behavior. When playing a computer, it may even be reasonable to act as if the computer has various desires—to control the center, to not exchange rooks for pawns, etc. But we would not say that we are thereby justified in making the following as a serious causal explanation: The computer’s move, e4, was caused by its desire to control the center and its belief that it could help to control the center by moving its king pawn to e4. Even if we have a justification for using rationalizing principles as a heuristic for understanding the machine’s chess moves, this does not imply that we are justified in using rationalizing principles as a method of finding correct causal explanations. Although the computer was designed to be a rational chess player, we would not use rationalizing principles as a heuristic for finding causal explanations of its behavior. This is just to emphasize that (2) is indeed a separate point over and above (1). Now let’s consider a more general case. Someday, computer programmers might go much further and program a fully functioning robot to act as if it is a rational being. Insofar as we had reason to think that the programmers had done their job well, we would thereby have reason to use rationalizing principles for the purpose of predicting and explaining the robot’s behavior. Even in the case of such a robot, this would not imply the analogue of (2); it would not imply that we have reason to use rationalizing principles as a heuristic for finding correct causal

90  Guido Löhrer and Scott R. Sehon explanations of the robot’s behavior. The chess program, after all, can be designed to act like a rational chess player, while it is nonetheless false that the computer has any desire-like state causing its behavior. For all we know, a robot with general artificial intelligence might use connectionist networks and machine learning, and there may be no state within the circuitry of the machine that can easily be identified with, for example, the belief that π is an irrational number, even though the robot is intelligent enough to affirm that π is irrational and to use this knowledge in inferences. So if we asked the robot, “is the square root of π rational?” we expect it to answer “no”, and our explanation will appeal to rationalizing principles: The robot knows that π is an irrational number and it knows (or can figure out) that the square root of any irrational number is also irrational. But none of this immediately licenses us in concluding that the robot has internal states such that the following is true, “The robot’s beliefs caused it to answer ‘no’ ”. Such a move would be justified if we could say something like this: The only way (or perhaps the best way, or the most plausible way) that one could program a robot to have artificial intelligence would be to program it such that it had states that could be identified with beliefs, desires, and intentions which caused its behavior. That seems to us a very substantive assumption that would require its own defense. Getting back to human beings and evolution, the analogous point is this: Even if our status as evolved beings would license using rationalizing principles as a heuristic, this by itself does not give the causalist what she needs in (2). As with the computers, it is one thing to say that humans act intelligently; it is another to say that we can therefore know that the behavior of the humans is caused by physical states that can be neatly identified with mental states. We can say that agents can believe propositions and desire states of affairs without thereby committing ourselves to the claim that the physical causes of behavior are identifiable with those beliefs and desires. (By analogy, we can truly say that a person is in a state of bankruptcy without thereby implying that this state will be something that scientists can pinpoint as the physical cause of certain behavior.) If the causalist simply asserts that (2) is true, that would seem to presuppose the rather substantive assumption that the only way (or only plausible way) that evolution could have “designed” us to behave rationally is to have made us as the causalist insists, namely, by creating us such that our reasons for action must be identifiable with states that actually cause our behavior. That’s tantamount to assuming the causal theory of action, and thus it would seem that this insistence simply begs the question. So even if we grant (1) above, the evolutionary causalist would still need further argument to get the needed claim in (2). But can the evolutionary causalist even establish (1)? Does the theory of evolution justify the expectation that human behavior will be in accord with rationalizing principles? The general story behind natural selection

Rationalizing Principles  91 is familiar. Variation occurs through heritable genetic mutations. If a genetic mutation allows an organism to have more offspring that survive, then that mutation is likely to come to dominate the population. In other words, traits that enhance reproductive fitness will be passed along. For organisms alive today, many of their traits are the result of the fact that these traits enhanced the reproductive success of their ancestors. Which traits enhance reproductive fitness? That will depend on the nature of the organism and its environmental niche. But within that context, the traits that have been selected for will typically be those that allow us to stay alive and healthy for a reasonable period of time—at least long enough to see to it that we have healthy offspring with a chance to likewise reproduce. To defend (1), the causalist appeals to evolution to explain why we would expect agents to act reasonably and rationally. The causalist would have that explanation if she could affirm the following for each application of rationalizing principles: 3. If rationalizing principles suggest that A’s reason for doing B was R, then evolutionary theory would predict that a disposition to act in accord with R in such circumstances would have been selected for— that a disposition to act in accord with R would have enhanced the reproductive success of the agent’s ancestors. If this were true, then this would explain the applicability of rationalizing principles. However, there are many cases in which a principle like (3) will look quite implausible. Consider, for example, a priest who intends to be celibate the rest of his life, or a couple who intends to use birth control and not have children. They will have reasons for those actions: The priest might hope to limit his earthly attachments so that he can be closer to God and better reflect Christ’s love for all humanity, and the childless couple might want to be unencumbered and have more time for hobbies and travel. Rationalizing principles could lead us to the conclusion that the behaviors of the priest and the couple are to be explained in this way. But a disposition to have no offspring would, in almost any circumstances, be bad for one’s reproductive success. So it is, to say the least, difficult to see how a disposition to act in accord with these reasons would have enhanced the reproductive success of that agents’ ancestors. These are extreme cases, where the actions seem positively detrimental to reproductive fitness. But there are also countless other cases in which a postulated goal will at best be neutral with respect to the agent’s chance of having healthy offspring. Consider an example from the history of philosophy (see Kühn 2001: 238–240). During a four- to five-month period of 1780, Immanuel Kant sat for long hours in front of his desk with the aim of writing a book; that’s the conclusion to which rationalizing principles would lead us concerning his behavior. (3) would require

92  Guido Löhrer and Scott R. Sehon that a disposition to write a book in such circumstances would have enhanced the reproductive success of Kant’s ancestors. This is, of course, absurd. Kant’s circumstances when he wrote the Critique of Pure Reason included being in his late fifties, and Kant had never had any thoughts of having children. In such circumstances, whether or not Kant finished the Critique of Pure Reason would have had no effect on whether copies of his genes came to dominate later populations. But the point is not specific to Kant’s circumstances. On the evolutionary timescale, books themselves are an extremely recent innovation, and even within these past few thousand years, it would be rather rare that the intention to write a book would somehow lead to more offspring. The evolutionary causalist can reply by retreating from (3) and noting that an evolved trait need not enhance reproductive success in all circumstances. Some dispositions might not be directly selected for but be the product of some further trait. Instead of (3), the evolutionary causalist might affirm: 4. If rationalizing principles suggest that A’s reason for doing B was R, then evolutionary theory would predict that either (i) a disposition to act in accord with R in such circumstances would have enhanced the reproductive success of the agent’s ancestors; or (ii) a disposition to have some other mental trait D would have enhanced the reproductive success of A’s ancestors, and mental trait D would have led the agent to act in accord with R. Consider a simple example where (4) is plausible. Suppose that Tom is going to the kitchen in order to get a piece of pecan pie. Clearly, the disposition to eat pecan pie is not a trait that was selected for in Tom’s evolutionary ancestors. On the other hand, it is plausible that a general disposition to like sweet food would have served those ancestors well, for they would have been encouraged to eat fruit when available. So Tom’s sweet tooth, which was selected for, might lead him to intend to eat more pecan pie, despite the fact that eating pecan pie will not positively affect Tom’s chances of reproductive success. This would be in line with (4): The disposition to like sweet things would be the other mental trait, D, and this in turn leads Tom to act in accord with the reason of getting a piece of pecan pie. But how would (4) help in the case of Kant writing a book? Which mental trait D will play the appropriate role? Of course, Kant probably did have some other desires beyond the immediate intention to write a book: the desire to respond to Hume, the desire to contribute to an important philosophical debate, the desire to convince philosophical colleagues of the view he was developing, perhaps even the desire to write a book that would be read by philosophers for hundreds of years.

Rationalizing Principles  93 However, such mental traits are similarly unlikely to have enhanced the reproductive success of Kant’s evolutionary ancestors, so a disposition to have such desires cannot serve the role of mental trait D in (4). Perhaps the causalist will posit a more general desire: the desire to be respected or approved of in one’s culture. It is not utterly fanciful that such a desire would enhance reproductive success, for those who are respected in their culture might have better luck at finding a mate; perhaps being respected would also help somewhat in successfully raising the resultant offspring to maturity. So, perhaps, a general desire to be respected by one’s peers could have been selected for. But this yields a rather puzzling picture when applied to the case of Kant and the writing of the first Critique. Rationalizing grounds lead us to conclude that Kant intended to write a book. By adopting something like (4), the evolutionary causalist would have us believe this: Kant had a general desire to be respected (a desire that was selected for), and this desire to be respected in turn led him to write a book. But we should ask about this last step, for a desire to be respected would not, irrespective of circumstances, immediately yield an intention to write a book (otherwise there would be a lot more people writing books!). Thus, the suggestion must be that Kant went from the general desire to be respected to the intention to write a book via some fairly deliberate practical reasoning. However, this looks rather implausible. Kant might well have denied having some general desire to be approved of, and even if he did have such a desire, Kant would probably have denied that this was his reason for writing the book. Kant surely did not think of himself as just figuring out what he could do in order to be respected by his peers. That would presumably seem a very crass motivation to the philosopher. If one nonetheless insists upon (4), this would have the consequence that Kant was significantly self-deceived: Despite his denials on the point, he had a general desire for cultural approval, and engaged in deliberate practical reasoning—of which he was entirely unaware—that led from this desire to the aim of writing a book. It might be of some comfort to Kant that, according to this evolutionary causalist view, all reasons for action ultimately trace to desires that somehow enhance reproductive success. But while Kant is no longer singled out, this might make the view seem even less plausible, for it means that nearly all agents are quite often selfdeceived: According to the causalist, our professed reasons for action will often be reached via deliberate practical reasoning of which we are unaware from general desires we might even disavow. Moreover, there is a further problem for the evolutionary causalist view. If Kant’s ultimate motivation was really the approval of his culture at the time, then he likely could have achieved this in a more efficient and productive way than by sitting in front of his desk for many hours and producing a book. Or at least that’s true for most of us writing books.

94  Guido Löhrer and Scott R. Sehon So, to make the account work, the evolutionary causalist needs to claim the following: • • • •

Kant had a general desire for the approval of his cultural peers. Kant engaged in practical reasoning to conclude that he could attain this approval by writing a book. Kant was nonetheless blind to the fact that there would have been better ways to achieve that general aim (especially given the Prussian state with its highly committed board of censors). All of this was subconscious: Kant might have denied the general desire, and certainly denied the alleged practical reasoning that gets him from the general desire to the specific aim.

So, to explain the applicability of rationalizing principles to human behavior, the evolutionary causalist postulates that people in general have rather sophisticated practical reasoning abilities, are rather stupid and self-deceived in the way they apply these abilities, but that all of this just happens to work out so that people are generally rational and reasonable. Perhaps one can consistently maintain all of that; we would not claim to have demonstrated that evolution cannot account for the applicability of rationalizing principles. But such a view faces substantial obstacles. We cannot simply assume that such an account will work somehow, especially if the account relies on very implausible psychology, psychology motivated only by the need to make the account work.

6. Conclusion History, insofar as it is the history of events in human culture, crucially involves the explanation of actions by individual people and by groups of people. It involves explanation by reasons. We have argued that reason explanation of action makes pervasive and ineliminable use of rationalizing principles, and thus historical explanation makes pervasive and ineliminable use of rationalizing principles. We have argued that this poses a prima facie problem for causal theories of action explanation. Insofar as the causalist allows that we use rationalizing principles to ascertain correct explanations of human action, she must claim that they function as a heuristic for finding the causes of human behavior. But why should rationalizing principles be an effective heuristic for finding causes in this realm when nobody would use them that way in any other realm? The causalist owes us an explanation here, and none seems to be forthcoming. Note that the question here is not: Why are rationalizing principles applicable to the explanation of human behavior? We have not claimed that the teleological account of action explanation has a better answer to this question than does the causalist. We have claimed that it is fundamental to the very project of explaining behavior in terms of reasons that we employ rationalizing principles, but we have offered no hypothesis as

Rationalizing Principles  95 to why this is so successful—why human beings are such that it works to prefer explanations that make the agent come out as rational as possible. We have taken rationalizing principles to be intrinsic to teleological explanation, but we have not derived them from some further, more fundamental, source. (There is an analogous mystery about the natural world in general: By the use of abductive inference or principles like Occam’s razor, we prefer explanations in the physical sciences that posit as few unexplained mysteries as possible. We know of no further metaexplanation for why this works as an approach to explanation in the natural sciences. It does not seem that the universe had to be this way.) The causalist might seize on this point and say that the teleological account is no better off than the causal theory, for neither can explain the basic applicability of rationalizing principles to the explanation of human behavior. But this is to misread the situation. The causalist needs to assume not only that rationalizing principles hold for human beings, but also that these principles work as a heuristic for finding the physical causes of human behavior, even though this would be a laughably silly method of finding causes in other contexts. The unexplained applicability of rationalizing principles to agency suggests, we think, that there is something sui generis about agency, something that is reflected in the fact that the explanation of action, including historical explanation, is irreducibly teleological.

References Bix, H. (2001). Hirohito and the Making of Modern Japan. London: Duckworth. Bloom, E. (2017). Trump Again Calls America ‘the highest taxed nation in the world’. CNBC Make it [online] July  26, 2017 ed. Available at: www.cnbc. com/2017/07/26/trump-claims-america-has-the-highest-taxes-in-the-world. html [Accessed Mar 20, 2018]. Bond, B. (1990). Britain, France and Belgium 1939–1940. Oxford: Brassey’s. Davidson, D. (1970). Mental Events. In: D. Davidson, ed., Essays on Actions and Events, Oxford: Clarendon Press, pp. 207–227. Davidson, D. (1963). Actions, Reasons, and Causes. In: D. Davidson, ed., Essays on Actions and Events, 1st ed. Oxford: Clarendon Press, 1980, pp. 3–19. Dennett, D. (1987). The Intentional Stance. Cambridge: MIT Press. DePaulo, B. (2017). I  Study Liars. I’ve Never Seen one Like President Trump. Washington Post [online] Dec 12, 2017 ed. Available at: http://wapo. st/2iDBdSh?tid=ss [Accessed Mar 20, 2018]. Eichenwald, K. (2017). Can Trump Tell the Difference Between Truth and His Lies? Newsweek [online] Feb 4, 2017 ed. Available at: www.newsweek.com/ trump-difference-truth-lies-552292 [Accessed Mar 20, 2018]. Evans-Pritchard, E. (1937). Witchcraft, Oracles, and Magic Among the Azande. Oxford: Oxford University Press. Fackler, M. (2014). Before Pearl Harbor, Emperor Warned against War. The New York Times [online] Sep 9, 2014 ed. Available at: https://nyti.ms/1pKWUcq [Accessed Mar 13, 2018].

96  Guido Löhrer and Scott R. Sehon Ghitis, F. (2017). Donald Trump is ‘gaslighting’ all of us. CNN [online] Jan 16, 2017 ed. Available at: https://edition.cnn.com/2017/01/10/opinions/donaldtrump-is-gaslighting-america-ghitis/index.html [Accessed Mar 20, 2018]. Grew, J. (1944). Ten Years in Japan: A Contemporary Record Drawn from the Diaries and Private and Official Papers of Joseph. C. Grew, United States Ambassador to Japan, 1932–1942. New York: Simon & Schuster. Ike, N., ed. (1967). Japan’s Decision for War: Records of the 1941 Policy Conferences. Stanford: Stanford University Press. Iriye, A. (1999). Pearl Harbor and the Coming of the Pacific War: A Brief History with Documents and Essays. Boston: Bedford/St. Martin’s. Kershaw, I. (2008). Fateful Choices: Ten Decisions That Changed the World, 1940–1941. London: Penguin Books. Konnikova, M. (2017). Trump’s Lies vs. Your Brain. Politico Magazine [online] Jan, 2017 ed. Available at: www.politico.com/magazine/story/2017/01/donaldtrump-lies-liar-effect-brain-214658 [Accessed Mar 20, 2018]. Kühn, M. (2001). Kant. A Biography. Cambridge: Cambridge University Press. Löhrer, G. (2006). Charakterstabilität und diachrone Kohärenz. Zurechenbarkeit im Prozess moralischen Umdenkens. Zeitschrift für philosophische Forschung, 60(4), pp. 528–553. Marcus, E. (2012). Rational Causation. Harvard: Harvard University Press. McLaughlin, T. (2017). For Americans Like me, Trump’s Questionable Relationship with the Truth Makes it Hard to Trust in His Presidency. Independent [online] Dec 2, 2017 ed. Available at: www.independent.co.uk/voices/donaldtrump-president-usa-lying-twitter-outbursts-government-distrust-a8084231. html [Accessed Mar 20, 2018]. Mele, A. (2010). Teleological Explanations of Actions: Anticausalism Versus Causalism. In: J. Aguilar and A. Buckareff, eds., Causing Human Actions: New Perspectives on the Causal Theory of Action, 1st ed. Cambridge: MIT Press, pp. 183–198. Mele, A. (2013). Actions, Explanations, and Causes. In: G. D’Oro and C. Sandis, eds., Reasons and Causes: Causalism and Anti-Causalism in the Philosophy of Action, 1st ed. London: Palgrave, pp. 160–174. Rothschild, M. (2013). No, Hitler Did not Let the British Escape at Dunkirk. Skeptoid [online] Apr 15, 2013 ed. Available at: https://skeptoid.com/ blog/2013/04/15/no-hitler-did-not-let-the-british-escape-at-dunkirk/ [Accessed Mar 20, 2018]. Sehon, S. (2016). Free Will and Action Explanation: A Non-Causal, Compatibilist Account. Oxford: Oxford University Press.

5 On an Imaginary Dialogue Between a Causalist and an Anti-causalist1 Giuseppina D’Oro

Davidson’s (1963) Actions, Reasons, and Causes brought the anticausalist consensus that had dominated the philosophy of action to an end with an argument aimed at establishing that explaining actions cannot be purely a matter of justifying them. He argued that it is not possible to account for the distinction between mere rationalizations and rationalizations that feature in the agent’s motivational set without appealing to causation. There are plenty of reasons for performing any given action and each of these reasons will justify/rationalize the action, but not all these reasons will be the reasons which motivated the agent to act and which truly explain the action. For example: I  may have accepted an invitation to speak at a conference in Dublin because it was on the theme of metaphilosophy—a theme close to my heart. Or I may have accepted the invitation because the conference was held in Dublin, a city I always wanted to visit. Both reasons rationalize/justify attending the conference, but which is the reason on which I acted? Davidson argues that there is a simple answer that is available to the causalist: The reason on which I acted is the reason which caused my action. Since anti-causalists claim (1) that explaining actions is a matter of rationalizing them and (2) that reasons are not causes, this solution is unavailable to them. As a result, anti-causalists are unable to make the important distinction between justifications (including ex post facto confabulations retrospectively concocted to present one’s actions in the best possible light) and genuine explanations. The causalist, by contrast, can distinguish between explanation and mere rationalization and claim that in genuine action explanations reasons are rationalizing causes of the action. Davidson’s construal of the distinction between justifying reasons and explanatory reasons has very important implications for what it is we are asking about when we enquire into an agent’s reasons for acting. When I returned from my conference in Dublin my partner Mark asked: “did you have a nice holiday then?”, implying of course that my real reason for going to the conference was its location and the theme of the conference was a mere rationalization which I made up retrospectively to justify leaving him in charge of the school run in rainy Stoke-on-Trent. So Mark

98  Giuseppina D’Oro has a very strong grasp of the distinction between reasons which can be invoked to justify my action and the reasons which may have actually motivated me. Now, if Davidson’s solution is to be accepted, it implies that when my partner Mark wonders to himself “did she go to the conference because she wanted to discuss metaphilosophy or because she wanted to visit Dublin?”, he is effectively asking “what reason caused her to go?” rather than “what reason makes most sense of her going to Dublin in the light of all the things I know about her?” In my view Mark is not asking “which rationalization made her move her body in the direction of the airport?”, as one might ask “what made the car engine start?” My partner’s question is about selecting the reasons which make most sense of my going to Dublin. If the question is construed in this way, then once Mark has reached a conclusion about the considerations which best rationalize my action, he has also reached a conclusion about the “real” reasons which motivated me to attend the conference: No additional appeal to causation is required. His question is about which, amongst a range of practical arguments, would be decisive for me, given the kind of person I am. It is not about what practical argument initiated the causal sequence which terminated with my landing in Dublin (for an account of how to disambiguate why questions see Tanney 2013). Now, suppose one construed Mark’s question as asking “what rationalization most compellingly accounts for my partner’s action?” rather than “what reason caused her to act?” On this construal of the question, the reasons which motivate me to go to Dublin are not different in kind from the reasons which justify my going to Dublin.2 There is, in other words, no distinction in kind between rationalizations invoked prospectively in the context of deliberation and rationalizations invoked retrospectively to justify actions (either by an interpreter or by the agent herself). This is not to deny that agents are prone to confabulation and that they may not, given the opportunity, cover up the prospective rationalizations which motivated them to act with different and more noble ex post facto ones which present their actions in the best possible light. It is rather to make the point that the prospective rationalizations which motivate the agent, and the retrospective rationalizations that are invoked by an interpreter (or by a confabulator), have the same logical form; rationalizations operate in the same way prospectively and retrospectively: by commanding an action as the rational thing to do. The reasons which prospectively motivate an agent to act (and which, according to the Davidsonian causalist, capture the agent’s “real” reasons for acting), are therefore not different kinds of reasons from those which are invoked to explain the actions retrospectively. Whether a rationalization is used prospectively by an agent or retrospectively by an interpreter, does not change the logical form of a rationalization. Reasons operate in the same way, i.e. they are species of justification, whether they are deployed prospectively for the purpose of deliberation, or retrospectively for the purpose of interpretation. Now, if

On an Imaginary Dialogue  99 reasons operate in the same way prospectively and retrospectively then one cannot say, as the Davidsonian causalist wishes to say, that the reasons which motivate an agent to act (prospectively) differ in kind from the reasons invoked to explain the action (retrospectively) because the former are rationalizing causes of the action, whereas the latter are mere justifications. The Davidsonian causalist believes there are two kinds of reasons: reasons which merely justify/rationalize, and reasons which also double up as causes. The latter genuinely explain the action because they are the reasons which cause the action. The anti-causalist position that I  wish to defend, by contrast, asserts that there is only one kind of (justifying) reasons used in different contexts: prospectively in the context of deliberation and retrospectively in that of explanation. Switching from one context to the other does not change the logical structure of a rationalization. Just as causal/nomological explanations have the same logical structure whether they are used prospectively (in the context of prediction) or retrospectively (in the context of retrodiction) so rationalizations have the same logical status (they are species of justification which command the action as the rational thing to do) whether they are invoked by the agent for the purpose of deliberation or by the interpreter for the purpose of understanding. It is important to note that the notion of justification invoked by the anti-causalist who denies there is a distinction in kind between explanatory and justifying reasons is neither moral nor epistemic; rationalizations (be they prospective or retrospective) justify neither the truth of the epistemic premise nor the moral acceptability of the motivational premise in the practical syllogism that the interpreter ascribes to the agent. When Mark tries to explain my action, he is not assessing whether the belief that the conference is in Dublin is veridical, and so whether the practical syllogism “There is a metaphilosophy conference in Dublin (epistemic premise); I want to hear metaphilosophical papers (motivational premise); therefore I  go to Dublin (practical conclusion)”, is sound. Nor is Mark assessing whether increasing one’s carbon footprint is desirable in the light of global warming and thus whether flying to Dublin is justified in the more robust sense of being morally justified. But while the notion of justification invoked in the explanation of action is neither moral nor epistemic, action explanation still is a form of justification because reasons command the action as the rational thing to do in the light of certain epistemic and motivational premises; they do not describe psychological processes. What explains my going to Dublin (on the view that the explanation of action is a form of justification) is neither my believing that the conference is held in Dublin nor my desiring to attend the conference, but the presence of a normative connection between the propositional content of the epistemic and motivational premises and the conclusion of the practical syllogism through which the action is explained. In fact, agents

100  Giuseppina D’Oro need not have silently recited any such practical arguments in order for their actions to be explained in this way.3 I defended the view that explanatory reasons are not different in kind from justifying reasons in D’Oro (2007) where I claimed that there are two distinct questions, which the Davidsonian causalist conflates. The first is a conceptual question: “What does it mean to explain something as an action?” The answer to this question is: it is to rationalize it, to provide reasons in favor of the action. The Davidsonian causalist and the anti-causalist agree on this point. The second is an epistemological question: “How do we know whether an agent really acted for a particular reason?” The anti-causalist answers this epistemological question by arguing that the real reasons why an agent acts are established through a hermeneutic process where one weighs one rationalization against the other and ascribes the agent the rationalization which makes most sense in a given context. The Davidsonian causalist, by contrast, answers the epistemological question by stating that the specific reason on which the agent acts is the reason which causes the action. I argued that the difficulties one may encounter in answering the second (epistemic) question should not tempt one to change the answer to the first (conceptual) question and claim, as the causalist does, that the agent’s real reasons are the reasons which caused the action. For to do so is to change the logical structure of action explanation. Mele (2013) has replied to this charge by arguing that the causalist is not guilty of conflating the conceptual and the epistemic question. What motivates the Davidsonian causalist to identify explanatory reasons with rationalizing causes of the action, Mele claims, is not an epistemic concern but an ontological one. The causalist, he says, is willing to concede the epistemic point that we may never know what the real reasons why an agent acts are. What the causalist is not willing to concede is the anti-causalist’s claim that the real reasons why an agent acts are not independent from the rationalization through which it is identified. The existence of a psychological process, for Mele, is what distinguishes real people from fictional characters. An account of action explanation that leaves these psychological processes out of the picture, he claims, treats real people as if they were fictional characters. In the following I put the bone of contention between Mele and myself in the form of an imaginary dialogue between a causalist and an anticausalist. In a nutshell my (anti-causalist) reply to the Davidsonian causalist is that the distinction between real people with real psychological processes and fictional characters with only make-believe ones is irrelevant to answering the question “what are x’s real reasons for acting?” Since reasons command actions, rationalizations necessarily present actions as obeying reason’s commands; they are not descriptions of psychological states of affairs. I do not deny that real people (unlike fictional characters) have real psychological processes. What I claim is rather that these psychological facts of the matter are not the kind of explanans that

On an Imaginary Dialogue  101 rationalizing explanations can invoke because the explanation of action is a normative affair. Thus, the attempt to re-state the Davidsonian challenge to the anti-causalist in ontological rather than in epistemological key does not succeed in showing that Davidson’s objection to the causalist is legitimate. Anti-causalists should not take Davidson’s bait. Rather than endorsing the concept of a rationalizing cause they should expose the Davidsonian challenge as failing to take seriously the implications of the normative character of reasons. Anti-causalist: If to explain an action is to rationalize it, then to choose between different action explanations requires choosing between different rationalizations, not between rationalizations and something else. If I asked you: which of these pears would you like? And you answered: I shall have the apple, then you would not have answered my question. Thus when my partner asks: “Now did you really attend the conference because it was on the theme of metaphilosophy or did you attend it because it was in Dublin?”, he is considering two practical arguments that could be ascribed to me to explain why I attended the conference. What he is asking is: “which goal (listening to metaphilosophical musings or visiting Dublin) is logically or rationally compelling”, not “which one of them is causally efficacious?” At least it seems to me that the question can be legitimately construed in this way, as asking not “which reason caused my action?” (as in “which electrical circuit caused the lights to go off?”), but “which practical syllogism makes most sense of my action?” And if this is the way in which the question is to be construed, then finding an answer to this question is not a matter of identifying what train of thought caused the agent to act. In fact agents may not have silently recited any internal monologues in order for the interpreter to reconstruct their reasons for acting by attributing a practical argument to them. Causalist: To stick to your example then: “How do you decide which pear to pick?” Or in other words: “How do you decide which rationalization should be ascribed to the agent?” If Mark is looking for a conceptual connection between the explanans (the motivational and epistemic premise) and the explanandum (the action), then he will find far too many since there are several different practical syllogisms which could rationalize the action. Davidson would grant that to explain an action is to rationalize it since he is adamant that the explanation of action has a normative dimension that has no echo in the sort of explanations at home in the physical sciences. But what he maintains is that the reasons which are genuinely explanatory are the causes of the action. It is this move that enables him to pick one pear (one practical argument) rather than another. If the real reason why you attended the conference in Dublin was your love of metaphilosophical reflection, Davidson would argue, then your thirst for metaphilosophical reflection caused (as well as rationalized) your visit to Dublin. If you are a causalist, you can pick

102  Giuseppina D’Oro this particular pear (meaning of course, practical argument). If you are an anti-causalist you cannot pick pears out of the bowl in this way because all pears (read: practical arguments) in the bowl will justify the action, but none of them will cause it. Anti-causalist: But the anti-causalist too has a way of distinguishing between generic reasons for acting and the reasons on which the agent actually acts. Maria could cycle to work to lose weight or she could cycle to be environmentally friendly. The more I know about a person’s profile (whether they care about their appearance, whether they care about their environment), the better I  will be able to choose between these different rationalizations. If Maria is a member of the green party, recycles religiously, and normally switches off all electrical appliances at night, then one might reasonably infer that she cycles to work because cycling is more environmentally friendly than driving, and that losing weight or keeping fit is not her reason for cycling to work. Simply stating that the reason why the agent acts is the reason which causes the action does not help in any way in determining what reason explains the action. The only factor which can help us determine which is the reason on which the agent acts is an account of whether the agent’s action coheres with certain goals. So if Maria’s cycling to work coheres with the goal of being environmentally friendly, cycling to work counts as a possible explanation of her cycling. If all other explanations are ruled out, then that possible reason will be the actual reason. But what makes us conclude that being environmentally friendly is Maria’s actual reason for cycling to work is not that it is the reason which causes the action but that it is the reason which best justifies it in the light of Maria’s overall profile. Saying that a reason explains an action if it is the one which causes the action does not actually help in any way in selecting the reason with greater explanatory power. And surely this is what we want to know: what reason is best suited to explain what Maria does. Going back to the original example, it does not help you to pick one pear rather than another. The claim does not do the work it is required to do. Causalist: But what if Maria were overweight and had a desire to lose weight and a thorough examination of the goals that her action might fulfill reveals that she might have indeed carried out the action to serve all these diverse goals? How does the anti-causalist decide between the different rationalizations? Maria’s motive could be losing weight as much as saving the planet. The causalist can say that the reason on which Maria acts is the reason which causes her to act. The anti-causalist cannot say that. Anti-causalist: I could use counterfactuals: Would she do it were she naturally thin? It is like my conference example: Would I have participated had it been held in Sheffield or Reading or at a hotel next to Milano Malpensa rather than in Dublin? We weigh one rationalization against another all the time quite effectively. When I say of my nine-year-old that

On an Imaginary Dialogue  103 she would not have come to the supermarket had she not known I was going to buy her a magazine, I  have determined what her motivation for coming to the supermarket is, without leaving the space of reasons and entering that of causes. Had she been willing to come along without being promised a reward, then I would have considered ascribing her different motives. I may not always be able to determine the real reasons in this way, but I can in most cases.4 Causalist: But surely there will be some cases where we cannot confirm or disconfirm the counterfactuals. In such cases appeal counterfactuals will not help establishing what an agent’s real reasons in acting are. Anti-causalist: Yes, I can see that there are such cases. If so, the anticausalist strategy will reach an impasse. But I do not think this is a big problem really, since the fact that the anti-causalist’s criterion for making the distinction between mere rationalizations and rationalizations which explain a particular agent’s action cannot be conclusive in every case does not entail that they are inconclusive in all cases. Causalist: My point is not that the anti-causalist does not have any criteria for making the distinction between explanatory and justifying reasons in most cases but that the causalist has a way of doing so at tie-break, when two or more rationalizations are equally compelling, whereas the anti-causalist does not. Thus causalism has an advantage over anti-causalism. So if you accept that there are some cases where the anti-causalist cannot come up with a definitive answer then you are effectively conceding defeat. Anti-causalist: I am afraid I am not willing to abandon the view that the explanation of action is a matter of justification simply because the criterion I apply leaves some cases indeterminate. Deontological theories reach an impasse when an agent is confronted with a conflict of duty. The agent cannot choose precisely because s/he cannot apply a deontological criterion to make the decision. The same could be said about utilitarianism. There will be situations in which the utilitarian principle will be unable to adjudicate between two actions which generate an equal amount of happiness. The anti-causalist is in a similar predicament: I cannot apply hermeneutic criteria to choose between two equally compelling rationalizations. True. In my view, however, this is not a good enough reason to abandon a commitment to a hermeneutic conception of explanation any more than the fact that the utilitarian principle cannot be invoked to arbitrate between two actions which produce an equal amount of happiness is a reason for forsaking utilitarianism. Causalist: It seems to me you are conceding defeat. Anti-causalist: Not really. What I am claiming is that you cannot settle the dispute as the Davidsonian causalist wishes to without changing the subject. To invoke the concept of causation to settle a hermeneutic dispute is to leave the space of reasons. Suppose that Serena Williams and Simona Halep had won an equal amount of games in a tennis set so that

104  Giuseppina D’Oro they had to play an additional game in order to determine who the winner of the set is. I put it to you that you could not ask them to play a game of football in order to decide who wins the tennis set. It would be equally wrong to settle a hermeneutic dispute about the meaning of an action by introducing a concept, that of causation, that does not respond to normative considerations. But this is precisely how the argument that it is the difficult cases (rather than the straightforward ones) which should settle the dispute between the causalist and the anti-causalist, seems to work. The causalist says that in order to decide between equally compelling hermeneutic explanations we need to make a move that does not belong to the game of the asking and giving of reasons. Reasons, as we have seen, command action. Explaining actions through reasons, therefore, requires presenting actions as abiding to the commands of reasons, not as caused by antecedent conditions, even if such antecedent conditions are psychological processes. Moreover, I am not sure you have an advantage over me: How do you know what the agents’ real reasons for acting are? I  doubt you would be able to see them even if you cracked their skull open. And if you cannot know what they are then you do not have an advantage over me. Causalist: I think you may have misunderstood the nature of my objection to anti-causalism. My worry is not epistemic. The point I am making is not that identifying explanatory reasons with the reasons which cause the actions, as Davidson suggests, will assist in determining which is the reason on which the agent acts, thus enabling us to know with absolute certainty, for example, that Maria’s reason for cycling to work is to save the planet rather than lose weight, or that you went to the conference because you wanted to visit Dublin rather than because you wanted to listen to some papers in metaphilosophy. The point I am making is that in your case there is a fact of the matter about why you went to Dublin, even if Mark may never find it out, whereas on your account there is actually no such thing. The question I  want answered is: “In virtue of what is it true that S acted for R?” not or not just “how can we tell that S acted for R?” (See Mele 2003: 39 and 51; and also Mele 2010: 195–196). On your account there are only practical arguments but no real psychological processes. In fact, one might go so far as saying that you interpret your daughter’s behavior as if she were a character in a Flaubert’s novel. You even interpret your own behavior as if you had to understand it from a third-person perspective. For the anti-causalist we are all fictional characters with no real psychological processes, just persuasive narratives made up by intelligent readers about our reasons for acting. But these narratives are only stories which, unlike real psychological processes, have no causal powers (see Mele 2013: 171–172). Anti-causalist: I am not sure whether I see a problem with the claim that I have to interpret my own actions. If this claim commits me to the view that I  am not infallible when it comes to understanding my own

On an Imaginary Dialogue  105 reasons for acting, I am very happy to concede that implication. I do not think that my knowledge of my own reasons for acting is incorrigible. Sometimes I know the reasons why other people act better than they do and vice versa. If the charge is that I do not endorse a Cartesian picture of the mind, I shall happily plead guilty to it. At any rate I think I am getting closer to understanding why you, and many others, have found Davidson’s challenge so compelling. The reason why many have found Davidson’s solution persuasive is that they think that in identifying explanatory reasons with the causes of the action Davidson provides a solution to the problem of mental causation which they claim is left unaddressed by the anti-causalist. Causalist: I do not think you have quite addressed my objection. My objection is not driven by a Cartesian view of the mind and a commitment to first-person authority. I am not arguing that agents have a privileged and incorrigible access to their reasons for acting. My point, as I just said, is that there is a fact of the matter about why (real) people act even when it cannot be known. There is something that makes the claim “Mark’s partner went to Dublin because of R” true even if we may never get to the bottom of the issue. Thus in my opinion there is a difference between your case and that of Flaubert’s character Madame Bovary. In Madame Bovary’s case if her actions are open to different interpretations, and the author does not write her true motivation in the text, then there is no fact of the matter about the true explanation for her actions, whereas in your case there is, even if it cannot be known. Your claim that the reasons which motivate agents to act are species of justification entails that there is no fact of the matter why an agent acted as they did. But there are thought processes which real people (unlike fictional characters) go through whether or not there is somebody to “read” or interpret them. And yes, I would also argue that it is these thought processes which are causally responsible for things happening in the real world. But my main point here is that in the case of real people, unlike that of fictional characters, there are such thought processes. Your interpretivism about action explanations ends up treating real people as if they were fictional characters. And this cannot be right. Consider the following scenario: At the end of the Battle of Trafalgar, Nelson went on deck in full admiral’s uniform, medals glistening in the sun. He was promptly shot and died not long after. Some historians think he wanted to die saving England in order to achieve everlasting glory. Others that he couldn’t face returning to England with the scandal currently breaking about his mistress Emma Hamilton. Others think he was just reckless and wasn’t expecting to get shot. But suppose this is what happened. Nelson had a temporary memory loss, forgot where he was, and thought he was going for an afternoon stroll in the park. Since he didn’t speak to anybody, give the game away etc., nobody

106  Giuseppina D’Oro will ever know the real reason Nelson went on deck. But surely it is perfectly coherent to suppose that there is such a reason. Are you denying this?5 Note the ingenuity of my objection, in case you missed it: in a real-life case, like my Nelson example, it seems obvious that there might be a mental event that nobody could ever know. In a fictional case, it is equally obvious that there couldn’t be. E.g. it couldn’t be a FACT that Sherlock Holmes acted in a certain way because he suddenly suffered a memory loss and believed he was going for a walk in the park, if Conan Doyle provides the reader with absolutely no reason to think that this happened. But it could be a fact that Nelson acted because he lost his memory for a moment and thought he was going for a stroll in the park, even though nobody could ever have any reason to think this. You draw the distinction between the rationalizations which motivate agents to act and other rationalizations in the manner in which Berkeley draws the distinction between ideas of imagination and ideas of sense, that is, by saying that both are types of ideas which are more or less coherent rather than by saying that ideas of perception represent something independent of them. By claiming that the rationalizations which motivate agents to act are species of justification your account leaves out an important ontological dimension of the problem, i.e. that there are real thought processes which real people have, and fictional characters do not have, unless of course they are written in by the author. Anti-causalist: I  shall come back to the question of mental causation later. For the moment let me comment on this alleged difference between real and imaginary people. I  am not convinced there is such a great difference between understanding fictional and real characters, not because I want to deny that whereas I exist, Sherlock Holmes does not really exist, at least in the weightier sense of existence, but because I think that we understand real people much in the same way in which we understand imaginary ones. When my daughter reads about the characters in Jacqueline Wilson’s novels she gets a good psychological training in how to understand real people. Moreover, is it not the case that we understand historical characters in the manner in which we understand fictional ones? What is the difference between understanding Cleopatra and Madame Bovary? What helps or hampers understanding is not whether the character is real or fictional but how much information we have about them. A  vaguely sketched character in a novel will remain opaque, not because it is fictional but because it is roughly described; a distant historical figure may be hard to fathom because the records are poor, even if it is a real rather than fictional person. You could of course say that at least in the case of real (and living) people one could check ascribed motives against their real thought processes by asking them: “What was your real reason for acting?”. But such checks can only be

On an Imaginary Dialogue  107 part of the process of ascertaining the reasons why an agent acts and cannot be decisive, any more than confession alone can provide uncontroversial evidence of guilt. I am not sure I would believe my daughter if she told me that she comes to the supermarket because she wants to help carrying the groceries. The fact that an explanation comes from the horse’s own mouth does not make it the “correct” explanation, not only because the agent might be confabulating, but because agents could be mistaken about their real motives. Suppose that, in order to put an end to his endless speculations about my motives, Mark confronted me with the question: “Did you go to the conference because it was on the theme of metaphilosophy or because it was in Dublin?” And suppose I answered: “I went to the conference because it was held in Dublin”. He could still choose to disbelieve me, suspecting that I am a self-deceiving workaholic hiding behind the facade of an easygoing person who likes to have some fun. Or suppose I answered: “I went because it was on a metaphilosophical theme”. Even if Mark believed that I genuinely rehearsed a piece of practical reasoning about the intellectual value of the conference and that I  did not explicitly or consciously think about my long-valued goal of visiting Dublin, he may still not be persuaded that my “real” reasons for acting had to do with the intellectual value of the conference. He might still suspect that, had the conference been in a less attractive city, I would have found my consciously rehearsed practical reasoning concerning the intellectual value of the conference less convincing.6 You think there is an important difference between real and fictional characters because, in your view, if two competing explanations of a fictional character’s action are equally plausible, there simply is no fact of the matter as to how best to interpret the character’s behavior, whereas in the case of real persons (so you say) there is a fact of the matter about their reasons for acting, even if it cannot be known. And this fact of the matter is a psychological process even if we may never know what this is, as in Nelson’s case. But if the explanation of the actions of real people were so different in kind from the explanations of the actions of fictional characters, why is it that we sometime still choose to disbelieve people’s avowed reasons for acting even when we believe they consciously entertained those reasons and did not deliberately seek to mislead us? If agents do not have the last word on the meaning of their actions, then being able to accurately represent an agent’s psychological process (assuming it did happen) cannot be the criterion we use when picking one explanation rather than another. Causalist: I said it before, and I say it again. My objection is not driven by the view that agents have privileged access to their own reasons for acting, which they know better than the interpreter. When I say that there is a fact of the matter about why an agent acted I  am not saying that the agent has incorrigible knowledge concerning this fact of the matter. All I am saying is that I am committed to the claim that there is a fact of the matter even if it cannot be known (by the interpreter or even by

108  Giuseppina D’Oro the agent) whereas you are committed to the claim that there is no such thing. You are committed to some form of ontological or metaphysical indeterminacy about what reasons an agent had for acting. Anti-causalist: Maybe so. Maybe I am committed to the view that the reasons why agents act are, or at least can be, indeterminate. Why should this be a problem though? Reasons do not exclude each other like causes, do they? I could have gone to the conference in Dublin both because of its theme and because it was in Dublin. My action could be genuinely rationally over-determinate. In this case there would not be a determinate answer to the question: “Did she go because she wanted to visit Dublin or because she wanted to listen to metaphilosophical papers?” But my reply to your worry about truth makers (“In virtue of what is it true that S acted for R?”) is that the presence of internal monologues, even granting they do occur (and whether or not they are incorrigibly identified by the agent as their reasons for acting) are simply not relevant to the correctness of the explanation. Note the ingenuity of my reply, in case you have missed it: I am not saying that there are no psychological processes but that truth, in the sense of adaequatio intellectus et rei, is not a criterion for the correctness of the explanation and thus cannot assist us in establishing whether “S acted for R” is the correct explanation and whether “R” is S’s “real” reason for acting. Once one rejects the idea that the criterion for the correctness of the explanation of action is correspondence to a psychological process (whether or not one is in a position to establish it), the difference between real and fictional characters becomes irrelevant to the task of determining what the correct/true explanation is. It is only if one accepts that correspondence (in the sense of adaequatio intellectus et rei) is the criterion to be used to determine which is the correct explanation that one has to deny (as you indeed do) that real people can be explained in the same way as fictional characters. Causalist: Likening the explanation of real people to that of fictional characters is a big chestnut to swallow . . . Anti-causalist: I am not so sure. To revisit your comparison between Berkeley’s arguments and mine: Berkeley claimed that we cannot draw the distinction between ideas of imagination and ideas of sensation by appealing to the fact that ideas of sensation represent something outside the mind whereas ideas of imagination do not. I claim that we cannot distinguish between the rationalizations which motivate an agent and other rationalizations by arguing that the former capture internal monologues whereas the latter do not. CORRECT. But the similarities between Berkeley’s argument and my own end here because my claim is not that real people do not have real psychological processes and thus that they are, to all extent and purposes, like fictional characters. My claim is rather that such psychological facts of the matter (even if one could know them) are not a criterion that can be invoked to select one rationalization instead of another because the explanation of action is

On an Imaginary Dialogue  109 normative, not descriptive. Since the explanation of action is normative, not descriptive (as Davidson himself would concede), psychological processes simply drop out as irrelevant—just as (dare I say it?) Kant’s transcendental object = x, or Wittgenstein’s beetle. And if psychological facts of the matter are explanatorily redundant then the distinction between real people and imaginary characters, important as it might be, provides no ammunition against the anti-causalist. Enquiring about the motives of real people is not an activity that is different in kind from that of enquiring about the motives of fictional characters in novels or films. The psychology of the fictional academic with a fictional daughter and a fictional obsession for metaphilosophy would have to be explained in exactly the same way as that of the real academic with a real daughter and a real obsession with metaphilosophical questions, even if one of them is real and the other is not. So, I rest my case. Davidson’s objection to anti-causalists is that they are unable to account for the distinction between justifying reasons and explanatory reasons without invoking the concept of causation. I initially construed Davidson’s objection as epistemic in nature, i.e., as stating that causalism is superior to anti-causalism because it is in a position to identify which, among a range of possible reasons, are decisive for the agent by invoking the concept of causation. I argued that this challenge is misguided because if one is to remain true to the concept of action, then choosing between one rationalization and another is a matter of determining which rationalization makes most sense of the action, not of determining which reasons caused the action (D’Oro 2007: 19–20). To invoke the concept of causation to settle difficult cases, I have argued, would be like asking Williams and Halep to play a game of football in order to decide the outcome of a tennis match. It was then suggested that I misconstrued the nature of Davidson’s objection. This objection, so it was claimed, is not epistemological, but ontological. On this ontological construal the objection states not that the anti-causalist is unable to determine which are the reasons that were determining for the agent, whereas the causalist can. The objection states rather that even if the psychological processes which caused the agent to act could never be known, in the case of real people, unlike that of fictional characters, there are psychological processes. The anti-causalist, so this reformulation of the Davidsonian objection goes, fails to distinguish between real people and fictional characters. Once again, I argued that this challenge is misguided: The appeal to psychological facts of the matter (whether they can be known or not) changes the nature of the explanandum under discussion. Reasons, as we have seen, command action. Explaining actions through reasons, therefore, requires presenting actions as abiding to reason’s commands, not as describing psychological processes. Real psychological processes, whether they can be known or whether they forever elude us, cannot be invoked as a criterion to sift real from possible reasons

110  Giuseppina D’Oro because it is in the nature of reasons to command rather than describe. The Davidsonian challenge to the anti-causalist, both in its epistemic and ontological incarnation, changes the nature of the explanandum and, in so doing, asks us to leave the space of reasons.

Notes 1 I am grateful to Scott Sehon, Guido Löhrer, Constantine Sandis, Fred Schueler, James Tartaglia, and Gunnar Schumann for their comments on this paper. 2 The view that action explanation is a species of justification was widely endorsed before Davidson. See Dray (1963) and Winch (1958). 3 This externalist position has been defended by Dancy (2000) and Alvarez (2010) amongst others. The target of my argument in this paper is the internalist position presupposed by Davidson. I have clarified how the notion of action explanation defended here sits in relation to the internalist/externalist debate in D’Oro (2017). 4 For an account of counterfactuals in the context of teleological explanations see Sehon (2005) and also Löhrer and Sehon (2016). 5 I owe this formulation of the causalist objection to my colleague, James Tartaglia. 6 I owe this point to Scott Sehon.

References Alvarez, M. (2010). Kinds of Reasons: An Essay in the Philosophy of Action. Oxford: Oxford University Press. Dancy, J. (2000). Practical Reality. Oxford: Oxford University Press. Davidson, D. (1963). Actions, Reasons and Causes. Journal of Philosophy, 60(23), pp. 685–700. D’Oro, G. (2007). Two Dogmas of Contemporary Philosophy of Action. Journal of the Philosophy of History, 1(1), pp. 11–26. D’Oro, G. (2017). The Justificandum of the Human Sciences: Collingwood on Reasons for Acting. Collingwood and British Idealism Studies, 23(1), pp. 41–66. Dray, W. (1963). The Historical Explanation of Actions Reconsidered. In: S. Hook, ed., Philosophy and History, 1st ed. New York: New York University Press, pp. 105–135. Löhrer, G. and Sehon, S. (2016). The Davidsonian Challenge to the NonCausalist. American Philosophical Quarterly, 53(1), pp. 85–96. Mele, A. (2003). Motivation and Agency. Oxford: Oxford University Press. Mele, A. (2010). Teleological Explanation of Action: Anticausalism versus Causalism. In: J. Aguilar and A. Bucareff, eds., Causing Human Actions: New perspectives on the Causal Theory of Action, 1st ed. Cambridge: MIT Press, pp. 183–198. Mele, A. (2013). Actions, Explanations and Causes. In: G. D’Oro and C. Sandis, eds., Reasons and Causes: Causalism and Anti-Causalism in the Philosophy of Action, 1st ed. Basingstoke: Palgrave Macmillan, pp. 160–174. Sehon, S. (2005). Teleological Realism: Mind, Agency, and Explanation. Cambridge: MIT Press.

On an Imaginary Dialogue  111 Tanney, J. (2013). Prologomena to a Cartographical Investigation of Cause and Reason. In: G. D’Oro and C. Sandis, eds., Reasons and Causes: Causalism and Anti-Causalism in the Philosophy of Action, 1st ed. Basingstoke: Palgrave Macmillan, pp. 124–140. Winch, P. (1958). The Idea of a Social Science and Its Relation to Philosophy. London: Routledge.

6 Reasons, Causes, Desires, and Dispositions Severin Schroeder

In his recent book Action, Knowledge, and Will, John Hyman offers a thoughtful and astute reassessment of the philosophical debate between Wittgenstein, who holds that explanations of human actions in terms of reasons are not causal explanations, and Davidson, who insists that they are (Hyman 2015, ch. 5). Hyman argues that both sides are mistaken in their commitment to a Humean account of causation.1 Only against such an unduly narrow account of causality—as a law-governed relation between events—are Wittgensteinian objections to a causal construal of reasons successful. Once we accept (1) that one can be immediately aware of causation, without having to observe regularities, and (2) that desires are dispositions, and dispositions can be causal factors too, causalism is safe from Wittgensteinian objections. Moreover, Hyman tries to defuse the problem of deviant causal connections by arguing (1) that it is just a special case of the general difficulty of distinguishing between manifestations of dispositions and mere side effects, and (2) that it doesn’t undermine the causalist construal: for whether its effects are deviant or not, citing a disposition explains them causally in either case (Hyman 2015: 116). I find Hyman’s account unconvincing for three reasons. I shall argue, first, that the deviant-causal-connection problem cannot be defused as easily as he suggests. Second, using words in their ordinary senses, intentions are not desires, nor are desires dispositions; but more importantly (granting for argument’s sake that intentions are desires of sorts, and desires are dispositions of sorts), dispositions are neither causes nor “causal factors”. Finally, Hyman’s account leaves out what is most crucial in Wittgenstein’s view, namely the observation that statements of one’s own reasons for acting are covered by first-person authority.

1.  The Deviant-Causal-Connection Problem A few days ago I bought a bike: a hybrid with alloy frame and 18 speed gear shifter. Why did I buy it? (R1) Because when it gets warmer I’d like to go on weekend cycling tours.

Reasons, Causes, Desires, and Dispositions  113 So I say, but perhaps you suspect that the real reason why I bought the bike was this: (R2) I wanted to support our local bicycle shop (from which I bought the bike). Or perhaps: (R3) I wanted to be seen as a bike owner by my neighbors in order to appear sporting and active. Davidson claims (and Hyman seems to agree [2015: 114]) that the “because”, picking out one of those reasons as the motivating reason, can only be given a causalist construal: The motivating reason must be the one that causes the action (Davidson 1980: 4, 9–12). However, it seems that we can easily imagine a case where such a reason is causally responsible for the action while not being the reason why I acted. For example, I casually mention my half-hearted intention to go on bike rides during the summer to an acquaintance who thereupon tells me of the economic difficulties of the local bike shop and persuades me to do something for that poor shop owner by buying a bike from him. Thus, my intention to go on bike rides, as mentioned in conversation, triggered a pitiful description of a shop owner’s plight, which then caused me, out of sympathy, to buy a bike. So my intention to go on weekend bike rides did cause the purchase, albeit indirectly. Hence, for an action to be done with a certain intention it is not sufficient that the intention cause the action. Clearly then, Davidson’s original question: “What makes a reason to do something the reason why it is done?” cannot be answered, or at least not fully answered, by saying that the reason must cause the action (since that can also be true when the action was not done for that reason). As far as this argument is concerned, it may still be held that the explanation “I bought the bike because I wanted to go on weekend cycling tours” identifies a cause of the action, but it cannot be said that that is all it does. There must be another element in acting for a reason than being caused to act by that reason. Hyman’s reply to this is that the relation between a desire and an action done because of that desire is just a special case of the relation between a disposition and a manifestation of that disposition. Thus, the “deviant” case described earlier (buying a bike) is like the following case: A man might take a soporific drug before driving, and the drowsiness induced by the drug might make him crash the car and knock himself unconscious. (Hyman 2015: 116) Here we can say that the drug caused him to lose consciousness, though not in the way it is supposed to do it.

114  Severin Schroeder Hyman seems to agree with Davidson that the “right” way for a reason to cause an action cannot be specified, but he insists that that does not undermine the causalist approach: there is no need to show how to eliminate the deviant causal chains between desires and acts in order to defend the claim that explanations of intentional action are causal explanations [. . .] I conclude that Davidson’s view that desires are causal factors is not cast into doubt by the impossibility of eliminating the “deviant” causal chains. (Hyman 2015: 127) The latter sentence may well be true. (In the following section I shall argue against the claim that desires are causal factors, but that has nothing to do with the phenomenon of deviant causal connections.) If A causes B only indirectly, via X and Y, it remains nonetheless true that A is a cause of B. However, that is not the issue. Rather, we want to know whether the meaning of the word “because” in “He ϕ-ed because of his intention to ψ” is entirely causal, or whether there is something else not captured by the causal claim. In the latter case it is not clear whether we should speak (as Hyman does in the first sentence quoted) of a causal explanation. In short, Hyman’s idea seems to be that although the “right way” for a desire to cause an action cannot be informatively defined, it can be understood to be the way in which a disposition causes its manifestation, which is just a particular causal pathway, but does not involve anything distinct from causation. Apart from the fact that I  don’t agree with Hyman’s dispositional analysis of desires (see section  2), there remain two problems for this proposed solution: First, it covers only one type of Davidsonian “primary reasons”: desires, but not beliefs. We may be persuaded that the desire to go on bicycle tours is a disposition, amongst other things, to buy a bicycle, but it does not follow that the same can be said about beliefs.2 Second, and more importantly, our understanding of the ways in which desires can cause actions is far too hazy to provide a useful analysis of the concept of a motivating reason. In particular, typically we are not really able to tell what the causal pathway was. But if, in fact, we are unable to tell whether a desire caused an action deviantly or directly, the causal analysis fails to provide a criterion by which to identify the motivating reason. The insufficiency of our knowledge of the causal impact of desires tends to be disguised by the fact that we focus on a few examples where the causal pathways, and the strength or weakness of particular desires are simply

Reasons, Causes, Desires, and Dispositions  115 assumed to be obvious. Earlier, I characterized my desire to go on cycling tours as half-hearted, but its verbal expression led to my being given another desire (to help a struggling retailer) that we just assumed to be operative in my action. But what if there is no such linear narrative leading from a patently weak to a patently strong desire, but instead two or three equally serious desires co-existing? I want to go on cycling tours, I want to support my local bicycle shop, and I also want to cut a dash with my neighbors. Suppose I can honestly name the first desire as my reason for buying the bike—does that mean that I know it to be the one that caused my action in the appropriate way? No. I can’t observe the causal pathways of my long-term wishes and preferences. For all I know, I might never have bought the bike had it not been for one of the other two considerations, even though I don’t regard them as weighty or respectable reasons. Suppose then that my desire to look sporting in front of the neighbors caused the purchase—did it cause it in a deviant way, just triggering what then became the main cause: my desire to go on weekend trips? Or was it the other way round: Was my desire to look sporting the main factor and the weekend trip idea just a necessary pretext, such that if for some reason weekend trips had not been an option I would have thought of some other respectable use for my bike? Who knows. How am I to know? Psychological causes are not like inner billiard balls whose paths can be followed by introspection. This may to some extent be the case with mental occurrences: I can fairly reliably report a succession of ideas: how hearing a familiar tune on the radio reminded me of an old friend— which memory in turn brings up a mental image of my school built by Arne Jacobsen—which then reminds me of the book on architecture I  just bought—which then makes me pick up that book. But Hyman’s concern is with dispositional desires, which are not mental occurrences and whose effects, therefore, cannot be read off their temporal position. Hence, if the criterion for something’s being a motivating reason were its causal role—not just that it caused the action, but that it caused it in a certain way—then that concept would be virtually inapplicable, at least where the candidates for the role of such a motivating reason are not conscious occurrences. In other words, the account in question turns the concept of a motivating reason into a philosophers’ pseudo concept— useless for real-life considerations and assessment of people’s reasons.

2.  Intentions, Desires, and Dispositions According to Hyman, intentions are desires, desires are dispositions (2015: 107), and dispositions are causes or causal factors (120). I find all three claims problematic. 1. My neighbor at table asks me to pass the salt and I do so. The Davidsonian analysis would be that my action was caused by a pair of

116  Severin Schroeder primary reasons: the belief that my neighbor would like me to pass him the salt and the desire to do what he would like me to do. To anybody whose linguistic sensitivity has not been blunted by reading too much philosophical prose this sounds slightly ridiculous, as the term “desire” is obviously a few sizes too big for the occasion. It is as if I described a mild dissatisfaction at seeing a pencil lying on the floor as “agony”. Hyman acknowledges that he and other philosophers use the word “desire” in an artificially broad sense (107), but he seems to regard that as a harmless extension of ordinary usage. It isn’t. For it neglects a crucial logical difference between desires (in the ordinary sense of the word) and philosophers’ desires. Desires proper are felt episodes, datable occurrences, and hence suitable items for a causal explanation of an action. But normally, when I politely pass the salt to my neighbor I have not experienced any desire or craving to oblige him. The only event that can plausibly be cited as a cause of my action is the utterance of his request. 2. It is for that reason, of course, that Hyman proceeds to construe the desire in question as a disposition. But that is not quite right either. Politeness may well be a disposition of mine that manifests itself in my passing the salt to anybody who asks for it, but such a disposition cannot be identified with the intention (in philosophers’ jargon: the “desire”) to oblige this particular person on this occasion. Moreover, I need not have a disposition of politeness in order to pass the salt on this occasion. Even ill-mannered people can be seen passing the salt when asked to do so. In civilized society, everybody passes the salt when asked to do so, so that it is not much of a manifestation of politeness. At most one could speak of a disposition not to be perversely rude or uncooperative. But even that won’t do, for an intentional action may well be out of character: not a manifestation of the person’s dispositions or character traits. An extremely selfish and stingy person may one day see a beggar and reach for his wallet to give away £50. The intention is to do something generous, but it is not the manifestation of a generous disposition. Again, Hyman is not unaware of this objection, but he doesn’t take it seriously. Perhaps, he wonders, such a specific intention (or “desire”) is too fleeting to be called a “disposition” in ordinary parlance (as it is not a character trait), but why should there not be “fleeting dispositions” (110)? Could one not be altruistic and generous for two minutes? Again, we can certainly decide to use the word “disposition” in an extended sense, but the price for that is explanatory emptiness. The point of explaining something by reference to a disposition is to present it as an instance of a regular pattern. When someone’s elaborate and old-fashioned politeness surprises you and makes you wonder whether he is being ironic or trying to flatter you before going to ask you for a

Reasons, Causes, Desires, and Dispositions  117 favor—it can be a useful explanation to be told that this is just his normal kind of behavior: no hidden agenda, he is just an exceptionally polite kind of person. That is the explanatory point of identifying an action as an expression of a disposition. And therefore dispositions (in order to have this explanatory value) must have a certain duration. To say that a one-off action (generously giving to a beggar) manifests a one-off disposition (momentary generousness) is not saying anything. In other words, invoking non-occurrent “desires”, construed as ad hoc “dispositions”, is a pseudo-explanation. Davidson, as is well-known, discussed and rightly rejected the socalled “logical connection argument” (1963: 13f.): the claim that A cannot cause B if they are not logically independent. That an event can be described as “fatally wounding X” logically implies that X was going to die, and yet it is perfectly true (though uninformative) to say that that event caused X’s death. The important point is that there are two distinct events, which could also be identified independently. The event of fatally wounding X could also be described, say, as “stabbing X with a pen knife”. However, as Rundle argued persuasively, the kind of “desire” or ad hoc “disposition” invoked to explain a humdrum intentional action, such as passing the salt, is not a distinct, independently identifiable event or factor (Rundle 1997: 12, 168). As the want (or “desire”) to do as asked is attributed to the agent solely on the strength of the action itself (which under the circumstances we easily recognize as intentional), there is no real basis for construing it as an independent mental entity and causal factor. It is simply a description of my action to say that I passed the salt in response to my neighbor’s request. In so far as such descriptions are justified solely on the strength of the observed pattern of behaviour, it is clear that the want does not figure as a hypothesized cause; we are speaking only of the character of the behaviour itself, not of a “distinct existence”. (Rundle 1997: 12) That is not to deny that my neighbor’s utterance (and my hearing it with understanding) may be said to have caused my action. What is to be rejected is only the idea that between (my perception of) the utterance— the obvious trigger of my action—and the action we need to postulate another cause (or causal factor): a “desire” or “disposition”. To think so—to invoke a “fleeting disposition” as a causal link between perceived request and action—, is to mistake the causal efficacy of the cause for a second cause. When a storm blows down a fence, the falling of the fence is caused by the storm. We don’t add anything to this causal explanation by saying that the fence was such that the storm would blow it down. This phrase (“the fence was such that the storm would blow it down”)

118  Severin Schroeder simply repeats the causal explanation (that the storm caused the fence to fall), but doesn’t specify a second causal factor (cf. Rundle 1997: 208). Of course, such a second causal factor could be specified by an independent description of the fence: as ramshackle or partly rotten. Similarly, it is perhaps possible to identify as a contributing causal factor in my passing the salt a desire to please my neighbor: Perhaps I want him to lend me some money and so I’m particularly anxious to make a good impression on him, to put him in a good mood over dinner etc. But here the desire is not just a philosophical postulate, but an independently identifiable psychological circumstance. I am very much aware of my eagerness to please my neighbor even before he asks for the salt. So, a desire (in the ordinary sense of the word) can indeed be a causal factor,3 beside the triggering event, but a “desire”, a postulated ad hoc “disposition” is not. It is simply a confused way of reiterating the fact that the triggering event did indeed have the causal efficacy to make me carry out this intentional action. 3. Finally, let us consider the case of a real disposition (not just a postulated “desire” misconstrued as an ad hoc “disposition”). As first pointed out by Ryle (1949: 117), to explain an action by reference to a disposition towards that kind of action is not a causal explanation. To say that somebody frequently scratches his chin is, obviously, not to give a cause of his chin scratching. Similarly, if somebody responds to some grave news with a silly joke, the comment that he just is a flippant and immature character does not name a cause of his frivolous remark, but merely describes it as part of a recurring pattern. He frequently responds to serious matters with a frivolous remark. It amounts to saying that one thing (grave news) regularly causes another (that person’s frivolous response); but the reliability of a causal link is not an additional cause. The closest Hyman comes to considering this Rylean objection is the following passage: it is sometimes said that explanations that refer to dispositions are vacuous or uninformative. [. . .] The complaint has some merit when the disposition mentioned is to produce the effect whose cause we want to know. Molière’s joke about the student who is congratulated for explaining that taking opium makes one fall asleep because of its virtus dormitiva—i.e. its soporific power—is a case in point. [. . .] But explanations that refer to dispositions are not all alike. For example, “Lead is poisonous because it is a neurotoxin” is a much better explanation than the one Molière lampooned because it excludes a larger range of alternatives. (Hyman 2015: 110–111)

Reasons, Causes, Desires, and Dispositions  119 However, this is not a reply to the Rylean concern at all. For one thing, the concern is not that explanations that refer to dispositions are uninformative, but merely that they are not causal explanations. More importantly, Hyman’s examples are not just explanations by dispositions, but also (proposed) explanations of dispositions. The causalist thesis he tries to defend is that dispositions can be cited in the explanans of the causal explanation of an action, whereas in these examples the explanandum is not an action, but is itself a disposition. To answer the question why Jones’s taking opium made him fall asleep by saying that opium has a soporific disposition, can be a perfectly respectable explanation. Molière’s question, however, is not about a particular event, but about a disposition: Why has opium a soporific disposition?—to which the answer “Because of its virtus dormitiva” is indeed vacuous—hence a joke—, as it merely repeats the explanandum in Latin. The explanation “Lead is poisonous because it is a neurotoxin” is of course more informative, but it is not a causal explanation either. It is informative by replacing a determinable predicate by a (more) determinate one, like “Smith is a member of the legal profession because he’s a barrister”. The explanatory link is not causal, but conceptual: being a barrister doesn’t cause you to be a member of the legal profession, it means that you are a member of the legal profession. Likewise, being a neurotoxin just means being a certain type of poison. A dispositional explanation is, roughly speaking, of the form: A ϕ-ed on this occasion, because A has a general tendency to ϕ. It is not a causal explanation: It does clearly not identify a cause of A’s ϕ-ing. Why then is it so often mistaken for a causal explanation? It seems to me that the main reason for that mistake is a failure to distinguish carefully between a disposition and its material basis; or perhaps an overhasty assumption that by naming a disposition one has implicitly identified its material basis (cf. Alvarez 2014: 86). The falsity of this assumption is the point of Molière’s joke. It is further facilitated by the fact that a disposition may be a disposition to cause something (e.g., being poisonous is a disposition to cause harm), so that an explanation naming such a disposition seems to imply causation and so may easily be mistaken for a causal explanation; especially when one uses the weasel word “causal factor”, instead of “cause”. One may be tempted to think that talk of a person’s general tendency to ϕ somehow refers to something that repeatedly makes that person ϕ. But that is not so. Although it is natural to think that a general tendency must have a cause, that is not implied (and the cause is certainly not identified) by mentioning the tendency. Observing that somebody tends to respond frivolously to serious matters is not the same as identifying— or even asserting the existence of—a (single) cause of that frivolous disposition.

120  Severin Schroeder Non-psychological dispositions are often known to result from a certain material basis. For instance, glass is well-known to be fragile, petrol is combustible, and aspirin is analgesic. Here we often switch from one to the other, or combine a dispositional with a causal explanation. For instance, when it is reported that: 1. The vase broke when dropped because being made of glass it was fragile. —the dispositional explanation is combined with a reference to its material basis (glass), which is appropriately regarded as a causal condition. If, however, all we are told is that: 2. The vase broke because it was fragile. —no causal explanation has been given. (2) only says that the thing did what things like that tend to do. And yet, even then it is natural to assume that a causal explanation is in the offing: In order to identify the relevant causal factor, underlying the disposition, we only have to inspect the object’s material or construction. Psychological dispositions, however, are very different. What is the material basis of frivolity, politeness, stinginess? We have no idea. We may speculate that there may be something in a person’s brain or genes causing the behavioral patterns in question, or perhaps, further back, a series of childhood experiences forming such character traits—but we don’t really know. With psychological dispositions, the material basis that might figure in a suitable causal explanation is not only not referred to nor implied by talk of dispositions, it is moreover something that currently we are entirely unable to discover or identify. Therefore, a dispositional explanation of human behavior is not only not itself a causal explanation; it is also, for the time being, utterly unrealistic to regard it as a signpost referring us to a corresponding causal explanation.

3.  Reasons and First-Person Authority At the core of Wittgenstein’s distinction between reasons and causes lies the idea that the grammar of the former is characterized by first-person authority: a person does not generally know the causes of his activities. [. . .] he will frequently be mistaken in specifying the cause. Strangely enough, he cannot be mistaken in specifying his reason. [. . .] That is, we call the reason that which he gives as his reason. The cause of an action is established by observation, namely hypothetically, i.e., in such a way that further experiences can confirm it or contradict it. (Wittgenstein 2003: 109–111)

Reasons, Causes, Desires, and Dispositions  121 Hyman sees in this (or a related passage in the Blue Book; Wittgenstein 1958) only the claim that causal explanations are conjectures, whereas we are immediately certain of our reasons—to which he rightly objects that causal statements can be certain and immediate too (2015, 118). In fact, that is something Wittgenstein himself remarked and illustrated by the example of being startled by suddenly seeing a light: in which case one does not hypothesize on the basis of repeated observations about the cause of one’s bodily movement, but is immediately aware of that cause (Wittgenstein 1993: 408). So it is indeed inaccurate to say that “one can only conjecture” the cause (as Wittgenstein does in 1958: 15), but that is not his main point. The crucial contrast is not between immediate knowledge and mere conjecture based on repeated observation, but between, on the one hand, an authority conferred by our concepts (hence not really a matter of knowledge at all) and, on the other hand, any kind of empirical finding, be it immediate perception or a tentative conjecture. In the latter case there is an independent state of affairs of which I come to know, be it step by step through repeated observation or immediately. Either way, my cognition is fallible. Error may be highly unlikely, practically impossible, but extraordinary circumstances that would prove my judgement erroneous are at least conceivable. Thus, it is at least imaginable that the car I see right in front of me in broad daylight proves to be an ingeniously constructed hologram. Similarly, when I  suddenly see a light or a face and startle, it is conceivable (however far-fetched and unlikely) that my startling reflex was triggered by some electrodes in my brain quite independently of what I happened to see at that moment. By contrast, first-person authority does not underwrite reports about independent occurrences, but marks certain kinds of sincere avowals of a person’s impression as constitutive of a certain concept. Thus, when I  can sincerely complain that something feels painful to me, then—by definition or by grammar—it is painful. This concept of feeling does not allow any room to a distinction between appearance and reality; that is, error in this case is logically impossible. (Not because I know so reliably whether I’m in pain, but simply because this is not really a case of knowledge at all; cf. Wittgenstein 2009: §246.) Similarly, according to Wittgenstein, there is a concept of a person’s reason that is simply defined, or constituted, by that person’s sincere avowal of his reason. Not because we take people to be so reliable at finding out what their reasons are, but because a “reason” (in this sense), unlike a cause, is not an independent occurrence at all: It is simply defined to be what an agent can honestly say in response to the question: “Why are you doing it?” or (with regard to the recent past) “Why did you do it?” The point of such a concept is easy to see. An agent’s proffered reason will give us an insight into his character. It tells us what considerations he regards as justifying the action in question (at least in a weak sense of

122  Severin Schroeder “justify”: as making the action understandable from the agent’s point of view), or would so regard, given the information and interests he had at the time. Assuming that people’s general views and dispositions remain fairly stable over short periods of time, we can generally trust people to be reliable in expressing subsequently what they would have been able to say at the time of action. Anyway, the justificatory aspect of explanations in terms of reasons is of paramount importance to us. By asking people to give reasons for their behavior, we challenge them to justify it; to tell us (if they can) why it wasn’t a bad (or silly) thing. The question of when this justification was (or would have been) thought of for the first time may be quite irrelevant.4 Wittgenstein’s account of the concept of a person’s reason for acting is simply an abstraction from the language game of giving one’s reason, that is, of justifying one’s behavior. So it is not surprising that it offers a straightforward solution to the problem that defeated Davidson: the problem of finding criteria by which to identify the reason for which somebody has done something.5

Notes A similar position has been taken by Moya (2014: 200–204). 1 2 For reasons not to regard beliefs as dispositions, see Hacker (2004: 202–219). 3 It has been objected that an action caused by a desire could no longer be described as voluntary, let alone intentional (Melden 1961: 128–129; White 1967: 148; Hacker 1996: 580). However, on this point I am inclined to side with Davidson, who queries: “Why on earth should a cause turn an action into a mere happening and a person into a helpless victim?” (1963: 19). There is nothing in our concept of a voluntary action that would exclude the idea of its having causes. Besides, it is far from obvious that a cause must always necessitate its effect. As Anscombe has pointed out (1981), we frequently make causal statements without any implication of necessity. Likewise, we might describe someone’s action as brought on by a sudden desire without denying that he might have resisted that desire. A red traffic light causes me to stop, but my stopping does not on that account cease to be a voluntary action. Even if my decision to stop is the result of rational considerations, a pedestrian who pushed a button to make the lights change may still observe with gratification that his action caused me to stop. 4 For a further discussion of Wittgenstein’s account of a reason, see Schroeder (2017). 5 I am grateful to Gunnar Schumann for his comments on an earlier draft of this paper.

References Alvarez, M. (2014). Ryle on Motives and Dispositions. In: D. Dolby, ed., Ryle on Mind and Language, 1st ed. Basingstoke: Palgrave Macmillan, pp. 74–96. Anscombe, E. (1981). Causality and Determination. In: E. Anscombe, ed., Metaphysics and the Philosophy of Mind: The Collected Philosophical Papers of G. E. M. Anscombe. Vol. 2, 1st ed., Oxford: Blackwell, pp. 133–147.

Reasons, Causes, Desires, and Dispositions  123 Davidson, D. (1963). Actions, Reasons, and Causes. In: D. Davidson, ed., Essays on Actions and Events, 1st ed. Oxford: Clarendon Press 1980, pp. 3–19. Hacker, P. (1996). Wittgenstein: Mind and Will: Volume 4 of an Analytical Commentary on the Philosophical Investigations. Oxford: Blackwell. Hacker, P. (2004). Of the Ontology of Belief. In: M. Siebel and M. Textor, eds., Semantik und Ontologie, 1st ed. Frankfurt: Ontos Verlag, pp. 185–222. Hyman, J. (2015). Action, Knowledge, and Will. Oxford: Oxford University Press. Melden, A. (1961). Free Action. London: Routledge. Moya, C. (2014). Les Aspects causaux de l’explication de l’action. In: R. ClotGoudard, ed., L’explication de l’action: Analyses contemporaines, 1st ed. Paris: Vrin, pp. 187–206. Rundle, B. (1997). Mind in Action. Oxford: Clarendon Press. Ryle, G. (1949). The Concept of Mind. London: Hutchinson. Schroeder, S. (2017). Reasons and First-Person Authority. In: J. Padilla Gálvez and M. Gaffal, eds., Intentionality and Action, 1st ed. Berlin: de Gruyter, pp. 123–138. White, A. (1967). The Philosophy of Mind. New York: Random House. Wittgenstein, L. (1958). The Blue and Brown Books. Oxford: Blackwell. Wittgenstein, L. (1993). Cause and Effect: Intuitive Awareness. In: J. Klagge and A. Nordmann, eds., Ludwig Wittgenstein: Philosophical Occasions 1912– 1951, 1st ed. Indianapolis: Hackett, pp. 370–426. Wittgenstein, L. (2003). Reason and Cause. In: G. Baker, ed., The Voices of Wittgenstein: The Vienna Circle, 1st ed. London: Routledge, pp. 109–111. Wittgenstein, L. (2009). Philosophical Investigations, 4th ed. Oxford: Wiley-Blackwell.

7 Objectivism and Causalism About Reasons for Action Hans-Johann Glock and Eva Schmidt

This chapter explores whether a version of causalism about reasons for action can be saved by giving up Davidsonian psychologism and endorsing objectivism, so that the reasons for which we act are the normative reasons that cause our corresponding actions.1 We address two problems for “objecto-causalism”, actions for merely apparent normative reasons and actions performed in response to future normative reasons—in neither of these cases can the reason for which the agent acts cause her action. To resolve these problems, we move from objecto-causalism to “objecto-capacitism”, which appeals to agential competences manifested in acting for a reason. We briefly apply this view of reasons for action to historical action explanations.

Introduction: Explanation in Theory of Action and Philosophy of History Philosophy of history as traditionally understood deals with the same subject matter as historiography itself. It differs only in the breadth of its aspiration, namely of providing a comprehensive account of the historical process in its entirety. In academic circles this endeavor is generally derided as “speculative”. Since the turn of the 19th century it has been replaced by a different type of philosophy of history. It is a second-order form of inquiry which reflects on the conceptual and methodological presuppositions of first-order historiography (historical narrative, explanation, and judgement). The roots of this second genre lie in the hermeneutic tradition. But methodological philosophy of history reached new levels of conceptual sophistication and dialectic acuity within analytic philosophy of the mid-20th century. Since the 1970s, it has suffered an undeserved decline. The current volume is an overdue attempt to reverse that trend. Our essay does not directly contribute to analytic philosophy of history: It is not an exercise in the methodology of the historical sciences. At the same time, we address a phenomenon that is

Objectivism and Causalism  125 fundamental to both human history and its academic retelling. Historiography deals to a large extent with the actions of human agents. It describes these actions not bio-physically as bodily movements; and it does not confine itself to describing their consequences. Instead, even descriptive or narrative histories pursue an explanatory agenda. They try to make sense of actions by specifying reasons for which agents performed them. Chamberlain signed the Munich agreement because he believed that it would guarantee “peace in our time”; the ANC promoted the armed struggle in the 1960s, since its leadership saw no prospect of overturning apartheid by nonviolent means; Merkel kept the borders open to refugees in 2015 in order to avoid a humanitarian catastrophe; etc. For the purposes of this paper we call explanations of actions which appeal to the reasons for which agents perform them “action explanations”.2 Our focal question is: What is the nature and status of action explanations? This question is central both to the theory of action and to the philosophy of history. Indeed, contemporary debates about explanation within the theory of action have an important source in the methodology of the historical and social sciences. The rise and professionalization of historiography in the 19th century gave rise to two contrasting paradigms. On the one hand there was the Humean project, later pursued by positivism and logical empiricism, of emulating the hard, natural sciences. J.S. Mill gave emblematic voice to this vision: “The Science of Human Nature may be said to exist, in proportion as the approximate truths, which compose a practical knowledge of mankind, can be exhibited as corollaries from the universal laws of human nature on which they rest” (1911: 6.3.2). On the other hand, there was the hermeneutic tradition which reached methodological maturity with Dilthey. It insisted that the explanations of actions in the social and historical sciences is a sui generis form of interpretative “understanding”, fundamentally distinct from the causal “explanation” furnished by the natural sciences. This hermeneutic Erklären-Verstehen dichotomy was supplemented by a methodological tenet of the South-West School of Neo-Kantianism (Windelband, Rickert): Natural sciences are “nomothetic”, i.e. concerned with general laws; social/historical sciences are “ideographic”, i.e. concerned with individual cases. Both ideas—that historiography explains in an interpretative rather than causal way and that it deals with individual cases agents, actions, and events—came together in Collingwood’s eloquent plea for the independence of historiography from natural science (1946). We concentrate on the former idea. And we do so by means of engaging with state-of-the-art debates about what causal aspects, if any, explanations of action possess. The final part explores some implications of our results for the nature and status of historical explanations.

126  Hans-Johann Glock and Eva Schmidt

Motivating, Explanatory, and Normative Reasons In the following, we will rely on the distinction, widely accepted in the literature, between normative reasons, motivating reasons, and explanatory reasons. Normative reasons are reasons for an agent to perform an action, i.e. facts or considerations that favor her acting in this way. Motivating reasons are reasons for which an agent performs an action, or in other words considerations in the light of which the agent acts, considerations which make performing the action attractive from her point of view. Explanatory reasons are reasons why the agent acts a certain way, that is reasons which explain her acting. Let us briefly elucidate these distinctions by providing an example. Yuki closes the window because it’s freezing outside. Yuki’s action—her closing the window—is explained by appeal to the fact that it’s freezing outside, so this fact is an explanatory reason or reason why she performs that action. That it is freezing outside is also a consideration which makes it attractive for her to close the window, and so it is a motivating reason/a reason for which she closes the window. Finally, the fact that it is freezing outside is also a reason that genuinely speaks in favor of her closing the window, it is a reason for her to close the window. Reasons in this sense have a normative dimension, for they contribute to the goodness or rightness of an action, and are thus typically called “normative reasons”. To simplify matters, we will follow Dancy (2018: 29–30) and Parfit (2011: 38) in limiting our discussion of normative reasons to fully objective normative reasons, that is, to actual facts that genuinely count in favor of the relevant actions.3

The Identity Thesis and Causalism One important question with respect to motivating reasons and the actions performed for these reasons is how the two are related. The answer we will focus on is that the relation between actions and their reasons is a causal relation. Philosophical orthodoxy, going back to Hume and epitomized by Davidson (1980a: 4), endorses this claim: The agent’s reason for performing an action is a combination of pertinent beliefs and desires that explain the action by showing the positive light in which the agent saw it; and this combination causes the action.4 Applied to our example, the view is that Yuki’s belief that it is freezing outside, her desire to keep the apartment warm, and her belief that closing the window will keep the apartment warm are her motivating reasons and cause her to close the window. We call this view “causalism about reasons” (Glock 2014a: 14). As the attentive reader will have noticed, on the orthodox picture causalism about reasons is combined with subjectivism or psychologism about the reasons for which the agent acts: These reasons are mental phenomena causing physical movements.

Objectivism and Causalism  127 But how do motivating reasons, as conceived by the orthodoxy, relate to normative reasons? The fact that it is freezing outside is a normative reason for Yuki to close the window, and thus an objective phenomenon— such as a fact, a state of affairs, or a situation rather than a mental event or state of the agent5 (See Smith 1994: 95; Dancy 2000: 3). The result is a bifurcation of reasons, since motivating reasons as mental and normative reasons as facts fall into different ontological categories (Smith 1994). Dancy (2000: 103) has argued that the bifurcation conflicts with the socalled “identity thesis” (see Alvarez 2017). According to this thesis, it must in principle be possible that the reasons for which an agent acts are among the facts that favor acting in this way; it must be possible that an agent’s motivating reasons are identical with the relevant normative reasons. In our example, this possibility is actualized: That it is freezing outside is a normative reason for Yuki to close the window; at the same time, she closes the window for the reason that it is freezing outside. The normative reason and Yuki’s motivating reason are one and the same. If the identity thesis is correct, it cannot be right that motivating reasons are combinations of mental states that cause an agent to act accordingly, whereas normative reasons are worldly facts, for this would preclude the possibility of identity. Combinations of mental states are the wrong sort of entities to be identical with worldly facts. And giving up the claim that normative reasons are features of the agent’s objective situation is not an option. So psychologism cannot be correct, and we are driven towards non-psychologism or, as we will say, objectivism about motivating reasons, the view that they are facts, purported facts, or possible facts (states of affairs).6 Rejecting the bifurcation between motivating and normative reasons leads Dancy to claim that motivating reasons are considerations in the light of which an agent acts, i.e., not her beliefs or other mental states, but what she believes, (purported or possible) facts. In other words, motivating reasons are not an agent’s mental state of believing, or her coming to believe; rather, they are the contents of her believing, what she believes. And if she believes truly, these contents are facts, what is the case. We have to put ourselves in the agent’s shoes and focus on what weighed with her as she made up her mind to act a certain way; this will be the considerations that persuaded her so to act. For instance, Yuki closes the window in light of the consideration that it is freezing outside. Such a consideration, where it is the case, may be the very fact that favors the agent’s action; objectivism about motivating reasons is therefore compatible with the identity thesis. Rejecting the bifurcation claim further motivates Dancy to deny causalism about reasons—the claim that the relation between the reasons for which the agent acts and her action is a causal relation. Dancy (2000, ch. 8) also rejects causalism about action explanation. The objectivist’s motivating reasons can be used to explain an agent’s action.7 For instance, that it is freezing outside, as Yuki believes, explains why she

128  Hans-Johann Glock and Eva Schmidt closes the window. What is contentious is how these reasons manage to explain the agent’s action: Is it a matter of the reasons causing the action, as claimed by causalism, or do we need a completely different account of reasons explanations? According to Dancy (2000: 163), reasons explanations are not causal, but normative. He holds that when we explain an action by adopting the perspective of the agent and figuring out which were the considerations in the light of which she acted, we do not try to figure anything out about the causal relations between these considerations and her action. We simply try to determine which were the considerations in the light of which the agent acted as she did, and there is nothing more to be said about the matter. We have to take normative explanations—explanations that appeal to the considerations in the light of which the agent acted, which make the action attractive in the agent’s eyes—as sui generis. The best we can do to explain an action is to understand what, from the point of view of the agent, favored or at least seemed to favor the action. Accordingly, explanations of an action that appeal to the reasons for which the agent acts are not causal explanations; they appeal to the (apparent) “normative situation” of the agent—hence the label “normative explanations”. Since the normative situation of the agent may be merely apparent, it can happen that we explain an action by something that is not the case, as when Yuki closes the window for the reason that it is freezing outside, even though it is really quite warm outside. This goes to show that normative explanations are not factive—they do not entail that the explanans is the case. But causal explanations are factive. Nothing can cause an action (and thereby explain it) if it’s nonexistent or does not obtain.8 So, Dancy concludes, the relation between the reasons for which and the action is not causal, and normative explanations are not causal explanations. The availability of different kinds of action explanation might suggest a pluralist account: In addition to normative, admittedly noncausal explanations in terms of reasons-for-which, there are causal explanations of an action which are different from yet in harmony with the normative ones. This proposal combines a rejection of causalism about reasons with an endorsement of causalism about action explanation—motivating reasons are not themselves causes, yet action explanations make reference to efficient causes associated with reasons (see Glock 2014a: 15). Here is one version of pluralism.9 Sure, the relation between an action and the reasons for which it’s performed is not causal—the contents of mental states do not stand in causal relations to actions. Nevertheless, given that the agent’s reasons are her reasons, which persuade her to act, mental states which have these considerations as their contents and which plausibly cause the agent’s action have to be part of the causal picture. And even if we refrain from calling them “motivating reasons”, these mental states can still be appealed to in an explanation of the

Objectivism and Causalism  129 action. To put it differently, there is no irreconcilable conflict between what Dancy has to say and what is at the heart of causalism. There are noncausal action explanations in terms of motivating reasons in Dancy’s sense which render the action intelligible from the perspective of the agent. Ascribing these reasons to her by taking that perspective is a case of the Verstehen emphasized by the hermeneutic tradition. But in addition, there is also a causal relation, not just between neurophysiological phenomena and bodily movements, but between the agent’s mental states that have motivating reasons as their contents, and the relevant action. Moreover, these mental states provide one type of causal explanation of why the action was performed, precisely the kind of causal Erklärung that historians, though not neuropsychologists, are interested in. Yuki’s reason for closing the window is that it is freezing outside, and that reason renders her action intelligible. But at the same time, Yuki’s believing that it is freezing, her desiring to keep the apartment warm, and her believing that closing the window will keep the apartment warm (taken as mental states, not their contents) causally explain why she closes the window. Dancy (2000, chapter 8; 2004) rejects this pluralist proposal. According to him, normative and causal explanations compete and are mutually exclusive, so they cannot be combined into one pluralistic picture. His key points against pluralism are that, first, the two explanations try to give different answers to the very same question (viz. why did she do it?), and that, second, they appeal to the very same features in the explanans, the mental state and its content, while giving mutually exclusive accounts of which feature does the explanatory work. In particular, psychologism ascribes central explanatory relevance to the mental state as motivator, objectivism instead to its content as motivator. But there cannot be two equally correct, yet distinct explanations of the same phenomenon, explanations which appeal to the very same features, while assigning them completely different explanatory roles. Moreover, in his (2004: 31), Dancy argues that the explanatory principles that supposedly connect desires and beliefs with actions on the psychologist/causalist picture are themselves normatively loaded and a priori, rather than simple causal principles. For we do not learn about them by empirical observation; rather, we bring our assumptions about rational norms connecting mental states and behavior to bear on the actions we explain. In the end, they reduce to the objective normative principles involved in normative explanations. Dancy’s arguments against causalism about reasons and about action explanation are not persuasive. For one, the pluralism we described acknowledges that the mental states that are invoked in a causal action explanation are not the agent’s motivating reasons, but have these as their contents. So, these are not competing explanations for the same explanandum. For another, even if the general principles by which we

130  Hans-Johann Glock and Eva Schmidt explain actions are a priori, it may still be the case that specific action explanations are causal, in that they take the agent’s motivating reasons or her mental states to make a difference to her actions, and hence to what happens. Finally, in line with the “logical connection” argument of earlier anti-causalists like Melden, von Wright, and Anscombe, Dancy seems to rely on the following assumption: If two phenomena— such as a reason or motivation to act a certain way and an action— are connected normatively or conceptually, they cannot be distinct and hence cannot stand in causal relations. But as Davidson (1980a: 14–15) points out, whether or not a statement normatively/conceptually links phenomena is a matter not of the phenomena themselves, but of how they are described. The very same phenomena may be normatively/logically related under one pair of descriptions, but not under another, as the statements, “The cause of B causes B”, and “A causes B” illustrate (for details, see Glock 2014a: 30–33). These objections are sufficient to warrant another look at causalism and how it might be combined with objectivism about reasons.

Combining Causalism With Objectivism Agents are often motivated by what is the case and they are able to act for normative reasons. In consequence, we endorse objectivism and reject psychologism about motivating reasons: Motivating reasons are facts or states of affairs (purported/potential facts), depending on whether the agent represents the world correctly or incorrectly. However, to abandon causalism (about reasons and about action explanation) in addition to psychologism is to throw out the causalist baby with the psychologistic bathwater. The option of combining objectivism with causalism has recently been pursued by several authors (e.g. Perner and Roessler 2010; Hyman 2015, ch. 5). Our variant is indebted to Mantel’s “competence account” (2018). But this is not the place to compare the respective merits of these approaches. Our aim is simply to explore the option of holding on to a (potentially modified) version of causalism in combination with objectivism about motivating reasons—which we call “objecto-causalism”. Its starting point is that even though motivating reasons are (purported) facts rather than mental states, they may well cause agents to act.10 In the following, we will first introduce objecto-causalism in some more detail by comparing and contrasting it with what Dancy (2000: 101) calls “the three-part story”; we then motivate objecto-causalism, before turning to two core problems it faces: future normative reasons and error cases. The three-part story has it that, in the order of causal explanation, the first element is a fact/normative reason, which causes the agent to believe that the fact obtains (second element), where the belief—taken as mental state—is identified as the motivating reason, which in turn causes the

Objectivism and Causalism  131 agent to act in accordance with the normative reason (third element). So, the normative reason causally explains the agent’s belief and thereby— due to the transitivity of explanation—also explains her action, at least partly. This connection is supposedly sufficient for the agent to act for the normative reason. Dancy criticizes this view by appeal to the identity thesis: In cases in which we are motivated by a true consideration, the motivating reason is identical with rather than a causal consequence—a mental effect—of the normative reason. By contrast, according to objecto-causalism, when all goes well, the agent’s motivating reason is identical with a normative reason favoring her action.11 Moreover, for the agent to act for the normative reason that p, she has to believe that p; and where she does so, the believing will be caused by the normative reason—the fact that p. Her believing that p, in turn, has to cause her to act in accordance with the normative reason that p.12 The agent can be said to act for the normative reason just in case this causal sequence from normative reason to action is in place; and the normative reason can be said to be her motivating reason. The normative reason then causally explains why the agent acts as she does. Objecto-Causalism: A Φ-s for the normative reason that p iff (1) that p favors Φ-ing, (2) that p causes A to represent13 that p, and (3) her representation that p causes A to Φ.14 The contrast between the three-part story and objecto-causalism can be displayed as follows: explains causes

causes

fact that p

S’s belief that p

normative reason for S to Φ

S Φ-s

S’s motivating reason

Figure 7.1  The Three-Part Story

explains and motivates causes

fact that p normative reason for S to Φ = S’s motivating reason

Figure 7.2 Objecto-Causalism

causes

S’s belief that p S’s way of possessing the normative reason p

S Φ-s

132  Hans-Johann Glock and Eva Schmidt Objecto-causalism is just like the three-part story, except that it identifies the agent’s motivating reason with the normative reason for which she acts. It combines Dancy’s metaphysical picture of normative and motivating reasons with a causal picture of what it takes to act for a normative reason: When the agent acts for a normative reason, the reason causes the agent to act in accordance with it; this involves the agent’s being aware of that fact. The normative reason, which is also the agent’s motivating reason, causally explains her action. Another way to think about objectocausalism is this: For an agent to act for a normative reason, the normative reason has to move her to act in accordance with it, and it can only do so if the agent “possesses” the reason, in the sense that she is aware of the fact that p, and acts in the light of that fact. Possessing the reason involves mental states of the agent’s which correctly represent the reason. What is more, these states are causally related to the fact that p, even if that relation does not fit the Humean restriction of causal relations to events. The fact is not just a reason for the agent to hold a belief, it is also a causal factor, in the same way in which in Davidson’s theory beliefs and desires are causal factors without being events that fit the Humean picture of causation (see endnote 3). The fact that p is an aspect of reality that makes a difference to whether or not the agent comes to believe it. The same holds in turn for the believing and the action. Accordingly, the fact (normative and motivating reason in one) is also a distal cause of the action.15 This does not rule out that there also distal and proximal events involved in the overall causal picture; in fact, according to the account defended below they are required to trigger the manifestation of agential capacities. In our example, the normative reason—the fact that it is freezing outside—causes Yuki to believe that it is freezing outside. This belief (together with her desire to keep the apartment warm, etc.) then causes her to close the window. In other words, Yuki possesses the normative reason, that it is freezing outside, and can do what the reason favors on its basis: close the window. We can therefore say that Yuki closes the window because it is freezing outside. This fact is what causes her to close the window; it is her motivating reason, and we can think of her as acting for that normative reason. Objecto-causalism, unlike the three-part story, is not affected by objections from the identity thesis, since it endorses that thesis. But what speaks in favor of the view? First off, where agents act for normative reasons, it gives the latter a robust role in the production of action. If the reasons cause the action, they are efficient causes in the weak sense of making a difference to whether or not the action is performed. Moreover, objecto-causalism does not have to make do with an unanalyzable “in the light of” relation as Dancy does, but can give a deeper elucidation of how normative reasons move us to act (cf. Setiya 2009). Finally, objectocausalism reflects our ordinary way of conceiving of agents’ interactions with the world and each other.

Objectivism and Causalism  133 Let us give two examples to illustrate this last point. First, think of a soccer player. She sees and believes that her teammate has passed the ball to her and that the ball is rolling towards her at a certain speed and angle; in reaction she moves her foot in just the way to kick it into the goal. Plausibly, we can tell both a causal story and a reasons-story about this interaction. That the ball rolls towards her in this-and-that way causes her to kick it in such-and-such a way, which causes the ball to go into the goal. At the same time, that the ball rolls towards her in this-andthat way is a normative reason for her to kick it in such-and-such a way, which the soccer player possesses thanks to her visual experience and belief. She kicks it in such-and-such a way in the light of her motivating reason, and thus does what the reason favors. On our ordinary way of thinking, the causal story and the reasons-story are neither mutually exclusive nor detached from one another; rather, they go hand in hand. We explain actions by reference to facts which both justify action and render its occurrence intelligible, even if we do not explicitly use the terminology of causation or justification. The soccer player’s normative reason causes her to act appropriately; she acts for a normative reason when she kicks the ball, in light of how it is coming at her, in just the right way to score a goal. We have a step-by-step causal chain with no gap in the places where agents move from reason to action. In the second example, a father sends his daughter to her room because she pushed her baby brother, telling her that she has only herself to blame that she is now stuck in her room—she brought this on herself. Assume that the fact that the daughter pushed her brother is a normative reason for the father to send her to her room, which he possesses because he saw her push her baby brother. His claim that she has only herself to blame, or that she has brought this on herself, involves not only a normative, but also a causal presupposition: It was her pushing her brother which made her father send her to her room. Generally, ordinary thought takes it that just punishment is an appropriate response to someone’s wrongdoing. But the idea of a response implies that there is a causal relationship between the fact that someone did something wrong (the normative reason that justifies the punishment) and the punishment imposed (the action done for this normative reason). So again, our ordinary way of thinking involved in our practice of sanctioning misbehavior has both a normative and a causal dimension; the action that is performed for the normative reason is also caused by it. By contrast, it is hard to see how Dancy’s view could account for the causal and normative dimensions of our acting seamlessly interlocking with each other. As sketched earlier, he rightly denies that normative explanations are themselves at the same time causal explanations, and further, and more problematically, denies that there could be paralleling causal and normative explanations of an agent’s action in terms of mental states and their contents. In light of this, objecto-causalism has an

134  Hans-Johann Glock and Eva Schmidt advantage over Dancy’s normative story, in that it allows normative reasons both to favor actions, and thus play a normative role, and to cause actions, and thus play a motivating, causal role. So much for the motivations for objecto-causalism. But what challenges does it face? It has potential problems with deviant causal chains, future normative reasons, and error cases. In this essay we shall leave aside the first obstacle, which is a festering wound in any version of causalism (see Hyman 2015, ch. 5). We harbor some hopes that an appeal to capacities can resolve it, but we shall not make a case for this claim here. Instead, we shall tackle the other two challenges in the following sections.

Trouble for Objecto-Causalism To illustrate the problems objecto-causalism has with future normative reasons, take the following example. Hasan will eat an eight-course dinner at a three-star restaurant tonight. That he will do so favors his not eating anything for lunch today—it is a normative reason for him to skip lunch. Moreover, that he will eat the eight-course dinner tonight is the reason for which he skips lunch now. However, that he will eat the eight-course dinner tonight cannot cause his action of skipping lunch now, for it is a future normative reason, that is, it is not now/not yet the case that he is eating an eight-course dinner. Since there is no backwards causation, this not-yetobtaining state of affairs cannot cause Hasan’s current action of skipping lunch. But according to objecto-causalism, a normative reason is the agent’s motivating reason only if it is involved in causing her to act in accordance with the reason. So here we have a counterexample to objecto-causalism: A normative reason is the reason for which the agent acts even though the normative reason does not cause (or causally explain) the action. To make matters worse, because of error cases we cannot generally understand reasons for which agents act as facts that cause them to act in certain ways. Here is an example. Yuki falsely believes that it is freezing outside and therefore closes the window—it merely appears to her that it is freezing outside. Still, the reason for which she closes the window is that (as she believes) it is freezing outside. Now since it is not actually freezing outside, there is no fact that it is freezing outside which might cause her to close the window; a merely purported fact, a state of affairs that does not obtain, cannot cause anything. Nevertheless, Yuki has a reason for closing the window and indeed closes it for that reason. Objecto-causalism cannot capture this, for it only ascribes motivating reasons when a normative reason causes the agent to act accordingly. At best, therefore, objecto-causalism has nothing to say about error cases; at worst, it is committed to the untenable claim that there are no motivating reasons in such cases. In response to the problem of error cases, it might look attractive to explicitly restrict objecto-causalism to the good cases—cases in which

Objectivism and Causalism  135 the agent succeeds to act for a normative reason. In fact, we phrased the view in just this way above. In parallel to disjunctivism about perception (the view that we need two fundamentally different accounts for veridical perception on the one hand and hallucination on the other), one might combine objecto-causalism about acting for a normative reason with a different, yet still causal account of acting for a merely apparent normative reason.16 In the success case, the fact that it is freezing outside is Yuki’s motivating reason and causes her to close the window. But in the error case, her false belief (in the sense of mental state) that it is freezing outside causes her to close the window. Now what is Yuki’s motivating reason in the error case? If the motivating reason has to cause the agent to act, it seems that she either has no motivating reason (for there is no pertinent fact that causes her belief and action) or that her believing, i.e. a mental state, is her motivating reason after all. Neither option is attractive, for on the one hand, agents have considerations in whose light they act in error cases as much as in good cases. On the other hand, Yuki acts in the light of the apparent fact that it is freezing outside, not in the light of the fact that she believes that it is freezing. This latter fact plays no role as premise of her practical reasoning.17 Moreover, disjunctivism does not solve the problem of future normative reasons. Because of its causal condition on what it takes for a normative reason to become motivating, it fails to give us a way of making future normative reasons, that is facts that do not yet obtain, but that favor a certain way of acting now, themselves our motivating reasons. This result is very problematic for objecto-causalism. For the core idea of objecto-causalism is to give the facts that favor our actions a robust place in what makes us act, by depicting these facts as causal factors bringing about our actions. It would be bad news for the account if there was a class of cases—actions for future normative reasons—where it simply did not apply.

From Objecto-Causalism to Objecto-Capacitism We need to find a better way to deal with the problem cases, one which provides both an account of agents acting for future normative reasons and an account of the motivating reasons in error cases. Let’s return to the example of Hasan skipping lunch. What is going on with him in this situation, in which he manages to skip lunch now, for the normative reason that he will have an eight-course meal at a three-star restaurant tonight? He knows that he will have the eight-course meal tonight (imagine that he has been invited by his extremely reliable mother) and therefore it is a reason he possesses. Moreover, he thinks that it would be a waste not to be able to eat all the excellent food, so that it would be good if he had a healthy appetite tonight. So he intends to make sure he will be hungry then. As a result of all this, he is able to adjust his actions

136  Hans-Johann Glock and Eva Schmidt in the light of his future circumstances, to act in accordance with them now. At present, Hasan has the capacity to know about (some of) his future normative reasons, based on indicators about what will happen that he has currently available; he has the ability to be motivated, in what he currently does, by what he knows will happen later; and he has the competence to act now on the basis of his motivation. That is to say, he has epistemic, motivational, and executional competences to act in accordance with future normative reasons (cf. Mantel 2018: 43–44). He further manifests all these competences in skipping lunch, and so he skips lunch for the future normative reason that he will have an eight-course meal tonight. Generally speaking, when an agent succeeds in acting for a normative reason, i.e., does what it favors, she successfully exercises her competence or capacity to respond appropriately to normative reasons to act. An agent’s capacities may or may not include a competence to do what future normative reasons favor, a competence that is manifested by present actions.18 There is nothing mysterious about agents being able to act now in light of facts concerning the future, provided only that we grant them the capacity to know about the future. This suggests that the causal element of objecto-causalism needs to be replaced with a competence condition on acting for a reason (though, as we will show, this is compatible with the normative reason causing the agent to act). We call the resulting account “objecto-capacitism”. Our outlook is inspired by Mantel’s (2018) normative competence account of acting for a normative reason.19 Objecto-capacitism differs fundamentally from Mantel’s view, however, in that it sticks with objectivism, identifying the reasons for which the agent acts with (purported) facts. Mantel, by contrast, identifies motivating reasons with mental contents, which she takes to be Fregean propositions, distinguishing them from the facts they represent. At the same time, objecto-capacitism takes from Mantel the idea that acting for a normative reason requires the agent to respond competently to that reason in doing what it favors. Let us next elaborate more on the picture painted by objecto-capacitism. To begin with, we can think of capacities or competences as complex dispositions, constituted by epistemic, motivational, and executional sub-competences or dispositions. When an object has a dispositional property, certain stimuli will at least be likely to trigger the manifestation of the disposition if the object is placed in certain circumstances (see Choi and Fara 2018; Mantel 2018: 45–47). Unlike the disposition of a fragile glass to break when struck, competences are more flexible—they may be set off by a broader range of triggers, for instance—, and they are features of agents that we value. Furthermore, the capacity to act for reasons is often what Kenny calls a “two-way power”. Its manifestation is not automatic but subject to the agent’s discretion, in some cases by way of explicit decision or choice.20 The capacity to respond appropriately to

Objectivism and Causalism  137 normative reasons is moreover normatively charged, since it is a capacity that is defined by the normative reasons to which it enables the agent to respond competently. We can think of such competences as a kind of virtue. As Mantel (2018: 46) argues, the competence to do what a normative reason favors is a tracking competence, that is, a competence that enables its possessor to make her actions match the normative reasons there are. When the agent manifests the competence, she succeeds in acting in accordance with the normative reason; this is the manifestation of the competence in response to the normative reason. The manifestation of this overall competence presupposes her manifestation of the subcompetences: She has to manifest the epistemic competence to represent the normative reason correctly, thereby possessing the reason, the motivational competence to be motivated by the possessed normative reason to act in accordance with it, and the executional competence to do what she is motivated to do, thereby doing what the normative reasons favors (see figure 7.3). The normative reason—the fact that p—is a causal factor, a stimulus that triggers the manifestation of the epistemic competence, namely the agent’s believing that p. That triggering may be regarded as an event causing the “onslaught” of the belief, that is the beginning of the state of believing. Furthermore, the agent can epistemically respond to the normative reason, the fact that p, either directly or in response to an indicator of the normative reason, another fact that q which indicates the obtaining of the fact that p. Objecto-Capacitism (1): A Φ-s for the normative reason that p iff in Φ-ing, A manifests her competence to do what the normative reason that p favors. We can make this sketch more concrete by applying the account to the examples discussed earlier. Yuki closes the window for the reason that it is freezing outside; that it is freezing outside is a fact that favors Yuki’s closing the window and she does what it favors. Moreover, she feels the icy cold outside and on the basis of this believes truly that it is freezing outside, thereby manifesting her epistemic capacity to represent this fact, explains and motivates

manifestation of S’s competence to Φ for the normative reason that p triggers/causes

fact that p normative reason for S to Φ = S’s motivating reason

triggers/causes

S’s belief that p S’s way of possessing the normative reason p

Figure 7.3  Objecto-Capacitism: Standard Case

triggers/causes

S’s motivation to Φ

S Φ-s

138  Hans-Johann Glock and Eva Schmidt a normative reason, correctly. She thereby possesses the fact that it is freezing outside as a normative reason to close the window. The possessed reason next motivates her to close the window, and in being so motivated she manifests her capacity to be motivated by a normative reason she possesses to do what this reason favors. Finally, she competently acts in the way she is motivated, thus manifesting her executional competence. Overall, then, she manifests the competence to match her action to the normative reason—to do what the reason favors. She therefore closes the window for the normative reason that it is freezing outside. Notably, in this case, the fact that it is freezing outside plausibly causes Yuki to close the window. It lies upstream in the causal chain leading up to her closing the window and can be seen as the stimulus that triggers the agent to manifest her competence to act for this reason and that by the same token causes her to close the window. So here, her acting for a normative reason is accompanied by her manifesting the competence to do so and by a causal relation between the normative reason and the action.21 This contrasts with cases involving future normative reasons. Hasan’s action of skipping lunch now matches his normative reason that he will have an eight-course meal tonight. In skipping lunch now, he manifests all the pertinent competences: He believes correctly that he will have an eight-course meal tonight thanks to his epistemic competence and therefore possesses this normative reason. In being motivated to skip lunch, he manifests his motivational competence to be motivated to do what the possessed reason favors, and in indeed skipping lunch, his executional competence to do what he is motivated to do. So, he manifests the competence to skip lunch now for the normative reason that he will have an eight-course meal tonight, and skips lunch for this normative reason. But what triggers the manifestation of his competence is not the fact itself that he will have the eight-course meal tonight, but a distinct explains

manifestation of S’s competence to Φ for the normative reason that p triggers/causes

fact that q

triggers/causes

S’s belief that p

triggers/causes

S’s motivation to Φ

S’s way of possessing the normative reason p

S Φ-s motivates and explains

fact that p indicates

Figure 7.4  Objecto-Capacitism: Future Reasons

normative reason for S to Φ = S’s motivating reason

Objectivism and Causalism  139 fact indicating this normative reason—in the example, the fact that his extremely reliable mother invited him. This is the stimulus that causes him to skip lunch now. This goes to show that a normative reason need not cause an action in order for the action to be performed for the reason. But in every case, an agent needs to manifest her competence to do what a normative reason favors in order for her to act for that reason. But what about error cases? Yuki believes that it is freezing outside and closes the window, but it is quite warm outside. Imagine that she has the belief because yesterday’s weather forecast predicted (mistakenly) that it would be freezing today. In this scenario, she cannot act for the normative reason that it is freezing outside because it is not the case that it is freezing outside; but the reason for which she acts is that it is freezing outside—this is the consideration that motivates her to close the window. Looking at the competences manifested by Yuki in this scenario, she does not manifest the epistemic constituent of the competence to close the window for the normative reason that it is freezing outside. Her belief that it is freezing outside is false and so not a manifestation of an epistemic competence to form beliefs representing correctly facts that are normative reasons.22 So, that it is freezing outside is not a normative reason to close the window that Yuki possesses. Next, she does not manifest her competence to be motivated to act in accordance with a possessed normative reason, as she is motivated to close the window even though she does not possess a normative reason to do so. However, she does manifest the executional competence to act as she is motivated. So overall, she unsuccessfully exercises, but does not manifest her competence to act for the reason that it is freezing outside. To adjust the error case to make it more similar to the original scenario, imagine that Yuki falsely believes that it is freezing outside because she is undergoing a tactile illusion of the outside air being icy cold. Her tactile illusion of the icy cold leads to her false belief (an unsuccessful exercise of her epistemic capacity), which triggers an exercise of her capacity to be motivated by what she takes to be the case and then a manifestation of her executional capacity to close the window.23 Overall, the competence to act for a normative reason is unsuccessfully exercised. On the one hand, it’s triggered by the illusory experience and does not involve a manifestation of an epistemic competence to represent correctly a fact that is a normative reason; on the other hand, what motivates her to close the window is not a normative reason but only a presumed normative reason; finally, she does successfully execute her capacity to act as she is motivated. In both versions of the story, we can explain the fact that Yuki closes the window by appeal to the fact that her competence to represent the fact that it is freezing outside is triggered by unsuitable circumstances, which consequently unsuitably trigger her motivational competence, which then suitably triggers her action.

140  Hans-Johann Glock and Eva Schmidt Objecto-Capacitism (2): A Φ-s for the reason that p iff in Φ-ing, A exercises her competence to do what the (presumed) normative reason that p favors. This provides a causal-dispositional explanation of Yuki’s action, but does not appeal to the reasons in the light of which she acts, as in Dancy’s normative story. However, there is no reason why proponents of objectocapacitism should be barred from going pluralist and additionally providing an explanation of her action in terms of reasons in the light of which. That Yuki’s situation, as it appears to her, is one in which it is freezing outside, which (apparently) favors her closing the window, motivates her to close the window, and so she does. By contrast to Dancy’s anti-pluralist stance, we believe that this account can be underpinned with the dispositional account. Due to her unsuccessful exercise of her epistemic competence, her situation appears to her to be one in which it is freezing outside. In response to the purported fact—a state of affairs mistakenly believed to obtain— that it is freezing outside, which throws an apparently favorable light on closing the window, she employs her motivational competence to be motivated to do what a presumed normative reason favors. She finally manifests her executional competence and closes the window. Rather than competing with the motivating-reasons explanation, the dispositional explanation complements it. That she unsuccessfully employed an epistemic capacity explains her apparent situation, and this explains the (apparently) favorable light in which she sees the action and which persuades her to act. Moreover, that, as far as Yuki can tell, her situation is the same in the success case and the error case, is explained by the fact that the same competences are employed in both cases, just once successfully and once unsuccessfully.

apparent fact that p

motivates and normatively explains

represents explains

unsuccessful exercise of S’s competence to Φ for the normative reason that p triggers/causes

fact that q

triggers/causes

S’s belief that p S’s way of possessing the apparent fact p as a reason

Figure 7.5  Objecto-Capacitism: Error Case

triggers/causes

S’s motivation to Φ

S Φ-s

Objectivism and Causalism  141 To sum up, objecto-capacitism is able to uphold objectivism about motivating reasons—the reasons for which we act are the facts or purported facts in the light of which we make up our minds and act. It concedes that facts may be the (normative) reasons for which an agent acts even though they do not cause the action, and thereby abandons the causal condition on motivating reasons. This still allows that there are many cases in which the normative reason for which the agent acts is the cause of the action. Instead of a causal condition on acting for a normative reason, it adopts a competence condition. Accordingly, an agent acts for a normative reason only if in so acting she manifests her competence to act for that reason, so that this explains why she does what the reason favors. Unlike the objecto-causalism we started with, objecto-capacitism can account for actions for future normative reasons, for the manifestation of the competence may be triggered by an indicator of a future normative reason. And it can handle error cases, in which the action is normatively explained by the purported normative reason for which the agent acts, where this explanation is supplemented by a causal-dispositional story. According to that story, the agent’s situation appears to her to be a certain way because her epistemic competence to represent normative reasons is unsuitably triggered; this then unsuitably triggers her competence to be motivated by the possessed normative reason, and finally (suitably) triggers her executional competence. It is notable then that despite its departure from our original causalist proposal, causal relations play a central role according to objecto-capacitism. For one, the view allows that there is a sequence of events that culminates in the triggering of the competence. For another, whether an agent has the capacity to act for a normative reason is itself causally relevant—in particular, it makes a difference for how she reacts to the presence of the reason. Capacities can be seamlessly integrated into a causal picture of the world. Accordingly, while objecto-capacitism rejects the core claim of objecto-causalism (that an action is performed for a normative reason only if the reason causes the action), it is still in the spirit of that view: It combines objectivism about motivating reasons with allowing that action explanation has an essential causal dimension.

Implications for Historical Explanations Explanations of actions by reference to reasons appeal to objective phenomena that weighed with the agent. Nevertheless, these explanations are a type of causal explanation. For they appeal to the cognitive, motivational, and executive capacities of rational agents. These capacities are not Humean causes, i.e. events. But like the onslaughts of beliefs and desires according to Davidson, their triggerings are events. And like the beliefs and desires in Davidson’s theory, the capacities themselves, as

142  Hans-Johann Glock and Eva Schmidt dispositions, are causal factors which form part of the causal history of the acts they explain. They make a difference to whether or not the agent acts in a certain way, to her behavior. A sustained survey of the interplay between different types of causes and causal factors and to the kind of difference they make is beyond our remit. Instead, we draw the conclusion that the conflict between Erklären and Verstehen can be resolved by acknowledging the teleological, justificatory, and causal aspects of the way we make sense of actions. In explaining actions we refer to causal factors, facts on the one hand, cognitive, motivational, and executional capacities on the other. But the motivational factors also have a teleological dimension, since they are typically capacities to pursue goals; and the cognitive factors have a justificatory dimension, since they are capacities to respond to reasons, to facts or potential facts concerning the action. Furthermore, we both understand the action by considering the presumed facts weighing with the agent and explain the action by reference to manifestations of epistemic, motivational, and executive competences brought about by facts. Accordingly, though causal, the explanation is different from mechanistic explanations exclusively in terms of physical events. The appeal to capacities for responding to reasons chimes with a view which is shared by common sense and professional historians: It is not just facts but also the cognitive and motivational powers of protagonists to respond to them that makes a difference to how history unfolds, yet not in the way in which e.g. mechanical causes do. Objecto-capacitism explains both the causal and the normative-cum-teleological side of this coin. And it can do justice both to being motivated by facts concerning the future and to being wrong about the facts. This again should be congenial to the study of history. For its protagonists are often motivated by anticipations of future events, and they are notoriously prone to getting the facts wrong.

Notes 1 The research for this paper was financially supported by the SNF within the DACH project The structure and development of understanding actions and reasons. We are grateful for discussions of earlier drafts with audiences in Bochum and Zurich. We would like to thank Susanne Mantel and Gunnar Schumann for helpful comments on earlier drafts, and Gunnar Schumann for his patience. 2 Other popular labels are potentially misleading: “Rational explanation” mistakenly suggests that the action is the rational i.e. optimal thing to do, all things considered; and “intentional explanation” is problematic because some unintentional acts are done for a reason, e.g. laughing out loud at a ludicrous sight. 3 Thus, we shall ignore the possibility of understanding merely apparent facts, which an agent takes to obtain, as normative reasons. We shall also ignore cases in which an agent acts on account of a fact, yet one which does not actually favor the action. As against extreme forms of relativism, we shall

Objectivism and Causalism  143 assume that some facts favor an action independently of whether the agent thinks so. 4 Strictly speaking, it is the “onslaught” of the combination—an event—that causes the action. Davidson’s (1980b) view is further that it is constitutive of an action (as compared to a mere movement of the limbs) that there is a description of it under which it is caused by the agent’s reason. We will leave this further view—a causal theory of what makes something an action combined with the view that all actions are “intentional” in the sense of done for a reason—to one side for the purposes of this paper. 5 They are typically concerned with the world, but may sometimes be concerned with the minds of people. For instance, that hitting a person hurts him is a reason to stop; that he believes to be followed by the CIA is a reason for him to seek psychotherapy (Hornsby 2008: 247). 6 As endorsed by Glock (2009) and, with respect to epistemic reasons, by Schmidt (2017). 7 Motivating and explanatory reasons can come apart, however. If John punches Peter because he has learned that Peter betrayed him, then the fact that he knows about this explains his punching; yet, it is a fact about Peter, namely, that he betrayed John, which motivates John to punch him (Alvarez 2017). 8 We will discuss this issue in detail below. 9 As suggested by Davis (2003, 2005) and Setiya (2009: 133). 10 A note on facts and causation: The Humean mainstream has it that only events can cause other events; but there are also more liberal views, according to which one fact may cause another to obtain. Cf. Dancy (2000: 159–160). We assume that a liberal view is correct and that facts (alongside events) can be causes or at least feature in causal explanations. 11 As will become clear, objecto-causalism needs to be restricted in this way to cases in which the motivating reason is indeed a fact which causes the agent to act. Another view which combines objectivism and causalism by treating objective facts as causes of the action is “teleology” (see Perner and Roessler 2010). But unlike our objecto-capacitism it lacks a sustained option for dealing with error cases and seems committed to the possibility of backwards causation. 12 While the concepts of belief and action are conceptually connected insofar as it is constitutive of beliefs that they can be adduced in explaining actions, there is for the most part no conceptual connection between individual beliefs and individual actions. One and the same belief can manifest itself in many different actions and many beliefs manifest in no actions at all, contrary to logical behaviorism. 13 Talking about “representing” in this context is not meant to commit us to a representationalism according to which the subject relates to a fact via a mental proxy that represents the fact. It is merely shorthand for cognitive intentional verbs like “believing”, “knowing”, “remembering”, or “perceiving”. 14 The right-to-left direction of the biconditional is subject to the caveat that the problem of deviant causal chains is resolved. 15 One must keep apart the distinction between event causation and causal factors of different types from the distinction between proximal and distal causes. Against objecto-causalism one might protest that facts can only be distal causes of actions, whereas it is the mental state that is the proximal cause. But the proximal-distal distinction is context-relative and gradual. While it is clear that the believing is causally closer to the action than the fact, it is not the proximal cause strictly speaking, the latter consisting of the

144  Hans-Johann Glock and Eva Schmidt enervation of motoric nerves or muscular contractions even further down the line. 16 A view along these lines is suggested by Hornsby (2008). 17 Alvarez (2017) bites this bullet by accepting that in the error case, agents have only an “apparent” reason. At the same time she holds (correctly, in our view) that this case involves an explanatory reason, namely the fact that the agent mistakenly believes something to be the case. 18 Plausibly, adult humans have a fully fledged capacity to act for future normative reasons. But there are other agents that lack such a capacity, such as infants or some non-human animals—they cannot adjust their current actions in light of what would be good in the distant future. 19 These days, capacities are appealed to explain quite a range of phenomena, and under many different labels; other examples are Schellenberg’s (2018) “capacitism” about perception and Glock’s (2014b) “capacity approach” to the mind. 20 With this acknowledgment, we can allow that agents are not always compelled by the strength of their normative reasons to do what these reasons favor, a predicament allegedly expressed thus by Luther: “Here I stand, I can do no other”. When we possess the capacity to act for a normative reason, it is often still up to us whether we do what it favors, which is captured by the idea of a two-way power, a power the subject has to Φ or not to Φ, given the opportunity (Steward 2012). As such, this capacity may resist identification with a disposition, since some authors—including Steward—argue that dispositions are tendencies of objects manifested automatically. Others, such as Mumford (1988), treat all potentialities, including two-way powers, as dispositions. This suggests that the issue is at least partly terminological. Deciding between the different alternatives would require in-depth discussion, which we have to reserve for another occasion. Keep in mind that we do not identify acting for a reason either with acting intentionally or with acting voluntarily. 21 Objectivism is often combined with agent-causation. Objecto-capacitism has the advantage of making that notion less mystifying. Agent-causation is not sui generis and inexplicable, but refers to the special intellectual capacities of a special kind of substance—rational agents. What is irreducible according to objecto-capacitism is the notion of capacities being actualized in response to circumstances. 22 But the belief is an unsuccessful exercise of her capacity to form correct representations of facts that are normative reasons. The way we use the terminology, manifestations of capacities are necessarily successful, but exercises of capacities may be unsuccessful. 23 Another way to think about this is to take the tactile experience to be constitutive of the capacity to act for a normative reason, and itself an exercise of a competence to represent a normative reason. Then the trigger will be what caused the agent to undergo the illusion.

References Alvarez, M. (2017). Reasons for Action: Justification, Motivation, Explanation. In: The Stanford Encyclopedia of Philosophy [online] Winter, 2017 ed. Available at: https://plato.stanford.edu/archives/win2017/entries/reasons-just-vsexpl/ [Accessed Dec 20, 2018]. Choi, S. and Fara, M. (2018). Dispositions. In: The Stanford Encyclopedia of Philosophy [online] Fall, 2018 ed. Available at: https://plato.stanford.edu/ archives/fall2018/entries/dispositions/ [Accessed Dec 20, 2018].

Objectivism and Causalism  145 Collingwood, R. (1946). The Idea of History. Oxford: Clarendon Press. Dancy, J. (2000). Practical Reality. Oxford: Oxford University Press. Dancy, J. (2004). Two Ways of Explaining Actions. In: J. Hyman and H. Steward, eds., Agency and Action, 1st ed. Cambridge: Cambridge University Press, pp. 25–42. Dancy, J. (2018). Practical Shape: A  Theory of Practical Reasoning. Oxford: Oxford University Press. Davidson, D. (1980a). Actions, Reasons, and Causes. In: D. Davidson, ed., Essays on Actions and Events, 1st ed. Oxford: Clarendon Press, pp. 3–19. Davidson, D. (1980b). Agency. In: D. Davidson, ed., Essays on Actions and Events, 1st ed. Oxford: Clarendon Press, pp. 43–61. Davis, W. (2003). Psychologism and Humeanism. Philosophy and Phenomenological Research, 67(2), pp. 452–459. Davis, W. (2005). Reasons and Psychological Causes. Philosophical Studies, 122(1), pp. 51–101. Glock, H. (2009). Can Animals Act for Reasons? Inquiry, 52(3), pp. 232–254. Glock, H. (2014a). Reasons for Action: Wittgensteinian and Davidsonian Perspectives in Historical and Metaphilosophical Context. Nordic Wittgenstein Review, 3(1), pp. 7–45. Glock, H. (2014b). Propositional Attitudes, Intentional Contents and Other Representationalist Myths. In: A. Coliva, D. Moyal-Sharrock and V. Munz, eds., Mind, Language and Action, 1st ed. New York: de Gruyter, pp. 512–537. Hornsby, J. (2008). A Disjunctive Conception of Acting for Reasons. In: A. Haddock and F. Macpherson, eds., Disjunctivism: Perception, Action, Knowledge. Oxford: Oxford University Press, pp. 244–261. Hyman, J. (2015). Action, Knowledge, and Will. Oxford: Oxford University Press. Mantel, S. (2018). Determined by Reasons. New York: Routledge. Mill, J. (1911). A System of Logic, 8th ed. London: Longmans, Green and Co. Mumford, S. (1988). Dispositions. Oxford: Oxford University Press. Parfit, D. (2011). On What Matters. Vol. 1. Oxford: Oxford University Press. Perner, J. and Roessler, J. (2010). Teleology and Causal Reasoning in Children’s Theory of Mind. In: J. Aguilar and A. Bucareff, eds., Causing Human Actions: New Perspectives on the Causal Theory of Action, 1st ed. Cambridge: MIT Press, pp. 199–228. Schellenberg, S. (2018). The Unity of Perception. New York: Oxford University Press. Schmidt, E. (2017). Possessing Epistemic Reasons: The Role of Rational Capacities. Philosophical Studies, 176(2), pp. 483–501, DOI: https://doi.org/10.1007/ s11098-017-1025-z. Setiya, K. (2009). Reasons and Causes. European Journal of Philosophy, 19(1), pp. 129–157. Smith, M. (1994). The Moral Problem. Oxford: Blackwell Publishers. Steward, H. (2012). A Metaphysics for Freedom. Oxford: Oxford University Press.

8 Are Reasons Like Shampoo? Constantine Sandis

Prologue Interminable debates concerning the nature of reasons for action (RFAs) typically arise when philosophers, in the grip of specific pictures, enforce Procrustean uniformity onto their respective theories of what an RFA is. Accordingly, competing accounts will highlight features prevalent under one concept or conception of an RFA (a distinction I elaborate in §II) at the expense of those prevalent under other equally legitimate ones. For there is no such thing as the concept of an RFA—or so I shall try to show. Four distinct pictures of an RFA have been particularly dominant in the literature, without exhausting it. One such outlook portrays RFAs as (1) considerations we act upon. Another views RFAs as (2) motivators of our behavior. A third conceives of them as (3) providing explanations of occurrences of some kind, e.g. of the event or process of someone doing something. A final, teleological, picture takes reasons to be (4) ends or goals. While not many philosophers identify (1) with (4) (but see §III), they think that whatever the relation between them is, both (1) and (4) can be further identified with (2) and (3). This conjunctive fact alone should raise our suspicions about how varying conceptions of RFAs are being grouped under a single concept. Conceptual analysts seeking to pin down the concept of an RFA tend to view two or more of (1–4) as different aspects of one thing called a “reason for action”. Under the resulting pressure to provide an account of RFAs, which meet the constraints of all three pictures, theorists are forced to conceive of one or more of these in ways that are at best deeply awkward, on pain of abandoning them altogether. We would do better to replace such “shampoo accounts” with a conceptual pluralism according to which there is no coherent concept of a reason under which all three of these distinct features of reasons can be united. That is to say, there could be no single thing that performs the relevant triple function, or even some of the double-function combinations on offer. Reasons for action simply do not function like two-, three-, or four-in-one shampoos, capable of performing a number of traditionally distinct functions.

Are Reasons Like Shampoo?  147 Conceptual pluralists should be purists about specific RFAs, employing different concepts of them for different purposes: normative, psychological, and socio-historical. An added bonus of doing so is that the question of whether RFAs are causes of action dissolves as a result of a conceptual pluralism that allows us to point to the socio-historical causes of human action without conflating these with either psychological phenomena or the considerations we act upon (see Sandis 2006). By contrast with the monist methods of conceptual analysis, which gave birth to such classics as Gilbert Ryle’s The Concept of Mind (1949) and H.L.A. Hart’s The Concept of Law (1961), as well as to the metaphysical theories (which replaced linguistic observations with modal intuitions) of later works such as David Chalmers’ The Conscious Mind: In Search of a Fundamental Theory (1996) and Phillip Pettit’s A Theory of Freedom (2001),1 the methodology behind my pluralistic approach is best viewed as a kind of analytic deconstruction.

I.  Concepts and Conceptions It is quite natural to distinguish between concepts and conceptions, yet in recent years philosophers have by and large used the two terms interchangeably.2 In his classic paper Essentially Contested Concepts, W.B. Gallie makes a throwaway comment distinguishing the concept of “the way a game is to be played” from “very different conceptions of how the game should be played” (Gallie 1956: 176). His suggestion is that a group might share the concept of the right way to play a game despite its members disagreeing what this way consists in. The distinction was subsequently taken up by Hart (1961), Rawls (1971), Dworkin (1972), and Lukes (1974, passim). Hart considers the original conception of law as “consisting of orders backed by threats” (1961: 37, see also 45) and Rawls famously argues for fairness as a conception of justice (1971: ch. 1 & 3).3 Dworkin (1972: §II) makes a similar move in relation to fairness itself, maintaining that the different members of a group may all share the same concept of fairness, yet differ in their conceptions of what would qualify as fair in certain cases (e.g. in terms of how utilitarian they are). Briefly put—and remaining as neutral as possible with regard to various competing conceptions of the ontology of concepts4—two people may share the same concept of something to the extent that they agree on the relevant uses of the word(s) expressing it, and yet differ in their approaches to that which is being conceived. So understood, conceptions are ways of thinking of phenomena subsumed under concepts, the latter being expressible through (typically linguistic) behavior. I am here largely in agreement with Peter Hacker, who takes the example of love: Possession of a concept is mastery of the use of an expression. A concept is an abstraction from the use of an expression. We have

148  Constantine Sandis relatively little difficulty and, for the most part, little hesitation in translating l’amour, Liebe, amor, eros, and ahava (and their siblings and cousins) as “love”. Whether the concepts expressed are the same or different depends upon the context of their use and on our fluid criteria for concept identity in this domain [. . .] There is no doubt that the ways in which thinkers, novelists, dramatists, and poets in different times or societies have conceived of love have been profoundly different. Some have thought of it as an ennobling emotion; others have conceived of it as a deplorable form of madness [. . .] Some have thought of it as a relationship between two human beings; others conceived it as perfected only when directed at a non human object, such as God or the Idea of the Good. (Hacker 2018: 268) It is tempting to respond to Hacker’s final suggestion by bringing up C.S. Lewis’ distinction between four different kinds of love: Storge, Philia, Eros, and Agape. Surely these are not different ways of conceiving the same thing but, rather, four different concepts of love. One might here object that to share a concept of x just is to not differ in our approaches to it. But this seems wrong to me. For one, the approaches we may differ in, need not be conceptual. We might, for example, agree that anger is an emotion with certain qualities, but disagree on whether it is ever good or justified. This disagreement doesn’t concern the concept of anger: We are not disagreeing about what anger is. It is just that one of us conceives of it as sometimes being justified and the other one doesn’t. Similarly, we may both agree in what a diet soft drink is, even though one conceives of such drinks as an innocuous part of a healthy diet, and the other as a nutritionally void symbol of capitalism gone wrong. Alternative conceptions need not be evaluative. To the extent that any given thing has properties that form no part of the concept of it, we may disagree in our conceptions of the nature of such properties without disagreeing about the concept itself. Thus, we may both share the concept of water, but disagree about its precise chemical constituents (of course those philosophers who think that being H2O is part of the very concept of water do not share the same concept of water as those who deny it).5 Or again, different people may conceive of God in radically different ways in relation to questions concerning time, emotion, interventionism, prayer, and so on, while nevertheless sharing the concept of God as an omniscient, omnipotent, omnibenevolent, and omnipresent being (for competing conceptions of omnipotence see Hart 1961: 149). The Ancient Greek concept of a God, by contrast, is an altogether different one, though not completely unrelated.6 For any difference in conception, one can of course introduce a new technical concept to include it. I might, for example, explicitly state that I shall be using the term “God” to just mean “a being that shares the four qualities mentioned above but is, by

Are Reasons Like Shampoo?  149 definition, interventionist”. Such conceptual stipulation has its benefits, but the more we engage in it the less we can expect that others to share our concepts and, pari passu, enter into meaningful exchanges with us. Let me illustrate further by looking at different conceptions of reasons for action. Suppose we both share the concept of an RFA as (1) a consideration that one might act upon, irrespective of whether or not this is a good thing. We may nonetheless disagree about whether we should further conceive of such reasons as facts, states of affairs, propositions, or whatever else best fits our ontological scheme. Our dispute here is no longer conceptual but purely ontological,7 though this is not to say that further conceptual explorations (e.g., ones related to facts, states of affairs etc.) might not shed light on our disagreement. Which of these conceptions we end up favoring will at least partly depend on whether or not we are conceptual monists about RFAs. Those who assume that there is such a thing as the concept of a reason are likely to further identify RFAs with one or more of the following: 1. Considerations we act upon 2. Motivators of our behavior 3. Explanantia of our actions8 Accordingly, some conceive of RFAs as having the double or triple function of (a) serving as considerations that we might weigh in decision making, (b) motivating our behavior, and/or (c) explaining why our actions occurred as and when they did. This leads to attempts to conceive of reasons as having the kind of ontology that would enable them to perform (a), (b), and/or (c), despite the fact that each of these functions is increasingly more causal than the preceding one. Alternatively, one might insist that the concept of a reason for action covers only one of the above. But how are we to settle which one? Intuitively (1–3) all seem to central to what we take reasons for action to be. Ordinary language can only help us so much here. We ordinarily talk of reasons as considerations we act for, as things that motivate us (often invoking belief and desire talk), and as things that explain why we acted as we did. There is no doubt that (1–3) are all associated with reasons for action in our everyday speech. The question we are trying to settle is whether there is one all-encompassing concept at play, or whether ordinary language allows us to switch between one and the other, with no explicit signaling.9 While what in some sense motivates an act can also explain it, and while the considerations we act for can in some sense be said to motivate us, these two senses are not the same. For example, I might be motivated by fear or jealousy but these are not considerations I  act for, though the fact that I am afraid or jealous may act as I reason for me to seek help. In the sense in which my jealousy explains why I acted as I did, the explanation does not appeal to an agential reason for action, though the

150  Constantine Sandis latter may be nested in the former (the same is true of character traits that aren’t motives e.g. shyness).10 Similarly, a reason may rationalize an action without explaining it, and explain an action without rationalizing it. In the former case we can render an action intelligible by offering reasons for which it could have been done, while falling short of explaining why it actually was. In the latter case we may explain an action by reference to the physical or psychological state of the person performing it (e.g. she was exhausted, confused, angry, exhilarated etc.), or to various features of the situation (e.g. it was dark, the handle was broken, time had run out), none of which necessarily rationalize the action when appealed to. I have previously (Sandis 2013) argued that the considerations for which we act do not, strictly speaking, explain why we so acted. This has important ramifications for the explanation of historical action, for it is one thing to try and render intelligible the thoughts and actions of a past figure, and quite another to explain why past historical events (including events of people acting) took place as and when they did (see Sandis 2006, 2015b, and 2016). Part of my aim of this current paper is to show that while this remains true in a sense, there may be other (equally legitimate) senses of “explain” in which it would be unproblematic to think of considerations as explaining action (e.g. if by “explanation” we simply mean any narrative, such as a myth or just-so story, that renders something intelligible). A conceptual monist who has the consideration “aspect” of RFA most clearly in sight, will quite naturally begin with an ontology best suited for such things and then attempt to stretch it to perform additional functions. Her philosophical task, then, is to place a “consideration function” constraint on the motivating and explanatory “aspects” of RFAs and only then proceed to explore how we might best conceive of considerations so that they might also perform additional functions. It is via such a process of reflective equilibrium, arguably, that Jonathan Dancy (2000) came to conceive of action explanation as non-factive, viz. to claim that something that is not the case may nonetheless motivate and thereby also explain one’s actions. By contrast, a conceptual monist, more focused on the “motivator” aspect of RFA, will begin with a “motivational function” constraint, as opposed to a consideration one. This initial conception of an RFA will thus be some kind of psychological state, and her philosophical task will be that of figuring out how a “mental state” might also act as not only an explanans of action but also a consideration we act upon. It is through such reasoning that Michael Smith and Philip Pettit, for example, conceive of reasons for action as belief-desire pairs, which double as springs of action in light of their “content”, which serves as grounds (Smith and Pettit 1997). Or, again, one might begin with an “explanatory function” constraint on RFAs and then try to figure out how something

Are Reasons Like Shampoo?  151 that explains the occurrence of an event (presumably causally rather than teleologically) can also double as a consideration we act upon and which might thereby be said to motivate us. Having briefly distinguished between concepts and conceptions—and how these might play out for the conceptual monist working on RFAs— I now consider some specific arguments in more detail. If, as I shall be arguing,11 none of them succeed in providing a unified concept of an RFA that meets all three constraints then this gives us prima facie strong reason to accept conceptual pluralism,12 thereby rejecting the promise of a shampoo account of RFAs. In what follows I confine myself to debates focusing on human action. The dogmas that already emerge here as a result of trying to make do with a one-size-fits-all conception of RFAs are bound to corrupt discussions of whether we can attribute RFAs to human animals and AI, but that is a topic for another paper.

II.  Four-in-One Reasons Consider the category of a reason for action that one may be said to have. These may come in a variety of forms: pro tanto, prima facie, all things considered, overall, instrumental, substantive, and so on. Philosophers have conceived of various kinds of reasons that fall under such categories as moral reasons (Dancy 1993), normative reasons (Dancy 1993, Raz 2009), exclusionary reasons (Raz 1990), justifying reasons (Raz 2009), good reasons (Smith and Pettit 1997), grounds (Wittgenstein 2009), grounding reasons (Bond 1983), practical reasons (McNaughton and Rawling 2004), con-reasons (Ruben 2009), enticing reasons (Dancy 2004), rationalizing reasons (Davidson 2001a), reasons the agent has (Davidson 2001a, Williams 1981), reasons why one should act (Sandis 2012a, Skow 2016), reasons for which one should act (Sandis 2012a), content-related and attitude-related reasons (Piller 2006), fact-type reasons (Hornsby 2008), favoring considerations (Dancy 2000), and reasons from which one deliberates and/or acts (Hyman 2011). Many philosophers take this set to overlap with that of reasons why we act (Hornsby 2008: 249, Skow 2016: ch. 6), which are in turn often defined as motivating reasons (Falk 1986a and 1986b, Dancy 2000), explanatory reasons (Raz 2009), agential reasons (Davidson 2001a, Hacker 2009, Sandis 2012a), teleological reasons—including purposes, aims, goals, and intentions (Sehon 1997, Bittner 2001, Schueler 2003, Skow 2016), considerations upon which one acted (Dancy 2000), facts by which a person can be guided (Hyman 2011), reasons that render action intelligible (Dray 1957), belief-type reasons (Hornsby 2008), and rational causes (Marcus 2012). According to many philosophers, there are important connections between the two sets of reasons, not least because it had better be possible to act for the sorts of reasons that fall under the first set (Dancy

152  Constantine Sandis 2000). We might then think that the second set of reasons is a sub-set of the first i.e. that one can only act for a reason that one in some sense has. We famously find this view in Donald Davidson’s seminal paper Action, Reasons, and Causes (2001a). Crucially, Davidson assumes that since it is possible to act for a reason that one has, the reasons we have and act for are indeed identical to the reasons why we act. He thus uses all of the following phrases interchangeably: i. ii. iii. iv.

The agent’s reason for doing what he did (3) The reason that explains the action (3) The reason why an agent did something (4) The agent’s reasons in acting (11)

For Davidson, these are typically all sub-sets of both of the following two co-extensive sets: v. The reason that rationalizes the action (3) vi. The reason the agent had (11) The only exceptions are cases of (v) or (vi) that are reasons the agent did not act upon. In all other cases, Davidson maintains, the reason also doubles as a cause of the action in question. In the first published response in print to Davidson’s article, V.C. Chappell writes that what is “chiefly disturbing about Davidson’s paper is that its central concepts—action, reason, cause, explanation—have not been delineated with sufficient clarity and detail” (Chappell 1963: 701). I have tried to show, above, that this is certainly the case with “reason”. It is easy to forget that Davidson was largely unconcerned with the nature of reasons for action after his initial 1963 paper (i.e. Davidson 2001a). While he spent half his academic life delineating the concepts of action, cause, and explanation, he felt no need to do this with reasons. Indeed, it is remarkable just how little Davidson engages with the notion of a reason for action throughout his remaining Essays on Actions and Events, with the brief exception of the clarificatory remarks in the opening pages of “Intending” (Davidson 2001d: 83–87). Even his famous discussion of so-called deviant causal chains (2001b: 78–80) is primarily concerned with “the causal conditions of intentional action” (80), not the concept of a reason, which he frequently conflates with that of a motive (e.g., Davidson 2001c: 264). It is part of his legacy that subsequent debates on the nature of “reasons for action” took it for granted that there was no real distinction between (i–vi). Even Davidson’s fiercest opponents, offering a radically different ontology of reasons, can only make sense of their opposition by assuming that all are agreed on the concept of a reason for action.13

Are Reasons Like Shampoo?  153 Yet RFAs, on Davidson’s conception, can explain and rationalize action. While Davidson himself doesn’t talk of reasons motivating, his term “primary reason” has effectively been replaced by “motivating reason” with little comment. This is unsurprising given Davidson’s abovementioned conflations of “reason” with “motive”, coupled with his view that RFAs move us to action (a view that is, incidentally, endorsed by many anticausalists, despite the fact that motivation is a causal notion).14 An early analogue of Davidson’s conflations may be found in W.D. Falk, who not only uses “reason” interchangeably with “motive”, but further states that reasons are mental antecedents to the actions they move, rationalize, cause, and provide a telos for: A reason or motive is a moving or impelling thought, the thought of that for the sake, or in view, of which, some act is done [. . .] no intelligible alternative to saying that it “moves” or “impels” in the sense that it functions as a cause of actions [. . .] a causa rationis, a mental antecedent which [. . .] will terminate in the action itself. (Falk 1986a: 25)1 While Davidson does not mention teleology explicitly, he cites Ducasse’s 1925 paper, Explanation, Mechanism, and Teleology, with approval (Davidson 2001c: 261), and there is no doubt that he takes all talk of teleology to be reducible to belief/pro attitude pairs (see Sehon 2010). The philosophy of action has thus come to employ a shampoo model of RFAs, according to which they (i) motivate, (ii), rationalize, (iii) explain, and/or (iv) serve as the purpose of any given action. I have purposefully not included the word “cause” in this list, despite the fact that it is Davidson’s claim that RFAs are causes of action that has received the most attention, both positive and negative. The chief debate between Davidsonians and their opponents has thus centered on the extent and sense (if any), to which reasons can achieve these four things causally. Davidson was responding to anti-causalists about RFAs who frequently appealed to Wittgenstein’s distinction between (i) the reason(s) for an action and (ii) its cause(s), to be further distinguished from (iii) the agent’s motive(s) (e.g. Wittgenstein 2009: §§475–485). But by (i) Wittgenstein had in mind the grounds or justification for a belief or action (viz. what we now call “good” or “normative” reasons),15 to be contrasted with the causes of a happening or event. In themselves, these distinctions tell us nothing about whether to group the reasons for which we act may ever be identical to any sort of cause or motive, or whether they form some fourth, sui generis, group. Whatever the merits of his own position, Davidson was right to be skeptical about many of the arguments he encountered in popular “small red books”, including the “logical connection argument” that the relation between actions and causes is

154  Constantine Sandis empirical, whereas that between actions and reasons is logical (for what we should still salvage from these books see Sandis 2015b; cf. D’Oro & Sandis 2013b: 13ff.). My question, by contrast, is the preliminary one of whether we can expect RFAs to achieve all (as opposed to any given one) of these things at all. If they cannot do this, then we need to open ourselves up to the possibility that RFAs might be causes in some of these respects and not others. This is not the thought that there is one kind of thing called a “reason for action” that performs some of its functions causally and some non-causally. Rather, we should be pluralists about the very sorts of things an RFA can be. We should abandon talk of the concept of a reason for action, replacing it with a conceptual pluralism that allows for different notions of RFAs that are closely tied-up, but without homologous resemblance.16 A reason why an action occurred must be a cause of sorts. To this extent, the explanation of historical events is unavoidably causal, though it is a mistake to here downplay (as Hempel and Davidson do) the causal powers of humans. A reason one acts in the light of, by contrast, doesn’t naturally double as a cause, though Anscombe (1957) was right to point out that “we should often refuse to make any distinction at all between something’s being a reason and, and it’s being a cause” of a certain kind, for whether or not a cause is a reason (or vice versa) depends “on what the action was or what the circumstances were” (§15), but that it doesn’t follow from this that a cause is a “mental event” (§10).

III.  Real and Apparent Reasons In asking why someone did something, I  may well be seeking to discover what her reason for acting was, in the sense of wanting to know the consideration or ground she acted upon. But, as already hinted at in §1, it does not follow from this that grounds capable of rationalizing action may thereby also function as reasons why. Bradford Skow (2016: 142) argues persuasively that teleological answers to why questions report reasons why. Thus, “Lou went to Pusateri’s in order to buy turmeric” is equivalent to “the reason why Lou went to Pusateri’s was [in order] to buy turmeric”. But while it is perfectly natural to refer to Lou’s purpose (viz. to buy turmeric) as her reason, the consideration she acts upon here is not an end or purpose but, rather, the purported fact that Pusateri’s stocks turmeric viz. what she believes.17 If it turns out that Lou was wrong about this, then it would not make her trip any less purposeful. She will have gone to Pusateri’s in vain, but not without an end goal or purpose. Indeed, her failure to find what she was after requires the presence of a telos. But what about Lou’s reason for going there? How does it relate to reasons why? “She went to Pusateri’s because they stock turmeric” is equivalent to “the reason why she went to Pusateri’s was that they stock turmeric”. The

Are Reasons Like Shampoo?  155 meanings of “because” and “reason why” are here ambiguous between reporting her reason for going, and reporting some unrelated reason why, as in “the reason why she missed the lecture was that her alarm failed to work” (cf. Sandis 2012b: 50, Skow 2016: 175). Either way, the equivalence only works, when Pusateri’s does indeed have turmeric (see Stow 2016: 174).18 So what of the case where Lou falsely believes that she will find turmeric at Pusateri’s? According to Maria Alvarez (2010), on such occasions the agent acts for no reason at all: When the reason why an agent acted is also a reason for which the agent acted, the explanation is what I have called a “reason explanation proper”. When an agent acts for an apparent reason, the explanation that cites this apparent reason is a Humean explanation: he Φ-d because he believed that p. (Alvarez 2010: 197) According to Alvarez, “Pusateri’s stocks turmeric” is only a reason for action if it is true that they stock it. Otherwise it is merely an apparent reason and an apparent reason is no reason at all (it is no more of a reason than an apparent duck is a duck). On such an account, the reason why the agent acted is identical with her reason for acting as she did when (and only when) her relevant beliefs are veridical. When they fail to capture the truth, they are neither a reason why nor a reason for. In such cases, the agent acts for no reason at all, though there is a reason why she acts (which cites the apparent reason). A similar view is defended by Clayton Littlejohn: I agree that we can explain Leo’s action by saying that he runs because he believes falsely that he is being chased. I agree that there are reasons why he acted as he did. I agree that when we see what these reasons are, we can see why Leo was perfectly reasonable in acting the way he did. What I deny is that Leo acted for a reason. There was nothing in the light of which he did what he did. We all know why Leo ran—he ran down the hall because he believed that the killer was after him. This does not explain his action in terms of motivating reasons because it does not tell us what his reasons were—it turned out that he had none. (Littlejohn 2012: 155) But it is misleading, at best, to claim that the person who so acts under a false belief does not act for a reason, viz. acts for no reason at all. Suppose Lou is certain that Pusateri’s will have turmeric when in actual fact it is merely likely that they will have it today. All else being equal, there is a (good) reason for her to go there, namely that it is likely they stock the ingredient she needs. But this cannot be her actual reason for going if

156  Constantine Sandis probability formed no part of her reasoning. We might then say that Lou failed to act for a good reason, even though there was a perfectly good reason for her to act as she did, and one that was not too far removed from her own reasoning at that.19 To infer from this that she is acting for no reason at all, however, is to falsely assume that all reasons we act for (viz. all agential reasons) are good reasons.20 As Anscombe and Hursthouse have argued, people sometimes act intentionally for no (apparent) reason at all, such as when they perform they “arational action” of throwing a plate in anger, or doodle during a lecture.21 But my scenario above doesn’t look like a case in point. After all, Lou is acting with a clear goal and intention and can explain why she acted as she did by citing the purported fact that Pusateri’s will have turmeric. To say that someone acted for no reason is to imply that their action had no rational motivation at all. This is simply not the case when we are motivated by false beliefs. The difference between the Anscombe/ Hursthouse and Alvarez/Littlejohn cases is that in the former agents do not even act for an apparent reason. We may capture the difference between Hursthouse cases and Alvarez cases with the thought that in the former people are said to act for no apparent reason whereas in the latter they act for some apparent reason. Yet if Alvarez is right to think that an apparent reason is no reason at all, there would be no real difference between the two kinds of case, for neither would be acting for an actual reason at all. Yet, unlike the arational agents, who act intentionally for no reason and with no purpose whatsoever, Lou’s trip to Pusateri’s is purposeful and based on reasoning from various considerations, one or more of which are in some important sense her reason for going to the store (see Sandis 2015a). We know what it is to act for a reason that is not a normative reason (viz. a reason why one ought to act a certain way) but this is not to act for no reason of any kind at all. It is perhaps tempting to therefore say, instead, that Lou acts for a real “motivating reason” that merely is an apparent normative reason. Whether this is right depends on what exactly is built into (and left out of) the technical notion of a “motivating reason”. One highly misleading feature of the expression is that there is an important sense in which what motivates Lou here is the false believing, as opposed to the falsehood she believes. This may sound odd to the philosopher trained in reasonology, but it is standard psychology to talk of being motivated by drives, states, and perceptions. Moreover, while it is perfectly grammatical to talk of being motivated by facts, in everyday language this is relatively uncommon compared to talk of being motivated by one’s thoughts, fears, desires, and beliefs, or indeed by motives such jealousy, greed, money, or love. Michael Smith (1987) is not making any kind of error, then, in claiming that the believing is what does the motivating in the veridical case too. The disjunctivist view of action explanations as being provided by (i) the agent’s reasons for action when

Are Reasons Like Shampoo?  157 we get this right and (ii) Humean belief/desire pairs (that are merely reasons why) when we get them wrong thus fails. The disjunctivist views of Alvarez and Littlejohn thus contrast with those of non-disjunctivists such as Michael Smith (1987), who claims that the agent’s reason is always a psychological state, and Jonathan Dancy (2008), who claims that it is always a (purported) fact.22 Dancy himself notes that “both are awkward in the sort of way that is usually due to a bad theory. As Aristotle said, they leave one saying things that nobody would say unless they were defending a theory” (Dancy 2008: 267). Quite; he might have added: in the grip of a picture. On Smith’s picture, false beliefs pose no problem, but we lose the sense in which an agent’s reason is a consideration she acted upon since considerations are things believed and not our so-called “motivational states” of believing them. Dancy’s view, by contrast, retains the latter at the cost of having to claim that either (a) falsehoods can motivate and explain action, or (b) reasons for action do not motivate or explain action (see Davis 2003, Sandis 2012b & 2013). Dancy initially opted for (a) but has since abandoned this in favor of something closer to (b): [T]he explanans is not identical with the reason for which the action was done [. . .] we can say that what explains the action is that it was done for the reason that p, without committing ourselves to saying that what explains the action is that p. It would remain true, however, that we explain an action by giving the reason for which it is done. All that we lose is the idea that the explanans is a proper part of the explanation as a whole [. . .] I said in Practical Reality that the explanans in a reasons-explanation does not have to be the case, and now I accept that this was wrong. (Dancy 2014: 90–91) Dancy’s u-turn reveals the benefits of abandoning the explanatory strand of a shampoo account of reasons, though he still wishes to hold on to the thought that “the relevant explanations are non-factive, since they have a contained clause which does not have to be true if the whole is to be true” (ibid: 91). Nonetheless, we can now see the way out of several impasses. The mistake was to want to allow that one can act for a reason that is not a normative reason without having to deny that reasons can both motivate and explain action. All the claims in question can be fine in themselves, but not when accompanied by the firm persuasion that that there is one thing called a “motivating reason for action”, which all these claims are about. Alvarez and Littlejohn are right to think that, in the case of false belief, we can truthfully state that what motivates and explains the action is not the agent’s reasons for doing it, but a reason why she acts as she does. Smith and Dancy are both right to think that these cases still involve an agent acting for a reason, but wrong to

158  Constantine Sandis maintain that the reason in question can either motivate or explain action in the relevant sense if it is to also serve as a consideration that the agent acts upon. Only by abandoning the shampoo picture of reasons can we find a way out of what otherwise appears as an impasse. The desire to hold on to some unified conception of a reason has led philosophers on different sides of these debates to all make a suspiciously similar kind of distinction: between the apparent and the real. Thus, Dancy distinguishes between apparent and actual facts or states of affairs (Dancy 2000: 131ff.), Alvarez distinguishes between apparent and proper reasons (Alvarez 2010: 197ff.), and I  have previously distinguished between apparent and genuine explanation (Sandis 2012a: 105ff.). The parallels between these moves suggest a common problem in need of a general treatment. While I presented my own distinction in the midst of trying to argue that we should not conflate the considerations we act upon with either explanatory or motivating reasons, my distinction nonetheless looks suspicious when placed alongside those of Dancy and Alvarez. It would have been better, I now think, to distinguish between different senses of explanation, motivation, and reason for action.23

Epilogue Eric Wiland concludes his book Reasons with the following claim: Reasons for action explain and count in favor of the actions for which they are reasons, and they belong to and are spontaneously known by the person whose reasons they are [. . .] our various criteria of what counts as a reason seem to struggle against each other—that is, in trying to meet some of the constraints, we make it more difficult to meet the others. This curious result suggests that we might not be able to fully understand the nature of reasons until we ourselves are fully reasonable [. . .] perhaps only the wise can do philosophy well. (Wiland 2012: 169, emphasis in the original) Wiland and I agree on the aporetic symptoms of theorizing about RFAs, but not the cause. I am consequently less optimistic that Wiland about the prospects of full wisdom enabling us to fully understand how a reason for action can fulfill the functional ranges mentioned by Wiland and others. This is not because I  claim to have reached full reasonableness and wisdom and it has failed to produce a consistent overarching account of reasons. My conviction that the various competing functions that philosophers have desired cannot be combined by any single thing called an RFA is, rather, the result of observing the various dogmas that emerge as the result of Procrustean attempts to meet all constraints imposed by shampoo accounts of RFAs. We might summarize the most prominent ones as follows:

Are Reasons Like Shampoo?  159 1. 2. 3. 4. 5.

Non-factivism: The Explanation of action need not be factive. Agentialism: Agential reasons are capable of explaining action. Apparentism: We can act for apparent reasons that are not reasons. Psychologism: All RFAs are psychological states. Anti-psychologism: We are never motivated by features of our psychology. 6. Disjunctivism: Whether or not someone acted for a reason could depend on something as trivial as whether there was a remaining jar of turmeric in the store when she got there. 7. Anti-disjunctivism: There is a single explanatory schema in terms of which all actions are explained by RFAs. If the main argument of the above essay is right, we have no reason to hold on to any of them, or others motivated by a similar concern. We should instead embrace a two-layered pluralism about both concepts and conceptions of RFAs. Such a pluralism facilitates the abandonment of the view that reasons resemble shampoo, with one caveat: Insofar as a function (e.g. shampooing) is weakened or distorted to the extent that it is combined with additional functions (e.g. conditioning and body-washing), we might think that RFAs are not so different from shampoo after all.24

Notes 1 Despite its name, John Rawls’ A Theory of Justice (1971) is a transition work that falls between these two approaches. For evidence that it originally developed as a result of conceptual investigations and was only subsequently transformed (as the tide was changing against Wittgenstein) into something approaching a theory, see O’Neill (2014). 2 Some exceptions are Ezcurdia (1988), Baker (2004), Lalumera (2014), and Hacker (2018: 267–269). Only the last of these, however, uses the two terms to mark the precise distinction I  am after. A  comparison with the others would, alas, take us too far afield. 3 See also his Preface to revised edition (1999: xi-xvi). 4 E.g., whether they are mental representations, abilities, or Fregean senses (see Margolis and Laurence 2014: §I). 5 In this, water is disanalogous to something like justice, for there is no question about what justice is that isn’t conceptual (see Hanfling 2000: 17ff.). Prima facie, reasons are more like justice than they are like water. And yet, I shall attempt to show that there remains an important sense in which two people can share the same concept of a reason but differ in the particulars of their conceptions. One explanation of how this is possible is that terms like “motivating reason” are highly technical and can be defined in terms such as “whatever it is that turns out to both motivate and rationalise action”. 6 See Wittgenstein (1980: §139, 1981: Z 646). I leave aside here complications introduced by the Wittgensteinian notion of a family resemblance concept, save to say that this arguably does more damage than good in (i) supposing that it marks out a special sub-set of everyday concepts, and (ii) blurring the distinction between singularity and multiplicity of concepts of any one given thing (see Beardsmore 1992; Sandis 2017).

160  Constantine Sandis 7 While the distinction between conceptual and ontological disputes is a popular, there are good reasons for thinking that it is—in its most general form— spurious. Hanfling (2000) gives the example of petrol. It is of course part of the very concept of petrol that it is a certain kind if liquid. But its precise chemical composition is a further, non-conceptual, question. In the case of petrol, this latter question is chemical as opposed to ontological. But there is no reason to suppose that certain kinds of ontological questions can’t also be bracketed from conceptual ones in the case of technical notions. 8 This third requirement is further complicated by competing conceptions if what an action is. 9 Hornsby (2008: 247ff.) distinguishes between two different “everyday conceptions” of reasons as (a) fact-based and (b) belief-based. These are not meant to be in competition with one another and so in some ways more closely resemble two distinct concepts (one could ask one’s interlocutor whether they were talking about f-based or b-based reasons; cf. Dancy 2008: 275). Indeed, Hornsby’s b-based reasons are akin to my agential reasons, and her f-based ones to what I would call “good grounds” (but see Hornsby 2008: 249–250). 10 I defend these and other distinctions in greater detail in Sandis (2012a: ch. 1 & 2, 2012b), where I further distinguish between different senses in which “action” is taken to be the explanandum in question. 11 For preliminaries see Sandis (2018, n.d.). 12 One need not be a monist or pluralist about the concept(s) expressed by all terms, though one could find evidence suggesting that most terms are best captured by one rather than the other. 13 See essays in D’Oro and Sandis (2013a). 14 To be motivated by something is to be moved by it to perform some action, or refrain from doing so. Not everything that moves us so is what we would ordinarily call a “motive”, the latter being a specific kind of motivator linked to character traits. While there is a conceptual connection between seeing something as valuable and being motivated to act accordingly (see Sandis 2015b), we are often moved to act against our better judgement and can, equally, remain listless in spite of it. Although we can evaluate motives but not causes as being good and bad, all this entails is that it makes no sense to speak of a motive as being good or bad qua cause. One sort of pill may be better at causing drowsiness than another, but we don’t thereby consider it to be a better cause of it. Likewise, greed may be a more efficient motivator of certain forms of behavior than temperance. 15 But see Smith and Pettit (1997) for a proposed distinction between reasons that are normative and reasons that are (merely) good. 16 See note 6 above. 17 See Dancy (2000: ch. 6) and Sandis (2012a: ch. 6). Skow (2016: 162) talks of purposes as (non-factive) reasons that are additional to (factive) agential ones, but enumerating them thus would seem to imply that they can be weighed or added, which cannot be the case. 18 Things would be different if Lou’s reasoning was epistemically cagier, e.g. “Pusateri’s might stock turmeric” or even “they stock turmeric but it’s possible that they have run out”. When modifying the example to discuss acting under false belief it is simplest to take the belief in question to be something like “they will definitely have it” (see further below). 19 Comesaña and McGrath (2014) argue that there are certain cases in which one can have a normative reason that p, even when it is not the case that p. I remain neutral on this question here, save to say that it is misleading to talk of “false reasons” in such cases, as if it is the reason itself that is true or false.

Are Reasons Like Shampoo?  161 The mistake is to think of “that p” as the name of some actual or imaginary entity (see Sandis 2012a: 115–119). 20 Hornsby (2008: 249–250) maintains further that an agent may have a reason to do something even when here is actually no reason for them to do it (cf. Dancy 2008: 275). 21 See Hursthouse (1991); cf. Anscombe (1957: §§ 17ff.) 22 Dancy’s anti-disjunctivism about RFAs leads to his trisjunctive account of acting for a reason (Dancy 2000: 140), eventually paired down to a disjunctive one (Dancy 2008: 268ff.). For complications that contrast a conception of reasons as premises or assumptions with that of reasons as facts see Hyman (2011). 23 Things become even more complicated once we introduce different senses of the term “action”, which any given sense of RFAs may be better or worse suited to (see Sandis 2012a, 2012b). On the conception of actions as events of our doing things, it would just be wrong to claim that there are reasons for action (Sandis 2012a: 29; cf. Stout 1996; Skow 2016: 160). 24 Earlier versions of this paper were delivered at the Reasons, Explanations, and Rational Accountability workshop at the University of Fribourg (November 6–7, 2015) and the Reasons, Rationality, & Rationalising Explanation Workshop at the University of Warwick (April 29, 2016). Many thanks to all of the organizers and participants for their help, especially Maria Alvarez, Joe Cunningham, Magnus Frei, Ulrike Heuer, Frank Hoffman, Clayton Littlejohn, and Guy Longworth. Special thanks to Gunnar Schumann for incredibly helpful comments on an earlier draft.

References Alvarez, M. (2010). Kinds of Reasons: An Essay in the Philosophy of Action. Oxford: Oxford University Press. Anscombe, G. (1957). Intention. Oxford: Blackwell. Baker, G. (2004). Wittgenstein: Concepts or Conceptions? In: K. Morris, ed., Wittgenstein’s Method: Neglected Aspects, 1st ed. Oxford: Blackwell, pp. 260–278. Beardsmore, R. (1992). The Theory of Family Resemblances. Philosophical Investigations, 15(2), pp. 131–146. Bittner, R. (2001). Doing Things for Reasons. Oxford: Oxford University Press. Bond, E. (1983). Reason and Value. Cambridge: Cambridge University Press. Chalmers, D. (1996). The Conscious Mind: In Search of a Fundamental Theory. Oxford: Oxford University Press. Chappell, V. (1963). Causation and the Identification of Actions: Comments. The Journal of Philosophy, 60(23), pp. 700–701. Comesaña, J. and McGrath, M. (2014). Having False Reasons. In: C. Littlejohn and J. Turi, eds., Epistemic Norms: New Essays on Action, Belief, and Assertion, 1st ed. Oxford: Oxford University Press, pp. 59–79. Dancy, J. (1993). Moral Reasons. Oxford: Blackwell. Dancy, J. (2000). Practical Reality. Oxford: Oxford University Press. Dancy, J. (2004). Enticing Reasons. In: R. Wallace, P. Pettit, S. Scheffler and M. Smith, eds., Reason and Value: Themes from the Moral Philosophy of Joseph Raz, 1st ed. Oxford: Clarendon Press, pp. 91–118. Dancy, J. (2008). On How to Act—Disjunctively. In: A. Haddock and F. Macpherson, eds., Disjunctivism: Perception, Action, Knowledge, 1st ed. Oxford: Oxford University Press, pp. 262–279.

162  Constantine Sandis Dancy, J. (2014). On Knowing One’s Reasons. In: C. Littlejohn and J. Turi, eds., Epistemic Norms: New Essays on Action, Belief, and Assertion, 1st ed. Oxford: Oxford University Press, pp. 81–96. Davidson, D. (2001a). Actions, Reasons, and Causes. In: D. Davidson, ed., Essays on Actions and Events, 2nd ed. Oxford: Clarendon Press, pp. 3–19. Davidson, D. (2001b). Freedom to Act. In: D. Davidson, ed., Essays on Actions and Events, 2nd ed. Oxford: Clarendon Press, pp. 63–82. Davidson, D. (2001c). Hempel on Explaining Action. In: D. Davidson, ed., Essays on Actions and Events, 2nd ed. Oxford: Clarendon Press, pp. 261–276. Davidson, D. (2001d). Intending. In: D. Davidson, ed., Essays on Actions and Events, 2nd ed. Oxford: Clarendon Press, pp. 84–102. Davis, W. (2003). Psychologism and Humeanism. Philosophy and Phenomenological Research, 67(2), pp. 452–459. D’Oro, G. and Sandis, C. (2013a). Reasons and Causes: Causalism and AntiCausalism in the Philosophy of Action. London: Palgrave Macmillan. D’Oro, G. and Sandis, C. (2013b). From Anti-Causalism to Causalism and Back: A  History of the Reasons/Causes Debate. In: G. D’Oro and C. Sandis, eds., Reasons and Causes: Causalism and Anti-Causalism in the Philosophy of Action, 1st ed. London: Palgrave Macmillan, pp. 7–48. Dray, W. (1957). Laws and Explanation in History. Oxford: Clarendon Press. Dworkin, R. (1972). The Jurisprudence of Richard Nixon. The New York Review of Books, 18(8), pp. 27–35. Ezcurdia, M. (1988). The Concept-Conception Distinction. Philosophical Issues, 9, pp. 187–192. Falk, W. (1986a). Ought and Motivation. In: W. Falk, ed., Ought, Reasons, and Morality: The Collected Papers of W.D. Falk, 1st ed. Ithaca: Cornell University Press, pp. 21–41. Falk, W. (1986b). Action-Guiding Reasons. In: W. Falk, ed., Ought, Reasons, and Morality: The Collected Papers of W.D. Falk, 1st ed. Ithaca: Cornell University Press, pp. 82–98. Gallie, W. (1956). Essentially Contested Concepts. Proceedings of the Aristotelian Society, 56(1), pp. 167–198. Hacker, P. (2009). Agential Reasons and the Explanation of Human Behaviour. In: C. Sandis, ed., New Essays on the Explanation of Action, 1st ed. London: Palgrave Macmillan, pp. 75–93. Hacker, P. (2018). The Passions: A  Study of Human Nature. Oxford: Wiley-Blackwell. Hanfling, O. (2000). Philosophy and Ordinary Language: The Bent and Genius of Our Tongue. London: Routledge. Hart, H. (1961). The Concept of Law. Oxford: Clarendon Press. Hornsby, J. (2008). A Disjunctive Conception of Acting for Reasons. In: A. Haddock and F. Macpherson, eds., Disjunctivism: Perception, Action, Knowledge, 1st ed. Oxford: Oxford University Press, pp. 244–261. Hursthouse, R. (1991). Arational Actions. Journal of Philosophy, 88(2), pp. 57–68. Hyman, J. (2011). Acting for Reasons: Reply to Dancy. Frontiers of Philosophy in China, 6(3), pp. 358–368. Lalumera, E. (2014). On the Explanatory Value of the Concept-Conception Distinction. Rivista Italiana di Filosofia del Linguaggio, 8(3), pp. 73–81.

Are Reasons Like Shampoo?  163 Littlejohn, C. (2012). Justification and the Truth Connection. Cambridge: Cambridge University Press. Lukes, S. (1974). Power: A Radical View. London: Palgrave Macmillan. Marcus, G. (2012). Rational Causation. Boston: Harvard: Harvard University Press. Margolis, E. and Laurence, S. (2014). Concepts. In: The Stanford Encyclopedia of Philosophy [online] Spring, 2014 ed. Available at: https://plato.stanford. edu/archives/spr2014/entries/concepts/ [Accessed Jan 31, 2019]. McNaughton, D. and Rawling, P. (2004). Duty, Rationality, and Practical Reasons. In: A. Mele and P. Rawling, eds., The Oxford Handbook of Rationality, 1st ed. Oxford: Oxford University Press, pp. 110–131. O’Neill, M. (2014). Ludwig Wittgenstein. In: D. Reidy and J. Mandle, eds., The Rawls Lexicon, 1st ed. Cambridge: Cambridge University Press. Pettit, P. (2001). A Theory of Freedom: From the Psychology to the Politics of Agency. Cambridge: Polity Press. Piller, C. (2006). Content-Related and Attitude-Related Reasons for Preferences. Royal Institute of Philosophy Supplement, 59, pp. 155–181. Raz, J. (1990). Practical Reason and Norms, 2nd ed. Princeton: Princeton University Press. Raz, J. (2009). Reasons: Explanatory and Normative. In: C. Sandis, ed., New Essays on the Explanation of Action, 1st ed. London: Palgrave Macmillan, pp. 184–202. Rawls, J. (1971). A Theory of Justice. Cambridge: Belknap Press. Ruben, D-H. (2009). Con-Reasons as Causes. In: C. Sandis, ed., New Essays on the Explanation of Action, 1st ed. London: Palgrave Macmillan, pp. 62–74. Ryle, G. (1949). The Concept of Mind. London: Hutchinson. Sandis, C. (2006, June). The Explanation of Action in History. Essays in Philosophy, 7(2), article 12. Sandis, C. (2012a). The Things We Do and Why We Do Them. London: Palgrave Macmillan. Sandis, C. (2012b). The Objects of Action Explanation. Ratio, 25(3), pp. 326–344. Sandis, C. (2013). Can Action Explanations Ever Be Non-Factive? In: B. Hooker, M. Little and D. Backhurst, eds., Thinking about Reasons: Themes from the Philosophy of Jonathan Dancy, 1st ed. Oxford: Oxford University Press, pp. 29–49. Sandis, C. (2015a). Verbal Reports and ‘Real’ Reasons: Confabulation and Conflation. Ethical Theory and Moral Practice, 18(2), pp. 267–280. Sandis, C. (2015b). One Fell Swoop: Small Red Book Historicism Before and After Davidson. Journal of the Philosophy of History, 9(3), pp. 372–392. Sandis, C. (2016). Period and Place: Collingwood and Wittgenstein on Understanding Others. Collingwood and British Idealism Studies, 22(1), pp. 171–198. Sandis, C. (2017). Family Resemblance. The Philosopher’s Magazine, 78(3), pp. 17–19. Sandis, C. (2018). Robot Reasons. The Philosopher’s Magazine, 79(1), pp. 22–23. Sandis, C. (n.d.). Action Explanation for Robots, work in progress. Schueler, G. (2003). Reasons and Purposes. Oxford: Clarendon Press. Sehon, S. (1997). Deviant Causal Chains and the Irreducibility of Teleological Explanation. Pacific Philosophical Quarterly, 78(2), pp. 195–213.

164  Constantine Sandis Sehon, S. (2010). Teleological Explanation. In: T. O’Connor and C. Sandis, eds., A Companion to the Philosophy of Action, 1st ed. Oxford: Wiley-Blackwell, pp. 121–128. Skow, B. (2016). Reasons Why. Oxford: Oxford University Press. Smith, M. (1987). The Humean Theory of Motivation. Mind, 96(381), pp. 36–61. Smith, M. and Pettit, P. (1997). Parfit’s “P”. In: J. Dancy, ed., Reading Parfit, 1st ed. Oxford: Blackwell, pp. 71–95. Stout, R. (1996). Things that Happen Because they Should. Oxford: Oxford University Press. Wiland, E. (2012). Reasons. London: Continuum. Williams, B. (1981). Internal and External Reasons. In: B. Williams, ed., Moral Luck, 1st ed. Cambridge: Cambridge University Press, pp. 101–113. Wittgenstein, L. (1980). Remarks on the Philosophy of Psychology. Vol. 2. Oxford: Blackwell. Wittgenstein, L. (1981). Zettel, 2nd ed. Oxford: Blackwell. Wittgenstein, L. (2009). Philosophical Investigations, 4th ed. Oxford: WileyBlackwell.

Part II

Causal vs. Teleological Explanation in Historiography

9 Counterfactual Causality and Historical Explanations Doris Gerber

Introduction It is an obvious fact that we often use in ordinary language and in everyday situations causal terms if we want to explain something—for example why a person has done this or that. And it is undeniable that also historians employ a causal language in their scientific explanations. For example, they discuss the causes of the First World War or the effects of a certain economic development during a certain period of time or the consequences of a specific decision that was settled on by a leading politician. Moreover, historians use such terms, namely “cause”, “effect”, and “consequence” in a way that it is clear that they actually want to speak about causal relations. However, whether the usage of a causal language really manifests a casual explanation depends obviously on the very concept or understanding of causality that is presupposed. Since the publication of Carl Hempel’s provocative article “The Function of General Laws in History” in 1942 and up to this day historians (and not only historians but also philosophers) often think that the so called deductive-nomological model of explanation, developed by Hempel, Oppenheim, and Popper, is the paradigmatic model of a causal explanation. According to this model an event is explained by deducing the sentence that describes it from, first, sentences that describe initial conditions and, second, sentences that express relevant general laws. It is, however, actually surprising that this model is often seen as the paradigmatic case of a causal explanation, for two reasons: First, in early critical discussions of Hempel’s thesis that also historians—like every other scientist—use general laws in explanations, it was emphasized that historical explanations do have or can have a causal but no nomological structure. For example, William Dray contends that in historiography the application of general and strict laws is neither necessary nor sufficient. According to Dray it is not necessary because non-strict generalizations are available; and it is not sufficient because Hempel’s model of explanation cannot provide any reference to

168  Doris Gerber rational reasons (Dray 1957, ch V; Dray 1974: 69–70.). Other early critics are for example Michael Scriven (1959) or Georg Hendrik von Wright (1971). Dray’s critical objection, however, was partially anticipated by Hempel himself (although of course not in Dray’s spirit), because he argued that historians often only implicitly refer to general laws and that some of these laws are rather trivial. One example for such a trivial law would be: People that have a job do not want to lose it. Thus, according to Hempel, what historians can provide typically is just a sketch of an explanation but no full nomological explanation—and it was this thesis that makes his article so provocative for historians, because it implies that historical explanations are typically insufficient. The second reason why it is surprising that the deductive-nomological model is often seen as a paradigmatic model of a causal explanation is that according to Hempel a causal explanation is no more than a special case of a nomological explanation. Thus causation is subsumed to lawlikeness and not vice versa and therefore Hempel explicitly said that to speak of causes and effects is redundant at best. However, it is a wellknown problem of the deductive-nomological model of explanation that there are laws that can provide a valid deduction according to the model but this deduction does not explain anything. By referring for example to trigonometric laws one can deduce the length of a flagpole from the position of the sun and the length of the flagpole’s shadow but the sunlight and the shadow is no explanation for the length of the flagpole. And, to mention another well-known example, one can deduce stormy weather by referring to the mere temporal regularity between the barometer reading and the occurrence of a storm, but this regularity provides no explanation at all. Such problems hint at the fact that causality cannot be subsumed to lawlikeness because they hint at the fact that causality has a special point beside the point of regularity. In other words: The supposition that every singular causal relation is an instantiation of a causal law, that has its origins in David Hume’s regularity theory of causality, is simply false, or so I want to argue. In the first part of my paper, I want to give a very general argument against the assumption that causality can be reduced to (strict) regularity; this argument stresses the point that causality has essentially something to do with change and transformation and that the concept of change is more fundamental than the concept of regularity. The counterfactual analysis of causation is more accurate to meet this crucial feature of causation. In the second part I will introduce very briefly the concept of action I am relying on and deal with the more specific problem of mental causation. And in the third and last part I turn to specific aspects of historical explanations, namely the fact that historians are typically interested in rather far-reaching consequences of human actions and the fact that the so-called historical meaning of an event can change over the course

Counterfactual Causality  169 of time. These aspects seem to challenge the assumption that historical explanations have a causal structure, however, one can embrace these aspects by modifying the general counterfactual analysis in a minor way.

Causality and Regularity History is the history of human actions. Moreover, in my opinion one can argue that historical events are action events and that certain histories are causally and temporally structured actions. Even if one would contend, contrary to this opinion, that social structures or cultural traditions or that also natural events have an explanatory function in historiography independent of the explanatory function of human actions, it should be uncontroversial nevertheless that history is at least partly constituted by actions, namely, individual, social, and collective actions. I will say later a little bit more on my concept of action and the structure of causal action explanations. First, I want to examine the conceptual relation between causality and regularity. In the context of his discussion about freedom and necessity David Hume has argued that we can observe in human actions and more general in all human affairs the very same extent of regularity than in nature (Hume 1911: 115). Moreover, regularity has to be understood here as connected with necessity, because according to Hume the difference between causal and mere temporal regularities is the fact that only causal regularities are necessary regularities. However, this causal necessity is justified by our perceiving abilities and our experiences and thus it is highly controversial how Hume’s concept of necessity actually has to be understood. But independent of this problematic point it is not controversial that Hume reduced causality to regularity. Very sketchily, his argument runs as follows: All our knowledge is grounded on experience and observation and because we can observe in causal connections nothing else than regular connections between types of events, causality is nothing else than regularity. I think that all nomological theories of causality are based on such a thesis of reduction, even if they would understand necessity in a more realistic manner than Hume. In my opinion, however, this reduction thesis does miss  the point of causality very fundamentally. The point of causality is change and the concept of regularity presupposes the concept of change and not vice versa. That certain types of events follow one another other in a regular way implies that something changes in a regular way. An event of type A is followed by an event of type B. But by the contention that it is a regular change, the concept of change itself is not at all explained but presupposed. Even if we would suppose that actually all change is regular change, we would not be able to understand the concept of change on this supposition alone. Regularity is a possible attribute of change and one cannot explain a concept by one of its attributes.

170  Doris Gerber But how can we understand the concept of change? Moreover, how can we understand the concept of a causal change? Causality essentially implies change and regularity does miss the point of causality because it cannot explain this essential feature of causality—but what is the special sense of causal change? David Lewis has emphasized that we can understand causality if we express causal connections by counterfactual or subjunctive conditionals. In his argumentation in favor of the counterfactual analysis he wrote: “We do know that causation has something or other to do with counterfactuals [. . .] We think of a cause as something that makes a difference, and the difference it makes must be a difference from what would have happened without it” (Lewis 1986: 160–161). Lewis also emphasized that Hume himself already referred to this alternative approach to causality. Hume wrote in his famous definition of a cause: “we may define a cause to be an object, followed by another, and whether all the objects similar to the first are followed by objects similar to the second. Or in other words where, if the first object had not been, the second never existed” (Hume 1902: 79). Nowadays, there is agreement on the point that Hume’s “other words” actually did not introduce any synonymous formulation to the first mentioned regularity thesis but instead define a very different concept, namely the counterfactual concept of causality that implicitly questioned his regularity thesis. It is the counterfactual account of a cause that can give observable regularities a causal sense first of all, because such an account defines a cause directly as the origin of a change: If a specific singular event A had not occurred then another singular event B would not have occurred. A is the origin of the change manifested by the occurrence of B. The counterfactual concept of cause can be characterized by the following necessary and sufficient conditions: A is a cause of B iff 1. A occurred and B occurred. 2. If A had not occurred, but everything else being equal, then B would not have occurred. Like the regularity thesis this analysis follows a reductionist program: A is a cause of B if the relevant counterfactual conditional is true. However, first, it should be clear that not every kind of counterfactual or subjunctive conditional expresses a causal relation between events. For example, conditionals that express logical relations between concepts rather than factual relations between events are excluded. And, second, according to this analysis nothing at all is presupposed concerning the questions whether causal regularities between types of events can be observed or whether such regularities are assumed for some metaphysical reason. It simply depends on the types of events whether we are able

Counterfactual Causality  171 to observe regularities or not; and it depends on our metaphysical convictions whether we think that the structure of our world is characterized by nomological relations or not. According to the regularity thesis every singular causal relation is, in virtue of a conceptual necessity, the instantiation of a causal law. According to the counterfactual analysis this conceptual necessity does not hold, however, there are perhaps causal laws. In other words: If we concede that a singular causal relation can be expressed by the relevant counterfactual, causality does not necessarily imply regularities—but such (strict or non-strict) regularities are not excluded likewise. Thus, to contend that human actions can have a causal explanation does not imply that there are laws of action in any sense. Hume’s thesis that human affairs show the very same extent of regularity like natural events has to be revised at least. Of course there are some regularities and therefore action explanations imply the reference on some kinds of generalizations, however, it is obviously false to assume that there are strict regularities.

Causality and Mental Causation The problem of lawlikeness is one challenge for causal action explanations; another challenge is the problem of mental causation. This problem arises because in the philosophy of science it is widely accepted that the physical world is causally closed, i.e. that every physical event has a sufficient physical cause. If we want to avoid at the same time systematic causal overdetermination of physical events it is questionable how it is possible that actions are caused by mental events like beliefs or intentions. Can we solve this problem by appealing to the counterfactual theory of causality? Before I turn to this question, I want to say something on the concept of action I am holding. Following the objections against the belief-desire model of action explanation for example by Michael Bratman (1987) or Alfred Mele (1992), I  suppose that actions are typically caused by intentions and that intentions are specific and irreducible mental states. Intentions in turn are conceptualized against the background of a theory of intentionality. Such a theory supposes that intentionality is an essential and discriminating feature of all mental states. That means that the directedness of mind is an essential and distinguishing feature: All mental states are directed onto something else, they are about something else, they deal with something else. Originally, the concept of intentionality is a medieval concept that was introduced into contemporary philosophy by Franz Brentano. Brentano argued that intentionality is one essential feature that distinguishes psychological phenomena from physical phenomena. He wrote: “In der Vorstellung ist etwas vorgestellt, in dem Urteile ist etwas anerkannt oder verworfen, in der Liebe geliebt, in dem Hasse gehaßt, in dem Begehren begehrt usw” (Brentano 1973: 125). The English edition

172  Doris Gerber translates this sentence in the following way: “In presentation something is presented, in judgment something is affirmed or denied, in love loved, in hate hated, in desire desired and so on” (Brentano 1995). The existence of a mental state implies that there exists something else and the mental state is directed towards it. That is, mental states are intentional states that have intentional objects and in virtue of being directed at their objects they do have an intentional content. In my opinion not only concrete objects, like persons or tables, but also fictitious or abstract objects, for example Hamlet or mathematical sets, can be intentional objects; moreover, if the mental state is not directed at a single object but at an event, the intentional object could be a state of affair or an event in the past, in the present or in the future. But what about “directedness”? What does it mean to say that the intentional state is directed onto an intentional object? Directedness should be understood as a relation, more exactly as a relation of representation, i.e., the content of an intentional state is a representation of the intentional object. This representational content has to be distinguished from the so-called mode of the intentional state, namely, whether this state is a belief, a desire, an emotion, or an intention. We now can apply these conceptual theses concerning the philosophy of mind to the concept of intention. Like other mental states intentions possess the feature of intentionality and that means that they have an intentional object and an intentional content and this content is a representation of the object. The intentional object of an intention is the intended action as an event in the future and the intentional content of an intention is a conceptual representation of this future event, namely the intended action. My intention, for example, to take the umbrella with me is a conceptual representation of this action as an action that will be executed by me. The content of this intention would be described for example by saying: “I  better take the umbrella”. And if the represented action really was executed, we can say that the intention caused the action. Intentions are causally connected to each other and intentions are causally connected to other mental states, especially to beliefs and wishes. And these other mental states, in turn, are connected to and influenced by external states of affairs and events. Everything that belongs to this mental and external context of an action could be more or less relevant if you want to explain the action. It depends on the focus of one’s everyday interest or one’s research interest. In every case, however, the identification of the intention that caused the action will be an essential step in the explanation of an action. Beside such individual intentions that represent an individual action there are also social and collective intentions. I take it that social actions are executed in situations that imply activities between two or more

Counterfactual Causality  173 persons but social actions have to be distinguished from collective actions in the proper sense. An example for a social action would be: You are ordering a meal in a restaurant; and an example for a collective action would be: Three persons are carrying a piano upstairs. Whereas the representation of social actions provokes no special problem of analysis, so called collective or we-intentions are categorically different from so called I-intentions insofar as they do not cause directly a collective action by representing it. Rather, collective intentions have to be conceptualized principally as prior intentions, i.e. as intentions that cause other intentions which in turn cause a partial action that partly constitute a collective action. There is no space to discuss the problems of collective actions here. I only want to stress the point that I think that collective actions are a real and irreducible phenomenon of the social world and therefore also a crucial part of historical explanations. Collective actions are indirectly caused by collective intentions and collective intentions can be integrated in a model of an intentional and casual explanation very well. I  have proposed to analyze collective intentions in the following way (Gerber 2012: 258): The intention “We-intention (that we do Y)” is a collective intention if and only if: 1. The intention represents a collective We as the collective We that does Y. 2. The intention has the relational property to represent the same Y like the intentions of those persons that constitute the collective We that is represented in the intention as the collective We that does Y. 3. The intention is connected with the constant readiness to cooperate. Let me now come back to the problem of causality and mental causation. This problem is twofold, at least. First, one has to show how it is possible that some mental event actually can be causally efficacious; and second, one has to show how such mental efficaciousness can be compatible with the thesis that the physical world is causally closed and with the thesis that a systematic causal overdetermination should be avoided in scientific explanations. I think that the suspicion that mental events or mental properties cannot be causally efficacious only evolves if it is supposed that causality is manifested in some kind of a causal mechanism or power. However, what should a general causal mechanism consist of? To suppose such a mechanism is tantamount to the senseless attempt to search a cause for a cause

174  Doris Gerber and these two causes are efficacious exactly at the same time and place. Either this means that we only use different levels in scientific theory, and this would be of course possible and unproblematic, or there is actually the probability to get into an infinite regress. Consider, for example, the proposal that causality implies energy transmission. Now the question concerning some singular causal relation between events surely would be: What is the cause for this supposed energy transmission? And if we find some cause for this cause this would not be the end of the story. And so on, possibly ad infinitum. I take it that the counterfactual concept of causation can prevent such useless considerations. Like the regularity thesis the counterfactual analysis also pursues a reduction program: Causal dependence is reduced to counterfactual dependence, i.e. the criterion for the fact that A is a cause of B is the truthfulness of the relevant counterfactual. A  is a cause of B means: The counterfactual conditional “If A  had not occurred then B would not have occurred” is true. Thus the criterion of causality is a semantic criterion. Of course, this does not turn the causal relation into a semantic one, but it implies that any questions concerning causal mechanisms or powers is senseless from the very beginning. And therefore there is no special problem regarding the causal efficaciousness of mental events. However, this line of argumentation cannot solve the second aspect of the problem of mental causation that I have mentioned. Even if we concede that mental events are really able to cause physical events, i.e. even if we contend that there is the possibility of an interaction between mental and physical events, in order to preserve the thesis that the physical world is causally closed, it would not be enough to contend that the same physical event can have a mental and a physical cause at the same time. Because in scientific explanations we want to avoid systematic causal overdetermination; an event can have several causes in the sense that two or three or more events are only together sufficient to cause this certain event. But if we allow that an event can have two different but likewise sufficient causes then we must concede that no one is necessary and this in turn would be identical with the judgement that there is no cause at all. However, scientific explanations depend on some or other scientific theory. Therefore, the thesis that we should avoid causal overdetermination has to be interpreted in a way that observes this dependence. In other words: Only in the context of a certain theory such an overdetermination should be avoided. But the historical and causal explanation of an action as an historical action belongs to a very different theory than the explanation of, say, the bodily movements of an assassin shooting his victim. Thus the methodological rule that causal overdetermination would question the explanatory force of every explanation is no real challenge for the possibility of mental causation.

Counterfactual Causality  175

Causality and Historical Meaning Until now the discussion concerns general problems of action explanations. However, there is also a special problem that emerges in the context of historical explanations. Historians are typically interested in rather far-reaching consequences of actions and, moreover, they are interested in the relevance of an action or a complex of actions in regard to the future course of events in the context of a certain history. Events are not explained one by one, but events are set into relation not only to temporally earlier but also to temporally later events in order to explain these events. In other words: Historians are trying to explain the historical meaning of an event, for example the historical meaning of the assassination of Sarajevo or the historical meaning of Putin’s decision to annex the Crimea or the historical meaning of chancellor Merkel’s decision to open frontiers for refugees. Concerning the last example one question for future historians would be: Has this decision the historical meaning to be responsible for the success of the radical right-wing party Alternative für Deutschland (AfD) in the elections for the Bundestag in September 2017? However, why does this special interest provoke a special problem concerning the concept of causation? We can see this by considering a puzzling example described by Danto in the context of his distinction between a history and a so-called Ideal Chronicle. An Ideal Chronicle entails about every event every possible piece of truth and information that can be transcribed in the moment it happens. This means that the Ideal Chronicle describes every event in full detail but without reference to earlier or later events. Such a Chronicle is both very rich and very poor and it seems to be clear why a Chronicler’s transcription of happenings cannot tell a history: Histories essentially represent the relations between events, describing events not one by one, but events as within their relations. It is exactly this essential property of histories that Danto’s fictional Ideal Chronicle cannot possess. Danto’s puzzling example of two scientists formulating the same theory independent of each other articulates these conceptual correlations: Suppose, for example, that a scientist S discovers a theory T at t-1. S perhaps does not publish T. At some later time t-2, a different scientist S* independently discovers T, which is now published and taken into the body of accepted scientific theories. Historians of science subsequently find out that S really hit on T before S*. This need take away no credit from S*, but it allows us to say, not merely that S discovers T at t-1, but that S anticipated at t-1 the discovery by S* of T at t-2. This will indeed be a description of what S did at t-1, but it will be a description under which S’s behaviour could not have been witnessed and it will be an important fact about the event which accordingly fails to get mentioned by the Ideal Chronicle. (Danto 1965: 155–156)

176  Doris Gerber The puzzling issue in this example is the fact that the first event, the formulation of T by the first scientist, S, seems to acquire a new property, the property of being the anticipation of T, in virtue and only in virtue of the occurrence of a later event, namely, the formulation of T by the second scientist, S*, at t-2. At t-1, when S discovers T, this act of discovering is still not an anticipation. It only becomes an anticipation when S* rediscovers T. And the problem exactly is: How can we explain this very fact causally without assuming backward causation? Danto’s discussion of the example and its implications is not very clear. The question he asks is: Does anything change with or happen to the earlier event E-1 at time t-2 when the second and later event happens? On the one hand Danto rejects explicitly backward causation but contends nevertheless that if a former event is a necessary condition for a later event, then it follows that the later event is a sufficient condition for the former event. On the other hand Danto’s solution to the puzzling example seems to be a switch from the factual to the descriptive level: What really changes at time t-2 is our description of the event at time t-1. After the happening of the later event we can describe the former event as an anticipation of a theory. However, in my opinion this solution is too trivial. It is surely right that our description of earlier events gets more and more rich in the course of time, simply because of what we know about what happened afterwards. But it is also surely right that the truth of our descriptions depends on the facts and not vice versa. That means, it depends on the fact that at t-2 the very same theory was formulated again that we can describe the earlier formulation of this theory as an anticipation of it. Before the occurrence of the second formulation this description would not be true. Moreover, even if no one would ever discover that the already published theory had been formulated some hundred years ago, to be an anticipation of a later published theory would nevertheless be a property of the earlier event at and since time t-2. The historical meaning of an event is a real property of the event independent of the richness or poorness of our knowledge and our descriptions. However, how can one explain the possibly changing historical meaning of events causally without assuming backward causation? I want to propose to revise the counterfactual analysis of causality in two respects. First, the time of the occurrence of the respective events is mentioned in the formulation of the conditions. This revision shall exclude backward causation and make the entire proposal more adequate in regard to historical explanations because in history the time of an event’s occurrence can be a very important fact. Second, I will allow calling an event a cause if its efficacious force only concerns particular properties of the effected event and not the occurrence of the other event itself. The consequence of this second revision is that the temporally first event in Danto’s example

Counterfactual Causality  177 is a cause of the temporally later event and not vice versa. Here is the formulation of the conditions: A is a cause of B iff: 1. A occurred and B occurred. 2. A occurred at time t-1 and B occurred at time t-2, i.e. A and B are standing in a temporal relation to each other and A occurred earlier than B. 3. If A had not occurred at time t-1, but everything else being equal, then the following holds: Either (a) B would not have occurred at time t-2, or (b) there is at least one essential property of B, which B would not have possessed, that is, C would have occurred. 4. If (b) in condition (3) is the case, then it also holds that A and C would stand in the same temporal relation as A and B. According to this analysis, the earlier event in Danto’s example can be seen as a cause of the later event because condition (b) in (3) is met. For, the later event would not be a rediscovery of a theory if the earlier event would not had happened. The earlier event is causally responsible for the later event having a particular essential property. In this sense, and only in this sense, the earlier event changes its causal properties at time t-2 actually. This means that the later event is causally dependent on the earlier event because the following counterfactual conditional is true: If E-1 at time t-1 had not occurred, then E-2 at time t-2 would not have had the property of being a rediscovery. In this sense, and only in this sense, E-1 is a cause of E-2. One may object to my discussion and solution of the problem to explain the historical meaning of an event that Danto’s example is actually very puzzling. And this objection is both right and wrong. Ordinary action explanations are indeed explanatory by identifying the intentions and other mental states that cause the action and have usually no explanations that imply relations between such events like E-1 and E-2. However, historical explanations are dealing with stories and not with single actions, i.e. they are dealing with the causal connections between actions that constitute a story by their temporal and causal relations. The concept of the historical meaning of an historical event covers exactly this special feature. E-1 and E-2 are parts of one story, namely the story that one may call The discovery of T. Thus concerning historical explanations Danto’s example is, puzzling as it is at first glance, rather ordinary.

References Brentano, F. (1973). Psychologie vom empirischen Standpunkt. Hamburg: Meiner.

178  Doris Gerber Brentano, F. (1995). Psychology from an Empirical Standpoint. London: Routledge. Bratman, M. (1987). Intention, Plans, and Practical Reason. Cambridge: Harvard University Press. Danto, A. (1965). Analytical Philosophy of History. Cambridge: Cambridge University Press. Dray, W. (1957). Laws and Explanation in History. Oxford: Oxford University Press. Dray, W. (1974). The Historical Explanation of Actions Reconsidered. In: P. Gardiner, ed., The Philosophy of History, 1st ed. Oxford: Oxford University Press, pp. 66–89. Gerber, D. (2012). Analytische Metaphysik der Geschichte. Handlungen, Geschichten und ihre Erklärung. Berlin: Suhrkamp. Hempel, C. (1942). The Function of General Laws in History. In: G. Hempel, ed., Aspects of Scientific Explanation: And other Essays in the Philosophy of Science, 1st ed. New York: Free Press 1965, pp. 231–243. Hume, D. (1902). An Enquiry Concerning Human Understanding. Chicago: Open Court Publishing. Hume, D. (1911). A Treatise of Human Nature. Vol. 2. London: J. M. Dent & Sons. Lewis, D. (1986). Philosophical Papers. Vol. 2. Oxford: Oxford University Press. Mele, A. (1992). Springs of Action: Understanding Intentional Behaviour. Oxford: Oxford University Press. Scriven, M. (1959). Truisms as the Grounds of Historical Explanations. In: P. Gardiner, ed., Theories of History. New York: Free Press, pp. 443–475. Von Wright, G. (1971). Explanation and Understanding. Ithaca: Cornell University Press.

10 Beyond Causalism and Acausalism1 Harold Kincaid

The debate around causalism-acausalism about human action has not been entirely fruitful, at least seen from my naturalist approach which is suspicious of both conceptual analysis and appeals to intuition and which proposes instead to look at broadly empirical criteria for evaluating the debate. This chapter thus tries to sketch what a naturalist approach to the debate looks like. The chapter is in two parts. Section 1 argues that the debate over reasons as causes cannot be resolved by analyzing concepts and testing them against intuitions and imaginary cases, that the behavior/action distinction is not some fundamental conceptual truth but a distinction of value only where and when cognitive science provides good empirical evidence supporting it, that appeals to reasons as causes are best justified if they fit into empirically well-supported results, that results across the cognitive sciences suggest that much human behavior is best explained by processes that do not look anything like reasons, creating problems for causalists and acausalists alike, that traditional acausal accounts border on mystical metaphysics, and, nonetheless, there is an approach based on work by Dennett and on revealed preference theory in microeconomics that can make sense of a certain kind of acausal story about human action. Section 2 elaborates this story by looking at causalism/acausalism approaches to social entities. The social sciences standardly treat macrolevel entities—states, bureaucracies, and interest groups for example—as agents in the same sense that humans are studied. Here, a causal story—that these social entities have reasons that causally explain their actions—seems highly implausible. There are some heroic attempts to defend this view (e.g. List and Pettit 2011), but in the end this approach is a large stretch, at least if you see it as trying to tell a causalist story, for it is hard to see what in the world would count as a group reason that explained the actions of collective entities. My Dennettian, revealed preference approach, however, makes good sense here. We need to ask: Do the choices of such groups make for a consistent utility of functions? Do these functions allow us to find explanations—real patterns—describing their behavior?

180  Harold Kincaid The issues in the causalism-acausalism debate are large and complex, and the third way I propose introduces even more complexity. This chapter does not pretend to have all these facets sorted out; instead it should be seen as a suggestive foray into an alternative approach.

Section 1: Conceptual Issues and Naturalist Approaches It is useful to start with a brief statement of what I take the causalismacausalism debate to be, though readers of this volume are surely of course familiar with the debate. However, the disputes are not entirely homogeneous and well defined, so it will be useful to be somewhat specific. So, causalism claims that all human action is explained by the reasons of individuals standing in a certain causal relation to action— and if agents are to be part of the story, conscious reflective endorsement seems necessary (Aquilar and Buckareff 2010)—and acausalism denies that actions are to be causally explained by reasons and claims instead that human action is to be explained in noncausal terms by appeal to the purpose or what actions exist in order to do (Sehon 2005). There are many variants of these claims. However, my goals in this chapter are modest in that I want to sketch a naturalist alternative to the debate and I do not pretend to engage in all the complexities in the literature. I approach the causalism-acausal debate from a broadly naturalistic perspective. There are serious debates about what naturalism comes to and how it can be best be defended. But for my purposes, a restricted set of claims will do. First among them is skepticism about the value of traditional conceptual analysis in terms of necessary and sufficient conditions tested against ordinary language intuitions as a way to make progress in explaining human behavior. Of course, clarifying concepts can be important for scientific progress, but progress is most likely with concepts that have scientific, empirical ties. Analysis of ordinary language concepts devoid of scientific connections is an odd enterprise with dubious payoff for explaining the world. Thus, deciding whether causal or acausal accounts of human action are most justified is ultimately a question of asking what the best science shows.2 Obviously, there is no a priori reason that there must be one answer for all phenomena whether causalism or causalism is most plausible. Likewise, looking at the science may show that some standard philosophical formulations need revised—this is conceptual analysis worth doing. I should note that many participants in the causal-acausalism debate will reject the naturalism presumed here, so the arguments that follow can be seen as pursuing a project that asks what does a certain kind of naturalism say about the debate and what kind of alternatives might it suggest? In other words, my project asks, “if natural were true, what would then follow for this debate?”

Beyond Causalism and Acausalism  181 A full defense of naturalism is beyond the scope of this chapter. In general, the defense is worked out in the course of the development of certain strands of 20th-century philosophy. Quine’s arguments against the analytic-synthetic distinction; the difficulties in making sense of synthetic a priori truths; the findings from empirical psychology that ordinary concepts are generally best not understood on a necessary and sufficient conditions model; the findings in the history and philosophy of science that key concepts such as “gene” likewise do not fit that model either and that the theoretical and empirical parts of science are often intertwined; and many other considerations raised serious doubts about a first philosophy based on the intuitions and conceptual insights of philosophers.3 So from the naturalist perspective the idea that the causal-acausalism debate is not an empirical issue is a nonstarter.4 Fundamental to the traditional philosophical debate over causalismacausalism—going back decades—is that there is a distinction between action and behavior. Then the puzzle is to explain action. But the naturalist asks what kinds of scientific evidence ground this distinction and for what specific phenomena? Moreover, typical philosophical examples of explanation by reasons are on the order of “Alice crossed the room to get her coffee—did she do that because reasons and desires caused that or is the best explanation the teleological one that invokes meaning and an in order to type account?” Now, a great deal is known about the cognitive and neurobiological processes explaining behavior which suggest that a much more complex and nuanced picture of the determinants of behavior is called for than can be found in common sense philosophical accounts. This is not to say that the behavior/action distinction is never appropriate, but judgements that it is have to pay attention to what our best behavioral science (where that no doubt involves results across multiple science) tells us. These empirical findings raise doubts about the reasons as causes approach. The best cognitive science accounts of human behavior are making distinctions and citing processes that do not map in any clear way onto the common-sense belief-desire models. The issues here are similar to many of those raised in the long debate over folk psychology. Debating causalism-acausalism without considering those issues seems uninformative. Key results are the following.5 Unconscious processing in the causes of behavior suggest at least that the range of actions is much more limited than philosophers of action assume. Experimental studies show that the brain processes involved in behavior begin before there is any conscious sense of intending (Wegner 2002). Behavior is the result of a complex set of psychological/brain processes: “The real causal sequence underlying human behavior involves a massively complicated set of mechanisms. Everything that psychology studies can come into play to predict and explain even the most innocuous wink of an eye” (Wegner 2002:

182  Harold Kincaid 25). “Belief” and “desire” are extreme simplifications of complex brain and psychological processes in which conscious intent is not the driving force and may often just be a side product experience of will. “Desires” are complex configurations of attention, memory, cognition, and affect involving multiple regions of the brain (Banfield et al. 2004). So, claiming that an act was caused by reasons is many miles removed from the actual psychological processes. It is not obvious those processes even count as “reasons”. Furthermore, the complex interactions of various brain regions in producing behavior makes it hard to tell what scientific correlates would make the behavior/actions distinction scientifically useful. Perhaps a distinction between “normal” prefrontal cortex functioning and “abnormal” as in the case of addiction (see Ross et al. 2008 for a discussion) might give some sense to the actions-behaviors distinction, but even then, the neuroscience has sometimes been overinterpreted, with misleading metaphors of brain regions being like little muscles fighting it out (see Kincaid and Sullivan 2010). The upshot of all these findings is that we cannot take as a fundamental datum that humans act for reasons. The reality is much more complex. This conclusion also gains further support from the extensive work in experimental psychology and experimental behavioral economics. That work involves giving subjects tasks where they select between alternatives, almost always with their selections having a financial consequence in experimental economics and almost never in experimental psychology. From those selections, models are fitted describing latent factors that allow for predicting the observed selections, where the quality of model fit is judged by various statistical means. The latent factors are such things as risk aversion, ambiguity aversion, probability weighting, anchor points, confirmation bias, and so on. These latent factors are not conscious beliefs and direct questioning of subjects often shows that subjects would not express having beliefs about these factors or not be able to even make sense of the questions. The latent factors explain or at least predict their actions or behaviors—selections in tasks—without in any clear sense resembling conscious reasons.6 Let me elaborate on the argument I have just given. It is in schematic form: •

• •

The empirical literature from psychology, cognitive science, cognitive neuroscience, experimental behavioral economics, and so on, arguably shows for a wide range of behaviors/actions that the processes involved in decision making do not look anything like phenomena that could be described as conscious reasons. Both the causalist and acausalist think that conscious reasons explain action. Thus, the scope for either view is much more restricted than either view recognizes. Neither view is plausible for much human behavior.

Beyond Causalism and Acausalism  183 One might object that the argument I  have given begs the question in favor of the causalist, because the empirical literature I cite is about the causes of behavior. It does not beg the question in this way. The scientists who produce the work I  have appealed to are generally modest about making causal claims; instead they typically describe the factors invoked as “involved in” the behaviors in question. The work I  sketched from experimental psychology and behavioral economics is likewise seldom put in causal terms; investigators think of their work as finding predictive patterns. Generally, all this work can be given what philosophers of science call an “instrumentalist” interpretation, instrumentalism being roughly the well-known doctrine in the history and philosophy of science that the aims of science are to predict the observational data, not causally or otherwise explain them. Moving now to teleological views, from the naturalist perspective they look—well—teleological, and that means, from a scientific viewpoint, suspect. There are ways to make “in order” talk clear in the behavioral and social sciences without reasons as causes via selectionist stories a la Darwin (e.g. Kincaid 2006). Yet philosophical defenders of teleological approaches make no such efforts, and it is hard to see how they would work for the kind of explanations they want give. There is a large, unexplained metaphysical mystery in teleological accounts. However, I  think there are resources for making sense of a certain strain of acausalism coming from Dennett and revealed preference theory in economics. The basic idea derives from behavioralism, but not from Skinner-style psychological behaviorism but rather from a logical or philosophical behavioralism that claims that the main evidence for explaining behavior is observed behavior itself. The view here is subtle. Critics may think “subtle” is the wrong word and that vague and confused are better descriptors. I do not think so, but it takes some work to make this view coherent, and Dennett himself could certainly be clearer. However, I do not want to get involved in the minutiae of Dennett interpretation. So here is a Dennett-inspired set of ideas. Explaining behavior happens or, more sophisticatedly, can happen by finding patterns in observed behavior. Not just any pattern is explanatory, for it can be all too easy to find patterns, as standard statistical practice in the social sciences shows. Dennett (1991) says we want “real patterns”, patterns that are more efficient descriptions of the data than a bit map of the data. Still, if that formulation is not helpful, the basic idea is that we want patterns that do explanatory work. That is where the intentional stance comes in. Accounts in terms of belief, desires, and reasons from the intentional stance get us good descriptions of patterns in behavior—how one behavior is connected to another, how behavior changes over time and over circumstances, and so on. So how does this support acausalism? Here things get a bit tricky, and I want to depart from Dennett. Dennett was often accused of being

184  Harold Kincaid an instrumentalist (Lycan 1988). Ultimately—after some evasiveness—he explicitly denied that this was his view (Dennett 1991). Beliefs are real, but they are characteristics of the patterns of the behavior as a whole. I worry that this answer is insufficiently naturalist in that it still wants to give a conceptual account of what beliefs really are. Dennett (2009) explicitly endorses the conceptual analysis project for intentional systems theory but then seems to take it back recently (2018). Whatever his views, there are certainly those who see intentional systems theory as a possible route to the conceptual analysis of what beliefs, etc. really are. See, for example, Tollefsen (2015) who is explicit about this and thinks this is the enterprise we should be pursuing. For me, instead, we need to look at what the best science is committed to rather than giving a philosophical account of what beliefs really are. If in scientifically explaining observed behavior we invoke beliefs with good experimental evidence, for example, then we are committed to them.7 Yet this is not the philosopher’s question of the “true nature” of belief; asking about the true nature of belief is like asking what is the true nature of life—e.g., are cells “really alive?”, a question biologists long ago rejected as scientifically unhelpful and a priori. Thus, the point here from the naturalist perspective is that one way which belief-desire explanations can be supported is by looking at empirical applications to find patterns in observed behavior. This does not mean that beliefs and desires can be “reduced” (whatever that might mean—there are many notions of reduction) to behavior but only that one way to make useful sense of belief-desire notions may be by describing patterns of observable behavior. The charge of instrumentalism against the project of finding real patterns in behavior can be defused by simply pointing out that we can use them to explain. We can identify certain patterns that not only predict well but also can ground causal explanations. Grounding causal explanations is realism, not instrumentalism. What do these points have to do with causalism versus acausalism? For one, if we are explaining in terms of observed behavior, then these explanations are agnostic about internal psychological processes and thus do not require that behavior be caused by reasons, beliefs, desires, etc. So, there is no commitment to causalism. Is this approach a form of acausalism? That is a bit hard to tell, since acausalism as a positive thesis is not all that clear. Yet, let’s focus on the basic idea that we explain behavior in terms of what it exists in order to do something. One way to cash out that formulation is by saying that the behavior is the apt or the best way to do the goal in question. Put this way, then there are ways to describe “in order to” in terms of observed behavior. I am not sure exactly how Dennett proposes to capture “fit” or situation appropriate behavior from the intentional stance, but appeal to a presupposed notion of rationality seems part of the story. I have my suspicions about the rationality requirement for the

Beyond Causalism and Acausalism  185 standard reasons expressed in the literature about the principle of charity, about what rationality requires, etc. However, I  am not in the Dennett interpretation business and fortunately think I have another body of work to use that gives a much clearer picture—revealed preference theory. Revealed preference theory comes from economics as it developed over a considerable period of time dealing with the basis of neoclassical account of economic behavior. The key theory of consumer behavior— utility theory—initially rested on maximizing internal subjective states such as pleasure. A  major strand of neoclassical economics found this kind of foundations for utility theory troubling—probably unscientific— and as the discipline began to develop more formal rigor, there was a natural move away from these hedonist interpretations to models that focused solely on the formal properties of observed choices. This trend culminates in a way in Samuelson (1983) in what he labels “revealed preference theory”. This name is misleading. The formalism Samuelson produced was entirely applied to observed choice or even more accurately, observed selections in tasks. The project was not at all about identifying underlying psychological states such as preferences, contra what some commentators (Hausman 2011) think. Rather, the whole point for Samuelson was to get away from underlying psychological states. What Samuelson showed with refinements by others afterwards is that if a set of observed choices obey these axioms: • • • •

the selections are complete—there are no, noncomparable bundles, the selections are transitive—if A is chosen over B and B over C, then A is chosen over C, the selections are monotonic—more is chosen over less and there is never a saturation point, the selections are continuous—if y chosen over z and x is close enough to y, then x is chosen over8 z,

then they can be described as maximizing a utility function (this is called a representation theorem). These results can be elaborated in many directions—for example, by adding in risk and time attitudes, but where these “attitudes” are just further features of observed choices. Revealed preference theory as just described is used in all kinds of experimental and observational situations in economics. It is not just a descriptive theory of observed human choices. Given a utility function describing an individual, then it is possible to predict, and I would say causally explain, for example (though “causally” is not essential to the account), changes in choices given changes in other variables such as prices, budget constraints, and so on. This approach works for both human and animal behavior (e.g. the behavior of fish in coral reefs—see any text in behavioral ecology).

186  Harold Kincaid How does revealed preference theory connect to causalism versus acasualism issues? It is a more worked out and rigorous version of Dennett’s attempt to diffuse the issues. It is not causalist because it makes no presumptions about psychological states behind behavior (and recall my comment above that it is not a theory about preferences as psychological states). And, unlike Dennett, it is not committed to the idea that there are reasons, etc. somehow in the pattern of choices as “virtual” entities. Similarly, it is making no claim to do philosophical conceptual analysis of what reasons, beliefs, etc. really are. Moreover, it does not bring the baggage still found in Dennett of talking about “our practices” of attributing beliefs; it is based on scientific assessment of observed behavior. So revealed preference theory is an account of behavior with good scientific credentials that avoids internal psychological states such as reasons as causes and does not depend on making contentious conceptual philosophical presuppositions. What about acausalism? Revealed preference approaches provide a sense of acausalism akin to the way that intentional stance accounts did, but shorn of some questionable assumptions of the Dennettian approach. Given consistent choices, those choices can be described as existing in order to maximize a utility function given constraints. Like Dennett, the appeal to rationality is doing the work here, but for revealed preference theory rationality has a well-specified meaning—consistency under axioms—rather than the much less precise talk of fit. I now switch to another defense of acausalism of a rather different nature, admittedly a nebulous one but one I think of great importance and surprisingly still puzzling. The interpretist tradition as exemplified by Taylor’s (1971) classic paper has consistently denied that human action is to be causally explained because both explanation of human action and human action itself is about interpreting meanings. Taylor’s classic piece puts the point in criticizing the behavioralist tradition in political science as that there are no “brute data” in the human sciences. Even in 1970 when this was written it was already well understood that “brute data” was not a happy description for the natural sciences either—there are many interpretive steps in going from raw data to inferences about hypotheses. In fact, there are a host of ways in which naturalistic social science can deal with “meanings”, depending on how that term is cashed out. Meanings may refer to perceptions, beliefs, language interpretations, intentions, symbols, and norms among other things (Kincaid 1996). There is a variety of standard social scientific methods to study these components of meaning. That said, there still remains a deep issue that the acausalist should appreciate. Both human action and its interpretation by social scientists involves a set of categories that are needed to categorize behavior— dances, signals, taboo, threats, offers, and on and on. Those categories are an intricate set of interrelated ways of conceiving things that cannot

Beyond Causalism and Acausalism  187 plausibly be reduced to ideas in peoples’ heads (individual internal states such as beliefs and desires); they are much more distributed and fundamental than that. My naturalist proclivities make me think there must be ways of tying these down to empirical observations. Yet I  think it would be quite presumptuous to claim we know how to do that. In particular, social categorization is not just a “third variable”—a variable that is just added on as a separate cause in addition to the other two. It certainly can be a third variable, as the interpretations of meaning and the ways of investigating them above illustrate. Yet the questions are still deeper—where do the categorizations come from? How do we get outside them to assess them? Dennett’s talk of “virtual entities” is suggestive but only that. I do not think this is comfort for the hermeneutical tradition, because that tradition only points to the issues but has no coherent story that I can see about how we should proceed. So, for the acausalist position these points say that explaining human behavior/action has to go beyond reasons as causes, even if that mode of explanation has important uses. Things are more complicated than the causalist admits, but how those complications should be dealt with is a deep and open-ended question. The implications of this for doing historiography are pretty obvious I think. All the complex questions about how we approach current social structures and relations equally apply to understanding past societies.

Section 2: Explaining Collective Agents There is a large literature that in part parallels the causalism versus acausalism debate about individual humans transferred to debates about the behavior of collective entities such as organizations or groups. Most of the literature is actually quite individualistic, seeking to determine the conceptual conditions for individuals to act together, have collective intentions, and so on. There is much dancing around trying to find some sense in which groups have beliefs or are agents. I  will walk through some of this below, but from my naturalist stance this is not a happy exercise. If explaining individual behavior in terms of beliefs and desires is a fraught enterprise, then doing so for groups is even more so. And, much of this literature really is not about groups at all—it is about, for example, individuals walking together, etc. (cf. Gilbert 1994 and Bratman 2001). This work is very far from anything that connects to contemporary social science research. So here are some aspects of the complex literature trying to provide a causalist account of collective agent behavior. There are two fundamental issues running throughout this literature that concern whether real group agency—group belief, group preferences, etc.—is being accounted for rather than some notions of individual agency when individuals interact. A full-bodied account of group agency would refer to groups that

188  Harold Kincaid are somehow more than mere sums—more than just aggregates—but are real concrete entities. The University of Cape Town is a concrete social entity. It has a complex internal structure and can stand in causal relations among other things. The set of all those earning between R10,000 and R10,001 does not have these characteristics and is merely an aggregate. An analogue at the individual level would be the difference between a mass of human cells and an individual or a person. The second related issue is whether group agency is cashed out in such a way that it is really just certain kinds of beliefs, etc. of individuals or whether beliefs are really being attributed to the groups themselves. I  will return to these questions below when discussing the problems of various accounts of group agency and the superior alternative I favor. Here are some of the more prominent accounts of group agency or its elements: •

•

•

Toumela (1995) defines group belief as occurring when there are “operative” members—members with a special position to speak for the group—whose pronouncements are at least tacitly accepted by all the members of the group. Gilbert (1994) analyzes group belief in terms of “joint commitment” which entails norms and obligations among the members of the group about the belief in question—each individual is willing to commit to all the members of the group in question. List and Pettit (2011) provide a general theory of group agency that starts from a conceptual analysis of agency in general that requires motivations (in other words, desires), representations (in other words, beliefs), and a process whereby these cause actions. So, they assume causalism from the start. A group agent is formed when there is a joint intention to form a group around a shared goal. List and Pettit provide various constraints on shared goals and joint intentions that need to be met and try to show how various aggregation procedures meet those requirements.

Nonetheless, these accounts are a real stretch. There are multiple problems. A first is that if the goal is to give a conceptual analysis of notions such as group belief, intentionality, and cognition, then it is not clear they have done so, because their accounts are in terms of individual belief, not group belief (see Tollefsen 2015). A  second set of problems deals with the conceptual analysis itself. We have far weaker intuitions about what it means to say the group believed, intended, or knew than we do about those notions applied to individuals. There is a rich set of interconnections and practices around the attribution of folk psychological notions to individuals; there is no such thing for groups, and thus appeals to what we would say—almost always the evidence in these analyses—is only very loosely constrained. A further problem with these conceptual analyses is

Beyond Causalism and Acausalism  189 simply the one presented earlier on, namely, that there is no particular reason to think that conceptual analysis tested against intuition will provide good explanations by scientific standards. A final problem is one of evidence and localization. Exactly what in the accounts given here counts as the group belief and how do we go about finding it? So, what is the upshot for causalist approaches to groups as agents? For the causalist approach, it is clear: It is hard to see how to explain the behavior of groups as the results of their beliefs, desires, etc. We do not actually know how to identify group beliefs, and even if we could in principle, in practice it is hard to see how we could show that specific, determinate beliefs caused behavior. Thus, the naturalist alternative for collective entities seems rather obvious. Trying to figure out some way to attribute “real” beliefs and reasons to collective entities (Tollefsen’s [2015] overriding project) seems just misguided. What we can do is take the intentional stance/revealed preference theory approach to collective social entities, individuating the latter pragmatically according to what allows us to find real pattern and eschewing any attempt to give a necessary and sufficient analysis of “is a collective entity” as does the standard analytic literature on group entities. Thus, the question is whether we can do something like ascribe a utility function to the behavior of collective agents that shows they have consistent behavior (in the sense elaborated here) and if we can then use those utility functions to explain their behavior in terms of how they respond to budget constraints and other factors in their environment. In the case of collective entities, it is much harder to have evidence for the kind of objective choice behavior we can sometimes get with individuals through experimental setups that ground explicit revealed preference explanations. Often, we have maximizing utility models that have to be thought of more in the Dennettian style of intentional systems analysis. Here are two sample areas where collective entities are treated as agents by either the intentional stance or revealed preference approach and examples of how they work. The first is work on the theory of the firm from standard microeconomics; the second, rational choice and game theory models from the political science of international relations. There is a theoretically well-developed and empirically well-supported body of work in microeconomics on the theory of the firm. Firms are treated as unitary agents in a way parallel to the way consumers are treated in revealed preference theory: They have in effect a utility function where the argument is usually profit maximization but can also include other things. They face constraints similar to the consumer budget constraint but in the form of production possibilities and the costs of inputs and the externally determined prices of outputs. From observed choices of inputs and of prices for outputs it is possible to describe firm behavior in revealed preference fashion: The internal workings of the firm can be ignored and yet still the observed behavior explained as a function of

190  Harold Kincaid external constraints. Models can be made more complex by adding in the risk attitudes of the firm just as consumer models can. The model of individual firm behavior can also then be complicated by embedding it in game theory models such as those describing monopoly or oligopoly. The second illustrative example comes from rational choice and game theory approaches to international relations. A  rich literature explains international relations by treating states as actors and applying rational choice theory with game theory extensions (Kydd 2015). Wars, tariffs, treaties, and so on are explained by states maximizing their utility function subject to constraints. What goes into the utility function can be decided case by case—it may be power, security, territory, wealth, and other ends. More sophisticated models can allow for risk attitudes and time discounting. Then these utility functions are used in further analyses via game theory where interaction is modeled by defining a set of possible strategies and identifying relevant equilibria of the game. How well these models are supported by substantive data is an interesting question, but not one directly relevant here, for my point is only to illustrate approaches with implications for the causalism-acausalism debate at the group level. What morals should we now draw from the above discussions about the causalism-acausalism debate extended to collective agents? We have seen that it is hard to make sense of attributing beliefs, reasons, and so on to groups per se. (Tollefsen [2015] is a detailed and well-argued illustration of these difficulties from the perspective of the traditional analytic project.) I think these difficulties are entirely obvious, but if they are not, just ask yourself how you would go about making sure of and providing evidence for the claims that: (1) Volkswagen (the company, not some specific individual) believed that it could go undetected in its efforts to deceive regulators? Or (2) The University of Cape Town believed that accommodation was a preferred strategy over confrontation in dealing with #rhodesmustfall demonstrators? These difficulties mean that a causalist story for group beliefs is implausible. It is harder to know what to say about acausalism at the group level. There is certainly a tradition—with Hegel as a prime example (though actually Hegel’s own views are probably quite a bit more subtle)—of talking about the telos of the group as explanation of its behavior and/ or existence. There is also a long tradition of functionalism in anthropology and sociology where group practices are explained by their fit in an overall system. These explanations in some ways parallel the teleological explanations that acausalists favor for individual behavior. These kinds of explanations can be broken into two different species: explanations via roles and explanations via selective mechanisms (Kincaid 2006). Classical role explanations are found, for example, across anthropology, where the various institutions of a specific society are shown to form

Beyond Causalism and Acausalism  191 an interrelated whole. Explanations vis selective mechanisms are about more than roles and are explicitly causal. So, for example, if I claim that unions exist because they solve free rider problems in negotiations and outcompeted other organizational forms because they do so, then I am not just describing a role but providing a mechanism explaining why the role exists. The former, role accounts seems closest to the acausalist story for individuals, because the main explanatory goal is to show how the behavior of institutions (which are collective agents) fits or is appropriate, given the other institutions in the society (and perhaps the natural environment as well). How these role explanations work, whether they are at least sometimes noncausal, and whether they are empirically sustainable is a matter of debate (Kincaid 2006). Thus, we are left with a situation similar to that we reached with acausalist accounts of individual behavior defended via some version of the interpretivist tradition: Causalism is implausible, there may be versions of acausalism that are defensible, but the exact theses and how their defense goes are still rather unclear. Of course, the issues will not be decided solely by conceptual analysis on my naturalist view but rather are empirical questions, though obviously conceptual clarification is needed, but clarification must be guided by what it takes to explain social behavior. Our intentional stance and revealed preference approaches warrant similar conclusions about groups as it did for individuals: (1) we can explain behavior without identifying internal beliefs, reasons, and so on as causes but instead can explain by using patterns of observed behavior as the basis of explanations, given constraints and causes in the environment and (2) there is a sense in which acausalism gains traction, because we have explanations of behavior as “existing in order to” in a sense that is warranted by the role of maximizing assumptions in the explanations. So, to ask whether collective agents really have beliefs is a bad or at least unnecessary question. A good question is whether we can explain their observed behavior by a utility function that describes their choices and how resource constraints and the behavior of other collective agents causally interact with their preference functions. Clearly sometimes we can.

Conclusion The causal versus acausal debate about explaining human behavior has not been very fruitful if the metric is making progress by scientific standards. Yet while our philosopher’s inclination is to look at deep conceptual issues, there are many other more concrete problematics that we can contribute to. I have tried to do so in this essay. No doubt the results are limited. Still hopefully something of value can be found in the discussion.

192  Harold Kincaid

Notes 1 Gunnar Schumann made numerous comments that helped clarify and deepen the arguments of this paper and those comments are much appreciated. 2 There is probably no general algorithm for deciding what is the best science as a long debate shows, but that does not mean there are not general scientific virtues which can be applied case by case. Of course, it is kind of judgment call, but fallibilist epistemology teaches us that this is generally the case for deciding what is relatively well supported by the evidence. 3 Science of course often starts from common sense concepts. But in prac tice what we see is that common sense concepts are—at least for scientific investigation—vague, inconsistent, and often need to be jettisoned altogether to explain empirical phenomena (see Wilson’s Wandering Significance [2006] for a nice example). The process is seldom one of getting clear on necessary and sufficient conditions for folk concepts first and then doing the science. 4 It might be thought that naturalism begs the question because it presupposes causal notions. But that need not be the case. Naturalism as used here is mainly an epistemic principle and then is silent on when and where causal notions are essential to explanations, so it does not stack the case in favor of causalism. Below I sketch a third alternative in the debate that does not have to be causal. 5 Bennett and Hacker (2003: section 8.1) deny that the neuroscience shows what I describe here, but they do not cite a single piece of empirical evidence. 6 I take it that causalists/acausalists want reasons that are conscious. Causalists are clear about this, acausalists not so much. Sehon (2005), for example, is never explicit about this in hundreds of pages of discussion. 7 A clear defense of the idea that belief-desire psychology could have a role in empirical psychology is found in Hockstein (2017). 8 These axioms can be put in psychological terms, but they can equally be put simply in terms of what subjects select in experimental tasks without any direct commitment to a psychological interpretation.

References Aguilar, J. and Buckareff, A. (2010). The Causal Theory of Action: Origins and Issues. In: J. Aguilar and A. Buckareff, eds., Causing Human Actions, 1st ed. Cambridge: MIT Press, pp. 1–27. Banfield, J., Wyland, C., McCrae, C., Muente, T. and Heatherton, T. (2004). The Cognitive Neuroscience of Self Regulation. In: R. Baumeister and K. Vohs, eds., Handbook of Self Regulation, 1st ed. New York: Guilford Press, pp. 62–84. Bennett, M. and Hacker, P. (2003). Philosophical Foundations of Neuroscience. Oxford: Blackwell. Bratman, M. (2001). Two Problems about Human Agency. Proceedings of the Aristotelian Society, 101(1), pp. 309–326. Dennett, D. (1991). Real Patterns. Journal of Philosophy, 88(1), pp. 27–51. Dennett, D. (2009). Intentional Systems Theory. In: B. McLaughlin, A. Beckermann, and S. Walter, eds., Oxford Handbook of the Philosophy of Mind. Oxford: Oxford University Press, pp. 339–350. Dennett, D. (2018). Reflections on Tadeusz Zawidski. In: B. Heubner, ed., The Philosophy of Daniel Dennett, 1st ed. New York: Oxford University Press, pp. 57–62.

Beyond Causalism and Acausalism  193 Gilbert, M. (1994). Remarks on Group Belief. In: F. Schmitt, ed., Socializing Epistemology, 1st ed. Lanham: Rowman & Littlefield, pp. 231–252. Hausman, D. (2011). Preference, Value, Choice and Welfare. Cambridge: Cambridge University Press. Hockstein, E. (2017). When does ‘Folk Psychology’ Count as Folk Psychological? The British Journal for the Philosophy of Science, 68(4), pp. 1125–1147. Kincaid, H. (1996). Philosophical Foundations of the Social Sciences. Cambridge: Cambridge University Press. Kincaid, H. (2006). Functional Explanation and Evolutionary Social Science. In: S. Turner and M. Risjord, eds., Philosophy of Anthropology and Sociology, 1st ed. Amsterdam: North-Holland, pp. 213–249. Kincaid, H. and Sullivan, J. (2010). Medical Models of Addiction. In: D. Ross, H. Kincaid and D. Spurrett, eds., What Is Addiction? 1st ed. Cambridge: MIT Press, pp. 353–376. Kydd, A. (2015). International Relations Theory: A Game Theoretical Approach. Cambridge: Cambridge University Press. List, C. and Pettit, P. (2011). Group Agency. New York: Oxford University Press. Lycan, W. (1988). Dennett’s Instrumentalism. Behavioral and Brain Sciences, 11(3), pp. 518–519. Ross, D., Sharp, C., Vuchinich, R. and ‎Spurrett, D. (2008). Midbrain Mutiny: The Picoeconomics and Neuroeconomics of Disordered Gambling: Economic Theory and Cognitive Science. Cambridge: MIT Press. Samuelson, P. (1983). Foundations of Economic Analysis. Cambridge: Harvard University Press. Sehon, S. (2005). Teleological Realism. Cambridge: MIT Press. Taylor, C. (1971). Interpretation and the Sciences of Man. Review of Metaphysics, 25(1), pp. 3–51. Tollefsen, D. (2015). Groups as Agents. Cambridge: Polity Press. Tuomela, R. (1995) The Importance of Us: A Philosophical Study of Basic Social Notions. Stanford Series in Philosophy. Stanford: Stanford University Press. Wegner, D. (2002). The Illusion of Conscious Will. Cambridge: MIT Press. Wilson, M. (2006). Wandering Significance: An Essay on Conceptual Behaviour. New York: Oxford Press.

11 Two Methods in History Thomas Keutner

Whenever we explain actions by mentioning intentions as their causes these intentions will not be the reasons why we did perform those actions. In the following this will be taken as the result of a critical discussion of Donald Davidson’s thesis that reasons are causes. The debate was initiated by himself with the first example (in a long row) of an intention causing an action without being its reason, though being the intention to that action. If reasons are not causes then this re-opens a different discussion which has been led especially in the philosophy of history about the relationship between intentional or motivational and causal explanation. What follows is a reexamination of that discussion before Davidson. This relation has often been understood as a non-contradictory one: Explanations by intentions and explanations resorting to psychological or sociological causes have been seen to coexist side by side, each of them having maybe its own domain. But that impression might be wrong. There is a special language game played with these two types of explanation: The expression of an intention might be used to contradict e.g. a causal psychological explanation and vice versa. Ever since Carl G. Hempel has tried to show that there is no methodological difference in principle between historiography and the natural sciences there is a debate going on also in the field of Analytical Philosophy as to whether historians offer causal and maybe nomological explanations, as Hempel thought, or whether they explain historical facts by hinting at the intentions of historical agents.1 In what follows it will be suggested that there is a certain logic in the use of these methods, a language game of contradictions which is shown also by the debates about the explanation of historical events. It is tried to elucidate this thesis mainly by turning back to some of the first approaches of the methodological historical discussion in the Fifties of the past century. First, Robin G. Collingwood’s criticism of the concept of causality in An Essay on Metaphysics will be taken to be as one of two monistic approaches: He states that historiography is the only legitimate place for

Two Methods in History  195 causal explanations, while its use in the natural sciences (and also in the applied natural sciences) is nothing else than a metaphoric derivation from this primary use. Therefore, historical explanations are causal, and only historical explanations are truly causal. Causality is the true definiens of historical explanation. As opposed to this, C. G. Hempel has tried to include historical explanations into the methodology of the natural sciences: Also historical explanations are deductive-nomological explanations, though elliptic in form, so that their premises should be completed by the causal law from which the explanandum is to be deduced. All explanations, in the end,2 are causal explanations, and this is the reason why historical explanations are causal.3 Now, if one does not accept the two monistic approaches, neither the unity of a causalist methodology nor the rejection of nomological causality throughout, then we are faced with a new problem situation: Maybe we have to accept the existence of a plurality of methods in history side by side, maybe historians give action explanations of a specific form, but maybe sometimes what has happened is explained by subsuming it under a psychological or sociological law or some other regularity, as Hempel had thought. In that case the question may arise about the relationship between the two methods. It will be shown that, on one hand, Collingwood suggested that the relationship is a metaphorical one or a relation of analogy; natural processes are explained as if they were social ones, as if a billiard ball would force another billiard ball to act. On the other hand, Hempel looked at motives as if they were causae efficientes and therefore part of a deductive-nomological explanation.4 But the question could ask for another type of answer instead of asking for the way of incorporating the alternative method. Second, therefore, it will be looked at some pluralistic approaches which leave untouched the rival method as such, attributing different status to different methods. Patrick Gardiner has identified different strata of explanation, so that intentional or motivational explanations and e.g. sociological explanations are compatible with each other and might therefore coexist side by side. Also William H. Dray accepts a coexistence of what he calls rational explanation and the covering-law-explanation: Where historians fail in the attempt to give rational explanations historical method is supplemented by explanations following the deductivenomological- or covering-law-model. Third it will be characterized a type of relationship, a language game of contradictions between the two methods. Speaking of a contradiction here is meant in a pragmatic sense. The hint to a causa efficiens might be used to contradict a motivational explanation put forward by the acting person itself or by a third party. And vice versa: A motivational explanation might be used to contradict an explanation given by subsuming a certain action or event under a law or general norm.

196  Thomas Keutner

I Monism. In Human Nature and Human History from 1936 R.G. Collingwood describes historical research as a psychological investigation.5 A truly historical event is an action, and actions are a unity of the outside and the inside of an event. The outside of an action are bodies and their movements; their inside is “that in it which can only be described in terms of thought”. The main task of the historian is “to think himself into this action, to discern the thought of its agent” (Collingwood 1993: 213).6 Historical explanation, it might be summarized, is the explanation by the intentions and beliefs of the historical agent: For history, the object to be discovered is not the mere event, but the thought expressed in it. To discover that thought is already to understand it. After the historian has ascertained the facts, there is no further process of inquiring into their causes. When he knows what happened, he already knows why it happened. (Collingwood 1993: 214) This is here taken to mean that “ascertaining the facts” leads to their description in the right way, and if the facts are actions this means that they are to be described as such, in the light of an intention.7 In that case, in fact, I know why something happened, namely in the pursuit of that intention. Things are different in the case of natural events. The scientist looks at an event in the context of other events, and therefore he reaches beyond the event to be explained. He “observes (the relation of the event) to others, and thus brings it under a general formula or law of nature” (Collingwood 1993: 214). To the scientist “nature is always and merely a ‘phenomenon’ [. . .] in the sense of being a spectacle presented to his intelligent observation”. The cause of an event in nature, thus, is another event, the cause of a historical event is “the thought in the mind of the person by whose agency the event came about [. . .] the inside of the event itself” (Collingwood 1993: 290 & 292). The historian penetrates into the inside of the event, his explanation is the demonstration of the inside of the action to be explained. Nevertheless “this does not mean that words like ‘cause’ are necessarily out of place in reference to history; it only means that they are used there in a special sense” (Collingwood 1993: 214). Collingwood, in Human Nature and Human History does not say what this sense consists in. It seems to be clear, however, that the use of “cause” for natural events is perfectly alright. This changes with his Essay on Metaphysics from 1942. Collingwood distinguishes three senses I-III of “cause”. In history the word is used in sense I; in this sense both, cause and effect are human activities, and the cause is a composition of a causa quod and a causa ut (Collingwood

Two Methods in History  197 1998: 285, 292ff.). The causa quod, again, “is a situation or state of things known or believed (to exist) by the agent”; and “the causa ut is not a mere desire or wish, but an intention” (Collingwood 1998: 293). Collingwood specifies this by calling “what is ‘caused’ [. . .] the free and deliberate act of a conscious and responsible agent, and ‘causing’ him to do means affording him a motive for doing it”. He underlines that a cause in sense I does not compel the agent to act; the person in question freely makes up his mind and decides now to move.8 This thesis is inter alia elucidated by the example of the letter of a solicitor causing a man to pay a debt, but also by the example of bad weather causing a man to return from an expedition. If Collingwood, therefore, is speaking of an activity causing an action, this may be the activity “of a second conscious and responsible agent”, but it might also be a natural event which affords the agent with a motive. So, the second example is one in which the agent is determined to act by a causa quod. There is a difference between the approach in Human Nature and Human History and the one in An Essay in Metaphysics. In 1936 the cause of the action is the agent’s intention. In 1942 there is still the agent’s intention (as the causa ut) but there is a cause preceding it, the action of another agent. What might be called a psychological approach in Human Nature and Human History—Collingwood’s concern for the historian’s grasp of the agent’s intention—has disappeared in favor of a social act, the persuasion of an agent by a counterpart. A second difference is Collingwood’s attitude towards the concept of causation. In Human Nature and Human History what was doubtful was the application of the concept in history. From sense I of the word as used in history he now distinguishes sense II as used in the practical natural sciences. The effect in this sense is not an action but a natural event, and Collingwood defines its cause as “an event [. . .] which it is in our power to produce or prevent, and by producing or preventing which we can produce or prevent that whose cause it is said to be”. He therefore calls this kind of cause a “handle” (Collingwood 1998: 297f.). Causes in sense II are to be found in medicine and in engineering (Collingwood 1998: 287); an example is the bite of a mosquito causing malaria. I can prevent being infected by malaria by protecting myself against the bite. This means that causes in sense II are under human control and can therefore be used to manipulate the natural event caused. To speak of causes in sense III “is an attempt (to look at them) as things that happen independently of the human will but not independently of each other” (Collingwood 1998: 287). In this case, if the cause exists, the effect also must exist and the effect cannot exist unless the cause exists: The cause is the sufficient and necessary condition of the effect (Collingwood 1998: 285f.). In this sense the word cause has traditionally been used in the theoretical sciences of nature, in physics and in chemistry.

198  Thomas Keutner Both uses, sense II and sense III, fail. They are both metaphorical, and in both cases “cause” is used in an anthropomorphic way. In sense II nature is seen as animated and therefore—following sense I, as something which could be motivated to act ((Collingwood 1998: 309); he therefore speaks of the “Greek sense” of the word). To speak of causation in sense III means to treat natural events as endowed with the power to motivate another event (Collingwood 1998: 324). But this would amount to the reduction of physical processes to social psychology. It is again an anthropomorphic idea and an idea which is nothing else than reaffirming the view it is thought to attack.9 Causation then is not a concept introduced into history following the model of theoretical natural science. Rather the concepts of a cause in sense II and sense III are presupposing the concept in sense I (cf. Collingwood 1998: 290ff.). In An Essay on Metaphysics Collingwood, therefore, takes a different stand from the one taken in The Idea of History. The historical sense of cause is not only the primary, it is the only legitimate one, the causing of a natural event being but a metaphor. In the end there is just one kind of causal explanation, historical explanation.10 The opposite view is brought forward by C.G. Hempel in his essay on The Function of General Laws in History from 1942. Following Hempel, all scientific explanations are deductive-nomological explanations, and the laws in question are causal laws.11 This holds also for explanations in history. Hempel sees one main problem for his thesis, namely that, most often, historical explanations do not mention a law from which a given historical fact would have to be deduced. His explanation is that the laws in question are presupposed, but simply not mentioned, being “a matter of course”. One of Hempel’s examples is that of the farmers from the Dust Bowl leaving their countries and migrating to California where they hope for better living conditions. The explanation for the farmers’ behavior here is that populations living under difficult conditions for their subsistence tend to emigrate and to look for better conditions. Such a hypothesis is tacitly taken for granted, says Hempel. Therefore, explanations in history and in the social sciences might quite generally be regarded as unproblematic elliptical explanations, if the presupposed laws can be easily reconstructed.12 Hempel’s thesis is that explanations in history and in the natural sciences alike are deductive-nomological explanations. Like with Collingwood, this corresponds to a monistic view, here however an explanation being the one referring to a causal law. In The Function of General Laws in History Hempel is not explicitly concerned with the question of whether intentional explanations can be subsumed under the deductive-nomological scheme. But he offers the definition of deductive-nomological action explanation in his Studies in the Logic of Explanation from 1948. Here, he argues against the view that purposive behavior cannot be causally explained because such

Two Methods in History  199 explanations refer to future goals of the agent, while physical explanations do not. Hempel’s argument is that while desires do refer to certain future events, it is not the future event which is thought to explain the action. In fact, the future objective often is not attained. The causal explanation rather is given by reference to the desire, not to the desired object, and the desire is part of the present, not of the future: When the action of a person is motivated, say, by the desire to reach a certain objective, then it is not [.  .  .] the future event of attaining that goal which can be said to determine his present behavior [. . .] rather it is (a) his desire, present before the action [. . .] and (b) the belief, likewise present before the action, that such and such a course of action is most likely to have the desired effect. The determining motives and beliefs, therefore, have to be classified among the antecedent conditions of a motivational explanation, and there is no formal difference in the account between motivational and causal explanation. (Hempel 1948: 254)13 Collingwood had declared that the causal explanation of natural processes is but a metaphor transferring social relations to the realm of nature. Historical events, however, are social events and are therefore legitimately to be explained by causes. Hempel, now, takes the counter position: There is but one science and human behavior (or human actions) are part of nature; therefore, they will both have to be explained by the same method, by deductive-nomological explanations.

II Pluralism. The different versions of monism have not been taken over in the immediately subsequent discussion. Even where Collingwood’s analysis of different senses of “cause” has been accepted, the same does not hold for his thesis of a metaphorical use of the term in the natural sciences. This will be seen later on esp. when turning to William Dray’s characterization of a “rational explanation”. If monistic solutions are rejected, then historians are confronted with a new problem situation: If there are two methods of explanation in history and if both methods are to be accepted side by side, what is their relationship? In The Nature of Historical Explanation from 1952 Patrick Gardiner asks for the exact meaning of “the real cause” which is often sought for not only in common life but also in history. In common life, to speak of a real cause is often meant to pick out a single condition by which an event might be produced or prevented. To say so does not mean that there are no other necessary conditions, but this one might be the one which can

200  Thomas Keutner be handled. In history, however, to say that one has found the real cause makes no sense: “The historical process is not like a machine that has to be kept in motion by a metaphysical dynamo behind the scenes. And there are no absolute Real Causes waiting to be discovered by historians with sufficiently powerful magnifying glasses” (Gardiner 1952: 109). There is, following Gardiner, not a single answer to the question about the outbreak of the First World War. This question will be answered in a number of different ways and upon different levels: The question [.  .  .] is answerable upon the level of individual purposes, desires, weaknesses, and abilities; it is answerable upon the level of national policies, traditions of diplomacy, plans; it is answerable upon the level of political alignments, treaties, the international structure of Europe in 1914; it is answerable upon the level of economic trends. (Gardiner 1952: 105) And as these questions and answers are located on different levels, following different perspectives and interests, they do not contradict each other. A journalist might declare the attack in Sarajevo to have been the true cause of the outbreak of the First World War. But a historian, in denying this incident to be the true cause is hereby not contradicting the journalist. The journalist will mean that Sarajevo set in motion the whole mechanism of actions and reactions preceding the war. But in offering an explanation the historian will deal with a large number of conditions; he will therefore accept the journalist’s analysis as justified as e.g. a description of the succession of facts. But he will also hint to various other factors on other levels which form the parts of a more complete picture of history. To speak of a true cause, Gardiner says, is more indicative of the perspectives and interests leading an explanation than of properties of the historical event. It is interesting to see that the answers enumerated are not only different in generality. They also differ in type. An answer referring to the purposes of an individual is obviously what up to here has been treated as a motivational or action explanation; answers referring to national policies, to traditions of diplomacy or international treaties would maybe have to be understood as being about instances of social action; and answers invoking a structure of Europe at a given time or economic trends might in fact be seen as being of the deductive-nomological or statistical type. If this is so, then Gardiner here highlights the compatibility of various causal explanations—sociological, economical—, but also the compatibility of these causal explanations on one hand and motivational explanation on the other one which he does clearly not regard as a causal one (Gardiner 1952: 124f.).

Two Methods in History  201 In criticizing Collingwood’s thesis that historical understanding is not the understanding of outward events but an understanding of the inward thought of the agent Gardiner makes a difference between two kinds of historical explanations: When the historian comes to ask why (human beings) did what they did, he sometimes answers by referring to general laws of human response to specified types of situation; and sometimes (he answers) in terms of what it would be reasonable to do in such-and such circumstances, and with such-and-such objectives in view. [. . .] The first type of explanation mentioned [. . .] may broadly be called explanation in terms of “causes” and “effects”; the second type, explanation in terms of “intentions” and “plans”. (Gardiner 1952: 49ff.) And Gardiner underlines that there is no contradiction between the two methods: There is no clash between the two: which we choose depends upon our interest and purposes. The beliefs that there must be such a clash, or that causal explanation is extrinsic to history, have their origin in the model of the historian as a man who treats exclusively of mental events and processes which are set behind the physical actions of human beings, pulling the strings, as it were. (Gardiner 1952: 50) In what follows, I will therefore speak of Gardiner’s position as that of a compatibility view.14 He explains his position somewhat later: Thoughts and intentions, being unobservable, cannot be causes. If, however, we insist on speaking of explanations by intentions and thoughts, then these explanations, not being causal, must be of some other type (cf. Gardiner 1952: 118 & 124). What is the historian doing if he explains an action by referring to the agent’s intentions? Gardiner’s answer is that by the motivational explanation the action is shown to be “an instance of how (a person) can in general be expected to behave under certain conditions. It sets (the person’s) action within [. . .] the pattern of his normal behavior” (Gardiner 1952: 125). It will be observed, here, that there is a difference between saying that a person acts in a certain way in conformity with her usual behavior—we might say: following a disposition—and saying that she acts following a certain intention. Maybe her intention does not at all conform with her usual behavior. Gardiner seems to agree with this when he says that “to say that an individual’s actions are planned or conformed to a

202  Thomas Keutner programme or policy may be very different from saying that they were intended”. Nevertheless, this is his thesis about following an intention: that “we are concerned with fitting a particular action within a certain pattern” (Gardiner 1952: 125). Is it correct to say that there is no clash between the two explanations given? I will turn to this later with an objection: There are certainly cases in which an explanation based on a regularity or a law is used to contradict a motivational explanation and vice versa. This is possible only because at least sometimes there is in fact a contradiction between the two methods. In Laws and Explanation in History, William Dray agrees with Patrick Gardiner that explanations given in history are not all of the deductivenomological type. Especially motivational explanations have a form of their own. This special form, following Gardiner, is that of a dispositional explanation, and, so Gardiner said, it is therefore not causal. Dray however, thinks that many of those deductive-nomological explanations adduced by Hempel and by Gardiner are not to be taken as such. A closer look shows that many of these must rather be analyzed in terms of individual motives and intentions. And he disagrees also with respect to the specific form of motivational explanations: Typically, they are neither deductive-nomological, nor are they dispositional explanations, that is, nor do they follow a weaker modification of the deductive-nomological type (cf. Dray 1957: 14). They are rather of a completely independent form which Dray calls that of a rational explanation. Dray follows G.M. Trevelyan in defining action explanation as the “reconstruction of the agent’s calculation of means to be adopted toward his chosen end in the light of the circumstances in which he found himself” (Dray 1957: 122). This calculation is not the deduction of a practical conclusion performed by the agent. And maybe it is even not a calculation which the agent would go through by himself before he acts. But “the point is that when we do consider ourselves justified in accepting an explanation of an individual action, (this explanation) will most often assume the general form of an agent’s calculation” (Dray 1957: 123).15 The model of “rational explanation” differs from the deductive-nomological and from the dispositional model by explaining not with respect to a regularity or a pattern followed by the agent, but by showing that the action to be explained was rational. And this is shown if the action is demonstrated to be a convincing means toward the agent’s ends, that he has good reasons to act as he did (cf. Dray 1957: 124, 127). That brings the argument back to the historian: The rational explanation is a good explanation if it starts from the purposes and convictions, from the true situation of the agent. But whether it is in fact a good explanation, this has to be judged by the historian (cf. Dray 1957: 126): “I wish to claim that there is an element of appraisal of what was done

Two Methods in History  203 in such explanations; that what we want to know when we ask to have the action explained is in what it was appropriate” (Dray 1957: 124).16 So, following Dray, this is the main characteristic to define rational from deductive-nomological explanations in accordance with certain laws (loose or otherwise): “The goal of such explanations is to show that what was done was the thing to have done for the reasons given, rather than merely the thing that is done on such occasions, perhaps in accordance with certain laws (loose or otherwise)” (Dray 1957: 124). This is also the line taken by Dray to answer Hempel’s original question: Hempel had looked at historical explanations as nomological ones, the laws in question stemming from psychology or sociology. Such psychological or sociological explanations, however, should better be looked at as rational ones. A “detailed description of the (dust bowl farmers’) aspirations, beliefs, and problems” would reveal their calculations in a certain situation rather more than to show us the laws followed by that population group (Dray 1957: 134). It is true that the explanation of the emigration from the dust bowl shows an element of generality. But this is not a nomological generality. It is, rather, the generality of principles of action: If exposed to situations of a certain kind, people have good reasons to decide for certain solutions.17 Dray does not specify how far his proposal should be followed, whether, to quote one of the examples given by Gardiner, also the politico-geographical structure of Europe is to be translated into motivational terms. How about Hempel’s second thesis, that also motivational explanations should be seen as being deductive-nomological in form? This thesis is also rejected. Action explanations cannot be reconstructed in the deductive-nomological form. Sometimes, we speak of our motives and purposes as causes, but if we do what we have in mind are not causal laws from which our actions could be deduced. It will be seen later that for Dray, as with dispositions, our intentions and beliefs are causes of our actions, but not in the nomological sense. In the focus of the historian’s interests lies the calculation of the historical agent. The historian wants to know this calculation, and he therefore regards as an explanation those answers which inform him about the agent’s purposes and motives. If a motivational explanation cannot be taken to be a nomological one, can a motive then be reconstructed as a disposition? An agent needs not add up his past actions if he wants to explain a present one. His explanation will not consist in subsuming his action under a regularity, it will be the expression of the agent’s calculation. This is therefore the point of rational explanation missed by the dispositional construction; when the historian wants to know why a person performed a certain action, what he wants to know is “the rationale of what was done”. And that an action was rational, this is not demonstrated by the fact that an agent did always perform it up to now under like circumstances. The agent’s calculation demonstrates how she will

204  Thomas Keutner reach her goal by the action performed: “There is a sense of ‘explain’ in which an action is only explained when it is seen in a context of rational deliberation; when it is seen from the point of view of an agent” (Dray 1957: 150). Dray does not deny the existence of dispositional explanations of actions. And he agrees that these explanations will reduce our surprise when faced with certain kinds of actions. These explanations, however, miss the point of the historian’s questions. Dispositional explanations are given from the perspective of a spectator. It is the spectator who would explain an action by showing that it was what a person would normally do under specific circumstances. What does it mean now to say that rational explanations cannot be reconstructed as dispositional ones, while there are nevertheless dispositional (and causal) explanations in history? Dray follows Collingwood in claiming that there are two standpoints from which human actions can be studied. When we subsume an action under a law, our approach is that of a spectator of the action; [. . .] but when we give an explanation in terms of the purposes which guided the action, the problem which it was intended to resolve, the principle which we applied etc., we adopt the standpoint from which the action was done: the standpoint of an agent. (Dray 1957: 140) So, the historian’s standpoint is the one of the historical agent, his typical questions would ask for the agent’s calculations, while an explanation in terms of theories of the historical process would simply be “uncharacteristic of ordinary historical writing” (Dray 1957: 142). Gardiner thought that purposes and beliefs are not causes; they are no occurrences at all and therefore, being mental unobservables, even no candidates for being causes. Dray rejects this thesis because dispositions are very often cited as causes by historians. What is then the meaning of “cause” in history? He called it a pragmatic and inductive understanding of causality: Something is called a “cause” if it is necessary for the effect and if it is regarded to be a relevant causal condition (Dray 1957: 98). Quite generally then, there is “no restriction whatever upon the type of thing that can qualify as a cause” (Dray 1957: 151): If dispositions are rejected as causes, then not because belonging to a certain type of things such as a realm of mental causes, but only because they might not qualify under the specific circumstances given. He now makes it clear that this is the meaning of “cause” in action theory: If John hit me because he is bad tempered, this motive is to be regarded as a relevant necessary condition for what John did. We may therefore in an appropriate context call his motive a cause (Dray 1957: 151f.).

Two Methods in History  205 The same holds true for reasons as given by rational explanations. It looks as if Dray would anticipate Donald Davidson when he says that “(to put it a little crudely) reasons, too, can be causes” (Dray 1957: 153). But this impression is misleading. Dray makes it clear that he is not talking of causes in Davidson’s sense: “What is required is a qualified restatement of Collingwood’s doctrine that in history the term ‘cause’ is often used in Sense I: the sense in which to cause someone to do something is to provide him with a motive for doing it (where ‘motive’ means ‘reason’)” (Dray 1957: 154). And this use of “cause” is not to be taken in the sense that the agent did not calculate his action. It is exactly his calculation which might also be addressed as the cause of his action (ibid.). The causal connection, Dray says, is to be understood in rational terms. What is the difference then between these two ways of describing one and the same process? Our way of speaking, says Dray, “is one of approach, or point of view, or kind of inquiry”: Speaking of a reason corresponds to the point of view of the agent (and, therefore, of the historian); speaking of the cause “is to adapt the point of view of a manipulator—although of one well aware that he is dealing with agents who act on rational considerations. [. . .] It is a fact of ordinary historical writing that historians do sometimes take up (such a) ‘stand-offish’ attitude in explaining even the rational behavior of their characters” (Dray 1957: 154). Is there not more to our distinguishing between reasons and causes, even when the concept of a cause is understood in Collingwood’s sense I? Dray, on the relevant pages raises the question as to how this sense of a cause is to be reconciled with our understanding that to be caused is closely linked with being necessitated. Collingwood had answered this question by underlining that the meaning of “necessary”, just like that of the word “cause”, goes back to the same original meaning, namely “sense I”: Just as the original sense of the word ‘cause’ is what I have called the historical sense, according to which that which is caused is the act of a conscious and responsible agent, so the original sense of the word ‘necessary’ is an historical sense, according to which it is necessary for a person to act in a certain way: Deciding so to act and acting therefore freely and responsibly, yet (in a sense which in no wise derogates from his responsibility) ‘necessitated’ to act in that way by certain ‘causes’, in sense I of the word ‘cause’. (Collingwood 1998: 320) Dray follows Collingwood’s observation: “To be caused to act in this sense does not imply that the agent did not make up his mind to do what he did on the basis of certain rational considerations” (Dray 1957: 154). And he adduces the example of a person with a pistol pointed to his head,

206  Thomas Keutner where the following behavior of that person is what we mean by “compelled”, nevertheless being free and responsible (a person which wants to die will not deliver his wallet to the robber). Dray concludes “that the necessity of a causal connection, when it is actions we are talking about, is very often rational necessity. [.  .  .] In history the relevant kind (of necessity) will often be that found in action done for a good reason (from the agent’s point of view)”. So, it looks as if following Dray in history on one hand there are rational explanations of actions giving the agent’s calculations. These are delivered from the agent’s point of view and they are at the center of the historian’s interest. There are also causal and dispositional explanations of actions; but these are but a way of speaking—a form of description chosen (also by historians) in a “stand-offish” attitude. And here, to speak of causation means to do that in a specific sense of the word “cause” (namely Collingwood’s sense I). A  description in these terms does not exclude a description of that very action in rational terms. Does this eliminate deductive-nomological explanation from history altogether, in favor of rational explanations and of causal explanations in sense I? There is a paragraph in which Dray makes it clear that the model of deductive-nomological explanation is not to be seen as driven out of history by the rational explanation. While introducing to the methodology of rational explanation, Dray explains that the historian starts his investigation from a “criterion of intelligibility”, from “a general presumption that a given action will be explicable on the rational model if we study it closely enough” (Dray 1957: 137). Is there a limit to this assumption? There is no sharp line. When the historian’s efforts fail, his reaction will rather consist in reexamining his tools than in doubting the rationality of the action under investigation. In the end, however, he might look for a different method and change his instruments. Dray mentions the case of psychoanalytical explanation: “There will be particular cases in which we find it impossible to rationalize what was done, so that (the explanation) [. . .] will be of another kind [. . .] We give reasons if we can, and turn to empirical laws if we must” (Dray 1957: 138).18 And he adds that such a switch to another method belongs to history: “If a psychological theory were necessary and available, it would be the historian’s business to use it, and it would be of interest to the reader to know it”. However, “non-rational explanation only supplements, it does not replace, the rational sort” (Dray 1957: 139). It might be said, therefore, that Dray takes into account a plurality of methods of explanation in history: “rational explanation”, the calculation of the agent; dispositional explanation which is to be regarded as a causal explanation, but then in Collingwood’s sense I. The historian, sometimes might be interested to look at human actions from the outside, but this will be just a way of speaking, the historian being “well aware

Two Methods in History  207 that he is dealing with agents who act on rational considerations” (Dray 1957: 155);19 and causal explanation in the nomological sense, as envisaged by Gardiner, but only as a candidate for the supplementation of rational explanations. It is clear, however, that this kind of supplementation of rational explanation cannot be subsumed under Gardiner’s complementation thesis. If the historian resorts to a psychological explanation he will do so to contradict a tentative, but unconvincing rational explanation. The final part of these remarks are dedicated to elucidate this kind of contradictions.

III Contradictions. After the monistic beginnings (Collingwood, Hempel) some philosophers favored pluralistic approaches: In history motivational or intentional explanation lies at the center of the investigator’s interest. But motivational explanation is accompanied by some other types, and here the deductive-nomological form assumes a paradigmatic role. It will be demonstrated, however, that this description is not complete by discussing the function Dray assigns to rational and to, as he says, supplementary law-based explanation. Historians follow a principle of rationality and try to explain actions by rational explanations, by the considerations of the historical agent. Sometimes, these actions seem to be incomprehensible. In this case the historian may look for e.g. a psychoanalytical explanation and adduce not a motivational explanation but an explanation by the unconscious motives of the agent. Will these be explanations of one and the same type, so that it is shown that the limits of rationality are wider than we normally draw them in ordinary life or in our normal use of the term of a “motive”? Psychoanalysis, in this case, would be interpreted as having discovered the existence of a new kind of motives, namely unconscious motives.20 But Dray underlines that in normal circumstances we do not assimilate rational and non-rational actions; rational actions are explained by rational explanations, non-rational actions by explanations of some other type. What is a psychoanalytical explanation if it is not based on the extension of our ordinary motivational vocabulary? Psychoanalysis is a theory, and if unconscious motives are not a new kind of motives, they should be seen as theoretical terms and their meaning as to be determined in the context of the theory. That this vocabulary follows the paths of our ordinary motivational vocabulary is a very effective means in the clinical domain: The patient can persuade himself (or can be persuaded) to accept a certain interpretation in terms of his, as it seems, newly discovered motives. Which entities correspond to these theoretical terms, is an open question; they might show a greater or lesser similarity to our ordinary motives. Whether psychoanalytic interpretations are successful

208  Thomas Keutner “mutative” interpretations, whether they can change the behavior of a person, this is a question which has to be answered in the ordinary academic way in the psychological lab and by epidemiological studies. Is the supplementation function now a case of the co-existence of rational and non-rational explanation? Sigmund Freud describes the following case of a compulsive neurosis: A 30-year-old lady several times a day leaves her room and changes to a table in the neighboring one; there she calls for her maid and gives her some irrelevant orders. Then she goes back to the room she had left (cf. Freud 1940: 268ff.). One day she shows to Freud that the tablecloth covering the table contains a big red spot. In the context of the therapy it becomes clear that her habit has a certain symbolic meaning: It goes back to her wedding night when her groom revealed to be impotent and therefore spilled a pot of red ink over the wedding bed. Hereby he tried to save his honor in the eyes of the maid. Freud’s interpretation of the lady’s compulsive act is that it consists in the identification with her husband by which she also tries to save his honor, showing the spot on the table to the maid. Symbolically interpreted the tablecloth is the bedsheet and the spot stands for the traces of her defloration. But Freud’s explanation does not stop with this interpretation in motivational terms. He makes it clear that it is a causal explanation which he has in mind: “Es hatten also seelische Vorgänge in ihr gewirkt, die Zwangshandlung war eben deren Wirkung; sie hatte die Wirkung in normaler seelischer Verfassung wahrgenommen, aber nichts von den seelischen Vorbedingungen dieser Wirkung war zur Kenntnis ihres Bewusstseins gekommen”. (“Thus, psychic processes had worked in her, the compulsive act had been their effect: She had realized this effect under normal psychic conditions, but nothing of the psychic pre-conditions of this effect had reached her consciousness”.) (Freud 1940: 286; transl. T.K)21 Now this lady is not forced to accept Freud’s interpretation. She might choose to prefer a different explanation in rational terms e.g. that she is only changing the room to give some orders to her maid. But Freud will contradict her, and he will call her answer an unsatisfactory one and a rationalization (in the psychoanalytical sense).22 In both cases then the explanation will be used to contradict the alternative one: the rational explanation given by the patient to contradict Freud’s causal explanation; and Freud’s causal explanation to contradict the patient’s rational explanation. What is now the exact meaning of these contradictions? We might say that his patient could reject Freud’s interpretation if she can truthfully say that she knew about that “unconscious wish” she is accused of,23 that the content of his insinuation was part of her considerations before acting,

Two Methods in History  209 and that she nevertheless decided to act as she did, maybe as a kind of defiant action. And we can say that Freud would be entitled to insist on his interpretation if he can truly presuppose that the patient is in fact determined by a wish or intention she is ignorant of. So, this is what the dispute between the two parties is about: whether the patient knows about or ignores the supposed intention. And the point of this question is not that she would have acted differently on the basis of this knowledge; her action might be quite the same. It is rather that on the basis of this knowledge the type of our explanation will change: The ignorance of her wish on the part of the patient will lead to a causal explanation, her knowledge to a rational one, even if what happens will be unchanged. Is this an analysis valid only for psychoanalysis and for psychoanalytical hypotheses about unconscious intentions? It has been said that what is unknown here is not a new kind of motives but rather the hypothetical entities presupposed by psychoanalytical interpretations. Now, psychoanalysis is not the only theory present in the field of historical research. Since the times of Hempel it has been said that the explanations given by historians make use of very different theories taken from various neighbor disciplines such as psychology, sociology, politics etc. These theories, one might say, imply different grades of theory loadedness, e.g. psychology maybe more so than politics. If in social psychology it is discovered that in groups individuals tend to change their opinions and to follow one of its members as a leader, then there is something to learn for the individual that is considering which way to follow. She might decide against taking the path designed by some other individual in the group or decide to follow that path nevertheless, notwithstanding her newly acquired knowledge. In this case the explanation for her behavior will be a rational explanation. If she is ignorant of such group processes, there should be a much higher probability that she will follow the leader of the group. The same will hold for such examples as those introduced by Hempel at the very beginning of the historico-methodological debate: If a person knows that government agencies have a tendency to perpetuate themselves and to expand, and if she is an influential agent in such an agency then she might try to stop this tendency for her own institution (it will be supposed that she has some reasons of her own not to let things go their course). But whatever her actions will be: After learning about such tendencies, the explanation of her action will be a “rational” one, and not anymore determined by the fact that she is a part of that organization. It is to the historian to decide whether what he has to explain is an informed or rather an ignorant action and, in the consequence, what type of explanation will be the relevant. Contradictions of the type mentioned then show a characteristic feature: There are certain laws or regularities which depend for their

210  Thomas Keutner effectiveness on their being unknown to the person they are dealing with. In the case of the explanation of a specific action they can therefore be rejected by showing that the agent has been aware of them in calculating the action. (She might then nevertheless act in the sense indicated by the law in question, but in that case the action will not be an instance of that law.) There are, of course, other types of removing the cause of an action than by informing the agent. If e.g. an agent acts unknowingly under the influence of a drug, then she might think that she acts as she does following some irrefutable reason. In that case we can remove the cause of her acting by interrupting the medication. In the case of the causal relations here envisaged, however, the cause is removed by information alone and not by information and a subsequent intervention. It is therefore of a type different from the one in which a corporeal intervention is necessary. Monistic approaches then, such as Hempel’s thesis of the unity of causal explanation or Collingwood’s idea about the exclusivity of causality in sense I in history fall short of an important use of both types of explanation, namely to contradict the alternative. Therefore, the alternative has to be a part of the game. But the same also holds for those pluralistic methodological interpretations which classify motivational or intentional and law-based explanations as compatible or as supplementary. These interpretations push aside the fact that these methods are not simply responsible for different domains of the scientific landscape. They are also introduced with the purpose to fulfill their role in the scientific dispute. And this role is that of argument and counter-argument.

Notes 1 For the prehistory of this debate cf. von Wright (1971: ch. I, Two Traditions) 2 Hempel agrees that there are statistical besides causal explanations; but, as well-known, he accepts as such only statistical explanations with a high degree of probability. Highly probable explanations form a special class “of a distinct logical character, reflecting, as we might say, a different sense of the word ‘because’ ” (cf. Hempel 1965: 391ff.). 3 This idea of a methodological unity of science can also be exemplified by Donald Davidson’s thesis that actions are events caused by “primary reasons” (cf. Davidson 1980: 4). Historical processes, the coming about of past actions, will then be causal processes. Hempel’s view of a deductive-nomological explanation of action has been further elaborated by Paul Churchland in his The Logical Character of Action-Explanation in 1970, where he tries to complete the premises of the deductive-nomological scheme of an action explanation as to make it really deductive. It has been objected, however, that such a completed scheme is nothing else than a definition of the meaning of an “intention such-and-such”, in which case the would-be-law covering the action to be explained could not be considered to be an empirical statement anymore.

Two Methods in History  211 4 Hempel gives no example for the explanation of a single action in his The Function of General Laws in History, but he does so somewhat later in Studies in the Logic of Explanation (cf. above). 5 Cf. Collingwood 1993: 213ff. Human Nature and Human History was published separately by Collingwood in 1936, but had then been included in the posthumous publication of The Idea of History in 1946 in the Epilegomena. It is here quoted from the revised edition of The Idea of History in 1993 (cf. the Editor’s Introduction, esp. pp. XIIff.). For Collingwood, however, historical investigation is not just a psychological one. The re-enactment of an action, as he calls it, is rather what has been called a rational reconstruction, it is a critical reconstruction: “The historian not only re-enacts past thought, he re-enacts it in the context of his own knowledge and therefore, in reenacting it, criticizes it, forms his own judgement of its value, corrects whatever errors he can discern in it” (Collingwood 1993: 215). Dray remarks: “It would be odd, surely, to say that understanding an action logically required the historian’s correction of errors of reasoning on the part of the agent” (Dray 1980: 24). 6 Collingwood, being concerned here with the re-enactment of the thought of the agent, does not say that this thought would also include the agent’s intention. But certainly the agent’s intention would have to be mentioned in the explanation of his action. 7 That intention, however, will not be the intention to perform that action. The intention to perform an action will make it clear that the action is intentional, that it is in fact an action. But the explanation of the action will mean to mention a further intention and showing thereby that the action was performed to reach a certain goal (cf. Dray 1995: 47–49). 8 Though Collingwood speaks of the causa quod as a causa efficiens. More specifically the causa quod is not a state of things but the agent’s belief that this state of things exists (cf. Collingwood 1998: 293f.). 9 Obviously, Collingwood does not want to say that there is no explanation of natural events. He makes it clear in a comment on Newton’s Principia Mathematica: In modern physics the class of events happening according to laws has swallowed up the class of events happening as effects of causes. Explanation is by laws, not by causes (Collingwood 1998: 326f.) 10 It is to be added that even causation in sense I might be illusive: If we think of it in terms of strict necessity, then we will find that also necessity here is only one which the agent is convinced of or of which he convinces himself (cf. Collingwood 1998: 293f.). 11 Obviously, most explanations not only in physics are statistical and, therefore, not deductive in character. Hempel however takes it that inductivestatistical explanations are to be regarded as such if they make sure that the explanandum follows from the premises with a high degree of probability. In such cases we are entitled rationally to expect a certain outcome, say of some experimental procedures. Maybe it can be suggested that Hempel regarded cases of high probability as quasi-causal situations (cf. Hempel 1942: 237). He will later agree, however, that “because” here might be used in a sense different from the use in the deductive-nomological explanation. Certainly we will not insist on highly probable statistical explanation anymore after the turn inaugurated in explanation theory with Michael Scriven’s famous example about the infection with progressive paralysis showing that also low probabilities can be explanatory depending on their relation to given epistemic status (cf. Hempel 1965: 381ff, 391ff.). 12 They are elliptical also in the sense that social and historical situations are complex in form. In this case it will be more difficult to give a complete

212  Thomas Keutner explanation. How exactly is the discontent of a population to be determined if it is taken to be the explanandum of a revolution (cf. Hempel 1942: 236ff.)? 13 Hempel’s thesis has been further elaborated by Paul Churchland in his The Logical Character of Action Explanation. As opposed to this, Davidson does not believe that there are psycho-physical laws. His explanation of an action is an explanation by a single causal statement. The justification of this explanation is then to be given by the law of some more fundamental discipline. 14 Cf. also Gardiner (1952: 50): “Nor is there any conflict or incompatibility between explanations of this kind, and explanations in terms of thoughts and intentions”. 15 Dray underlines that “there are special dangers in the construction of such calculations after the fact”, but he does not explain what these dangers might consist in. His argument reminds of Wittgenstein’s distinction of reasons and causes in his Blue Book, published one year after Dray’s book, but circulating in Cambridge since the Thirties of the past century. Wittgenstein says that “giving a reason for something one did or said means showing a way which leads to this action. In some cases it means telling the way which one has gone oneself; in others it means describing a way which leads there and is in accordance with certain rules” (Wittgenstein 1958: 14). Maybe the special danger emphasized by Dray consists in the possible conflation of these two ways: We explain an action done by mentioning the calculation performed before acting. But we justify an action also by mentioning a calculation leading to it which we did not perform. The action is then justified in the sense that it is rational, that it corresponds to some means-to-end-calculation. We can show that there are good reasons supporting the action though these were not the reasons why we did it. If this calculation is then also called an explanation, then not in the sense that it has been the calculation that had led us in our acting. 16 The idea of an evaluation of the reasons of the agent by the historian is strongly reminiscent of Collingwood’s thesis that historical knowledge consists in the re-enactment of past thought in the context of his own knowledge. Dray explores this thesis critically in Dray (1980: 23ff.), see his critical remark quoted above, en. 5.) and again in Dray (1995: 52ff.). 17 Von Wright here speaks of “universal motives”, the motivational background being “of a very “crude” and universally human character, so much so that it need not merit special consideration by the historian (cf. von Wright 1971: 144). Dray sees the following difference between the universality of motives and the generality of a law: “If a negative instance is found for a general empirical law, the law itself must be modified or rejected”; whereas “if a negative instance is found for the sort of general statement which might be extracted out of a rational explanation, the latter would not be necessarily falsified” (Dray 1957: 132). Von Wright has followed Dray with this interpretation of the relationship between motivational and causal explanation. He takes it for granted that in history and in the social sciences we find both methods of explanation. He speaks on one hand of the first type as of explanations by Humean causes, their main characteristic being their logical independence from their effects. Intentions, on the other hand, are logically dependent on their consequences in the sense that an intention is not definable without relying on the intended result. Explanations by intentions, therefore are explanations by non-Humean causes (cf. von Wright 1971: 93). What about those explanations which Hempel and Gardiner had termed as causal—sociological, socio-psychological, psychological explanations etc.

Two Methods in History  213 A  relevant example is Hempel’s example of the farmers leaving the “dust bowl”. Following von Wright, the tendency of these farmers is not a Humean cause, this is rather their motive. And this is shown by the fact that it will be reasonable to emigrate under the given conditions, but that a person with different goals might decide to stay, e.g. a person born in this region which is not willing to leave its familiar surroundings. To speak here of “cause” and “effect” will be perfectly appropriate, but it will be a confusion to think, therefore, that this is a case of Humean causality (cf. von Wright 1971: 144). One more example is the outbreak of the First World War. This example is of special interest because the events in the summer of 1914 look as if being linked by a kind of mechanism—actions and reactions. The difference between this mechanism and a physical one, however, consists in the fact that the new moments of the changing situation form “motivations for further actions” (von Wright 1971: 140). And this is absent in a physical mechanism like the explosion of a powder-barrel initiated by a spark. Von Wright, therefore, speaks of the quasi-causality of such explanations. 18 “Rationalization” in the sense of giving appropriate reasons for the action performed. 19 In a critical examination of Collingwood’s analysis of the concept of a cause, Dray is mainly interested in the question as to whether Collingwood is right in excluding sense II and sense III from history. He is not interested in correcting Collingwood’s radical thesis about a purely metaphorical use of these two senses all over (cf. Dray 1957: 116ff.). 20 In the Fifties of the last century such an interpretation has been brought forward by Toulmin (1954: 138) and by Flew (1954: 145). 21 Freud here also underlines that he understands psychic causes in a realistic way: “Wir werden (aber an der Annahme unbewußter seelischer Vorgänge) festhalten und wir müssen es [.  .  .] als unbegreiflich abweisen, wenn uns jemand einwenden will, das Unbewußte sei hier nichts im Sinne der Wissenschaft Reales, ein Notbehelf, une facon de parler. Etwas nicht Reales, von dem so real greifbare Wirkungen ausgehen wie eine Zwangshandlung!” “We will stick to the hypothesis (of the existence of unconscious psychic processes) and reject it as incomprehensible if somebody would object that the Unconscious would be nothing real in the sense of science, just an expedient, une facon de parler. Something unreal with such sensible effects as a compulsive action!” (Freud 1940: 287; transl. T.K.). 22 That is, not in Dray’s sense of describing the considerations of the agent but in the sense of giving a rational explanation to an irrational behavior. 23 This is of course a too simple description of what is going on. A  neurotic symptom is not deleted by simply informing the patient about the existence of the underlying unconscious wish (cf. Strachey 1935 about so-called “mutative interpretations”).

References Churchland, P. (1970). The Logical Character of Action-Explanations. Philosophical Review, 79(2), pp. 214–236. Collingwood, R. (1993). The Idea of History, 2nd ed. Oxford: Clarendon Press. Collingwood, R. (1998). Essay on Metaphysics, 2nd ed. Oxford: Clarendon Press. Davidson, D. (1980). Actions, Reasons, and Causes. In: D. Davidson, ed., Essays on Actions and Events, 1st ed. Oxford: Clarendon Press, pp. 3–19.

214  Thomas Keutner Dray, W. (1957). Laws and Explanation in History. Oxford: Oxford University Press. Dray, W. (1980). R. G. Collingwood and the Understanding of Actions in History. In: W. Dray, ed., Perspectives on History, 1st ed. London: Routledge, pp. 9–26. Dray, W. (1995). History as Re-enactment. R. G. Collingwood’s Idea of History. Oxford: Clarendon Press. Flew, A. (1954). Psychoanalytic Explanation. In: M. Mcdonald, ed., Philosophy and Analysis, 1st ed. Oxford: Basil Blackwell, pp. 139–148. Freud, S. (1940). Vorlesungen zur Einführung in die Psychoanalyse. Gesammelte Werke. Vol. XI. London: Imago Press. Gardiner, P. (1952). The Nature of Historical Explanation. Oxford: Clarendon Press. Hempel, C. (1942). The Function of General Laws in History. Journal of Philosophy, 39(2), pp.  35–49 (reprinted in: C. Hempel (1965), ed., Aspects of Scientific Explanation and Other Essays in the Philosophy of Science, 1st ed. New York: Free Press, The Macmillan Company, pp. 231–242). Hempel, C. (1948). Studies in the Logic of Explanation. Philosophy of Science, 15(2), pp.  135–175 (reprinted in: C. Hempel (1965). Aspects of Scientific Explanation and Other Essays in the Philosophy of Science. New York: Free Press, The Macmillan Company, pp. 245–295). Hempel, C. (1965). Aspects of Scientific Explanation. In: C. Hempel, ed., Aspects of Scientific Explanation and Other Essays in the Philosophy of Science, 1st ed. New York: Free Press, The Macmillan Company, pp. 331–496. Strachey, J. (1935). Die Grundlagen der therapeutischen Wirkung der Psychoanalyse. Imago, 21, pp. 486–516. Toulmin, S. (1954). The Logical Status of Psychoanalysis. In: M. McDonald, ed., Philosophy and Analysis, 1st ed. Oxford: Basil Blackwell, pp. 132–139. Von Wright, G. (1971). Explanation and Understanding. New York: Cornell University Press. Wittgenstein, L. (1958). The Blue Book. In: R. Rhees, ed., The Blue and the Brown Book, 1st ed. Oxford: Basil Blackwell.

12 An Anti-causal Theory of Action as Basis for Historical Explanations. A Sketch1 Gunnar Schumann

Causalism is widespread in the theory of historical explanations and explanations of human actions in general. Overwhelmed by the success of the sciences, causalists think that all explanations must be of the same kind: causal explanations as they are employed in the sciences, foremost physics. Often, they do not even argue at length for it but just silently assume that explaining an event is identical to citing its causes. So they think why-questions always ask for causes and that sub-clauses beginning with “because” always state causes. They assume that in order for historiography and the humanities in general to be scientific at all, those disciplines have to take over the explanation methods of the sciences, i.e. their causal mode of explanation. Although Hempel’s notorious “Covering-Law-Model” is not seriously debated as a paradigm of historical explanation anymore, the scientific Zeitgeist, attempting to transfer the methods of explanation of physics to historiography and the social sciences, lives on and finds its expression in the view that human actions are to be explained by causes. The idea is: History is made up of human actions and these are ultimately causally determined, such that there are perhaps no global laws of history (as in Hempel, Popper), but one day history can be explained scientifically because human actions can be explained causally. Causal explanations might not involve any known regularities or not even regularities at all, nonetheless they are taken as the appropriate form of human action explanations. Even the concept of narrative explanation which became prominent in philosophy of historiography after the peak of the reasons-causes debate, is only directed against the idea that historiographical explanations are nomological in character. Nonetheless narratives are still conveyed the task of delivering a causal story about how a historical event came to be. Thus the doctrine of narrative explanations seems to embrace causalism (Stueber 2012: 24). Against this view, I want to argue that historiography deals with human actions and that these call for a teleological or intentional explanation. Human actions (be they past, present, or even future) must be explained by reasons, i.e. by reference to the goals, purposes, and intentions

216  Gunnar Schumann (and beliefs), that rationalize the actions of the agents and make them understandable. I shall argue that the true form of explanation of human actions takes the agent’s intentions and means-end-beliefs as premises of a practical syllogism, from which the explanandum, the action, follows logically, not causally. To show this, I will make use of several anti-causalist arguments, especially a qualified version of the so called “Logical Connection Argument”. My result will be that the explanation of a human action requires an investigation of the action’s context to determine the elements of the explanans, i.e. the agent’s volitional and doxastic premise. Causalist theories of historical explanation rest on causalist theories of human action explanation. If it can be shown that human actions in general require a causal explanation, then the central premise of a causalist theory of historical explanation can be taken for granted. Or, conversely: If it can be shown that human actions cannot appropriately be explained causally then the causalist view of historiographic explanation collapses and the notion of a “historical cause” (or “social cause”) cannot be taken in the same meaning as “cause” in the sciences. Causalist theories of human action take reasons for actions as causes of actions. In the contemporary debate, it was Donald Davidson who most prominently argued for this position in his 1963 article Actions, Reasons, Causes. The picture of the causalist of action is often: actions are the effects e.g. of “the will”, of intentions, of decisions, of desires and wishes of the agents. This picture is old and widely spread: Originating in Descartes, it stretches to Hobbes, Spinoza, Hume, and is a mainstream view in analytical philosophy of today. As Mele puts it: “According to a familiar causal approach to analyzing an explaining human action, our actions are, essentially, events [.  .  .] that are suitably caused by mental items, or neural realizations of those items. Causalists traditionally appeal, in part, to such goal-representing states as desires and intentions (or their neural realizers) in their explanations of human actions” (Mele 2000: 279). This picture is fuelled from several misconceptions, namely: (I) that actions are identical to physical events, (II) that actions are physical events preceded or accompanied by volitional mental occurrences, (III) that these volitional mental occurrences are causing actions just like natural phenomena cause each other. In the following I want to show that all these assumptions are mistaken. My arguments will be directed against the causalist picture in general, not against a specific causalist theory of human action explanation and I will limit my discussion to human action explanations in terms of intentions and beliefs.

I.  Actions Are Not Physical Events First it should be clear that human actions are not simply bodily movements, i.e. physical events. Bodily movements can be explained adequately

An Anti-causal Theory of Action  217 only by citing causes. But human actions are, while often accompanied by bodily movements, not identical to them. 1. Actions do not have to be accompanied by physical happenings or bodily movements at all. When I  hold a rope firm or prevent a door from being shut with my foot, then no physical movements are involved but nonetheless I act (Hacker 2010: 157). When somebody falls into a frozen lake and I stand on the shore doing nothing but watching then I perform a very bad action—although I do not move any limb at all. Omissions are actions as well. 2. Actions do have a physical aspect, but actions are not identical with their physical aspect, because a description of a bodily movement or muscular activity may be as detailed and precise as one wishes— but usually it will not show by itself the action of which it is the physical aspect (Von Wright 1985: ch. II.9). Even if we knew everything there is to know about the physics of a bodily movement, there still is a sense in which we could not be said to know which action is performed by this movement. Let’s say we watch how the arm of a person rises. What action does she perform? Does the person greet someone? Does she order a drink? Is she bidding at an auction? Does she perform a dance move? Is she doing a physical exercise? Does she point towards the ceiling? Which of these actions the person performs cannot be decided by any physicalist description of her movement alone. This is because we do not know yet anything about the context of the action in which the movement is embedded. Only when we widen the perspective from her movement and look at further behavior of the person and the context under which she makes her movements does it become clear what she is actually doing (Schroeder 2001: 156f.). It could be objected that she is obviously performing the action of raising her arm—but not even that could be said for sure. For why could it not be that her arm rises just because of a spasm or someone else raises it with invisible threads, so that it is not her action of raising the arm? To decide if she is performing the action of raising her arm or if it is some item of involuntary behavior (which would be no action) also depends on the context of the bodily movement. 3. Actions are not identical with their physical aspect, because many actions can be realized physically in a variety of forms (Von Wright 1971: ch. III.9). The many concepts of actions that we have rarely imply any specific bodily movements. Let’s take the action description: “He turned left into the main road”, Is “turning left” identical to a specific bodily movement? No, for he could have turned by moving the steering wheel of his car with two hands, with one hand, even with one finger only. He could have turned left with a bike, which requires a totally different sort of movement, or even as a pedestrian,

218  Gunnar Schumann in which case the turning does not involve a special movement of his arms at all. Usually, when we describe what action someone performed we do not report what movements his body carried out but report the result of the action. Even in our own case, when we are asked to describe in retrospect our actions, we usually remember what we did—not our precise bodily movements (Wittgenstein, PI, § 648). (Note that also in linguistic actions we usually do not remember our precise words but rather what we said.) Bodily movements and actions can be seen as standing in the same relation as spoken or written words and their meaning. Scribbled marks on a paper or uttered noises do have causes, but it would be odd to say that their meaningful content is caused. The content of some item of speech is not a physical event. It is no metaphysical, mental event either. It is no event or object at all. Meaning is conveyed to noises and marks by a rule-guided practice of a speaker community. The uttering of a specific sound gains its meaning by being used in accordance with an existing rule in a speaker community. Analogously, actions have semantic properties. They are something interpreted and understood; something that is conceptualized as a specific action when we recognize it as something that belongs to our familiar form of life (Von Wright 1971, III.9, p. 114.). We can recognize an item of behavior as some form of rational or goaldirected behavior, i.e. we can see what the agent was after, what was in some sense of value to him. We can do so because we classify it in familiar patterns of everyday human life. If an item of behavior is so strange to us that we cannot see any goals or purpose behind it, we not only cannot explain the action by reasons—we will hardly be able to see it as intentional action at all.

II. Actions Are Not Physical Events Accompanied by Mental States Causalists who want to reduce reasons to causes usually assume that actions are physical happenings preceded or accompanied by some mental occurrence. What distinguishes the rising of my arm from the raising of my arm? This question invites the answer: my will or an intention or a motive or something else which is conceived of an antecedent or accompanying mental state (maybe a so-called “volition”) which is the cause of the bodily movement. How this mental state causes an action is not quite clear, but, according to the causalist’s assumption, there must be something like mental causation, maybe some sort of volitional intervention into physical causal chains. But this answer is in fact misguided for several reasons: 1. To act intentionally is not the same as to act with an intention. Not all of our intentional actions are preceded by an intention,

An Anti-causal Theory of Action  219 will, decision, choice, plan, a.s.o. Of most of our everyday trivial actions, like walking, tying our shoes, or driving a car it is appropriate to say that we just perform them—without intending each single intentional action, let alone pondering, deliberating with ourselves, or doing calculations prior to the action (Melden 1961: 202f.). The same holds e.g. for the single movements of our hand when we brush our teeth—each movement is intentional, but it would be inadequate to say that we formed an intention for each single intentional movement (Hacker 2010: 148f.). Actions can be voluntary and intentional without being preceded by something that could be properly called “will”, “intention”, or a process of intention forming. 2. Willing, deciding, and intending are no mental states. Mental states are paradigmatically pains or feelings (Hacker 2010: 228f.). They can be more or less intense and may start suddenly and fade slowly or the other way around. They cease to be when we lose consciousness or when we fall asleep.2 But nothing like this can be said of volitions and intentions. One can have the strong will to do X but that means only that the agent overcomes obstacles in the realization of X, not that she has a strong mental occurrence. I can have the intention to write a doctoral thesis or to build a house—actions that take months or years to perform. But do we want to say that I  was therefore in a certain mental state for say three years, even when I was asleep? (On what criteria would we ascribe such a mental state—independently from the displayed behavior in an appropriate context?) While asleep, I do not lose my intention to write my thesis, despite very different neural patterns occur in my brain. One may feel a desire or an urge—but these feelings are not identical to intentions or the will, for one can want or intend something without having any feelings whatsoever shortly before the action and while performing it. When someone intends to kill someone and prepares the foul deed, the only thing present in his mind might be the hate for the victim or the sweet thought of revenge. But he does not even need to feel or think that and just be calmly occupied with his preparatory work. Anything could be present in his mind while, or shortly before, he acts; nothing specific must be present in his mind when he acts in order to say that he acted with an intention. 3. If volitional states were mental occurrences (construed in a dualist way) we could never be sure if others have an intention, will, or came to a decision. According to the Cartesian dualist picture, intentions would be private inner objects, locked in each person’s mind, so that in the end we only could guess if some other person has an intention or not, whereas we for ourselves have the best epistemic access there can be towards our own intentions. But this picture is totally misguided. According to the way we actually use the expressions of “intention”, “will” a.s.o. we can determine if a person intends something or

220  Gunnar Schumann not on the basis of everyday, publicly observable behavioral criteria. When an agent does something and displays all behavioral signs of e.g. being non-surprised, of awareness to the environment and being responsive to relevant circumstances, then we cannot help but ascribe to her intentional behavior, i.e. an action. Objection: “But is it not always possible that we are deceived by someone? And does that not show that we cannot be sure about the mental states of others?” No, for it is not always possible to be deceived: First, if deceit is uncovered then it will be uncovered by publicly observable behavioral everyday criteria and not by metaphysical investigations into his “mind” nor by neurological investigations of his brain. Second, our concept of “intention” is not such that it would allow for the constant danger of being ascribed erroneously: The question if someone really intends to do X, when he displays behavior over years that points to his intentions does not make sense. It would not make sense to ask, e.g., if someone who speaks of building a house for years, saves money for it, buys a parcel of land, commissions a constructing company and supervises the execution of the project for months or even years, really intends to build a house. We would not know what it could mean if we were told by a Cartesian dualist who took a look in the agent’s mind or a neurologist who took a look in the agent’s brain, that this agent in fact lacks the intention to build a house.

III.  Actions Are Not Caused by Mental States First, To intend something can furthermore be no “inner action” for Ryle’s famous infinite regress argument (Ryle 2000). The will or an intention could also not be anything that we experience, since experiences happen to us and could not explain what they pretend to explain: intentional actions. Intentions are not something passively experienced, like a perception—for otherwise we could not distinguish between involuntary and voluntary behavior. There is indeed something like mental causation, but this is not the right kind of thing to think of voluntary action: when I think of something disgusting and have to vomit, or when I am convulsed with pain. But these are just paradigm cases of involuntary behavior, not of intentional action. In the same way, an intention (maybe the intention of killing someone) may make me sweat or cause me to shiver. But it is mostly not an intention to shiver that causes me to shiver and if it were, it would be only an accidental fact that it made me shiver. Here the causalist faces the well-known problem of deviant causal chains. Second, Causalists might be divided on the issue of on how intentions are to be understood—as entities in Cartesian mental realm or as neurophysiological entities. They might also disagree on whether causality is

An Anti-causal Theory of Action  221 a counterfactual or a nomic relation or even something else. What they usually all accept is what may be called a “Humean notion of cause”, i.e. that the cause can be described and identified in a conceptually independent way from its effect. Since Hume’s criticism of the Rationalist notion of cause, it is commonly held that the relation between cause and effect is a contingent one and can be known of only by experience. As Hume’s argument goes, we could analyze the notion of a specific cause as much as we want—yet never will we be able to derive its effect unless we have seen that cause in operation. If a person with normal perception and intellect but no experience would be placed in the world, she never would be able to predict the effect of an observed cause (Hume 2008: Section IV, Part I). Now, if this is true, a decisive argument against causalism in action theory can be found. For the connection between an intention and its respective action is conceptual, not contingent. This has been called the “Logical Connection Argument”. Its point is not, as it has been often misunderstood, that if an agent has an intention, he will as a matter of necessity perform the intended action. From an intention one cannot derive logically that the relevant action will take place. The point of the “Logical Connection Argument”, as I  see it, is that our concept of intention is such that we would not ascribe an intention to an agent when he does not perform the relevant action although he has the opportunity to do so. “Opportunity” here just means that the agent has the required skills to perform the action, has not forgotten about his intention, and is not hindered by external factors (as when he is tied up) or by internal factors (as when he is psychically paralyzed). It would make no sense to ascribe to an agent an intention of doing X when he does not do X—although he has the required skills to perform the action and has obviously not forgotten about his intention, the time to act has come, and he is not hindered. Also, it would not make sense to stay neutral about the ascription of an intention—his behavior under these circumstances shows that he has not the intention to do X. No Cartesian metaphysical look into the agent’s mind and no neurological examination of the agent’s brain could tell us that he in this context still has the intention to do X—for we could not know what this was supposed to mean. Let me explicate this a little further: For a start, note that “to intend x” or “to have a goal” are no ordinary predicates. They do not refer to what could properly be called a “mental state”. Intentions are not something felt; they are not occurrences in our mind, nor in our brain. They also describe no mere disposition to act but express an agent’s commitment towards an action. This becomes obvious in the first-person case: When we say: “I have the intention to do X” or “I have the goal of doing X” then usually we do not report that we

222  Gunnar Schumann have a certain mental occurrence or describe ourselves as being in a mental state. (We also do not report the occurrence of a neurological state.) These sentences cannot be taken to be reports or descriptive statements, for they cannot be taken to be true or false at all. What is true or false is that which can be known or what can be doubted and investigated and be found out. But it makes no sense to say: “I thought I intended to do X, but now I see that I was wrong” or “My goal is either A or B—I haven’t found out yet. Let me check again”. Admittedly, we can say “I  don’t know if I want A or B”—but by this we usually mean only that we have not made up our mind—not that the matter is settled but we have not discovered yet what we really want. It also does not make any sense to say “I know that I intend to do X” in contrast to “I doubt that I intend to do X”—as if it would be a matter of good evidence for that claim. Only maybe when someone else annoys us by repeatedly telling us that he has found out that we intend to do X it could make sense to say: “I know that I intend to do X”—but by this we are just telling him to stop telling us. Or when we watch ourselves on video and describe what we were doing earlier—in such a case “Now I  have the intention to do X” is a descriptive statement. But these cases are very unlike the usual sense in which we say that we have an intention. In the regular first-person use it is simply not a matter of finding out, having evidence, believing, doubting, or knowing what we intend. Nonetheless, we can say what we intend. This is not due to some knowledge gained by introspection, and neither due to some special knowledge at all (“practical knowledge” as Anscombe would have it), but simply due to the authority to express our self-commitment. It is misleading to speak here of knowing or being aware of our intentions at all. We set intentions and goals ourselves, we form an intention. We do not describe a fact about our mind or our brain when we say that we intend something or pursue an end, but we express a commitment of ourselves, a norm under which our (future) action is to be seen. In short, it could not be said that someone intends to / has the goal of doing X unless he performs X or at least starts to perform X. This is a conceptual truth about our concept of intending/of having a goal. We would not know what it means to ascribe to someone an intention/goal when the agent does not perform the action when she has the opportunity to do so. When the agent fails to perform the action although she is capable of doing the action, knows what means she is to employ, is not physically or psychically hindered, and has not forgotten about her intention, then the ascription of having a goal loses any credence and we would take it back. The ascription of goals to an agent only makes sense insofar as she displays behavior of being committed to the performance of the action. We have no independent criteria of goal/intention ascription apart from the fact that the agent performs the action when she can. It would be misconceived to think of the expression of an intention as

An Anti-causal Theory of Action  223 a report of an accompanying state that “lies beyond” the behavior. The intention does not lie behind its behavior, but manifests itself in it (Baker/ Hacker 1993, 54f.). It is helpful here to differentiate between basic actions on the one hand and non-basic actions on the other. Basic actions, like raising an arm, nodding your head, or moving your leg, are those that can be performed directly without help of another action. The physical aspect of basic actions are simple bodily movements. Regarding these, the only criterion to ascribe an intention is usually the performance of the movement—in a context that allows for the ascription of intentional action at all, i.e. the agent has to show the behavioral criteria of voluntary action such as being not-surprised by her bodily movement; acting in response to request or instruction; the ability to repeat the action; sometimes displaying care in performing the movement. Non-basic actions like “going to the opera” or “writing a doctoral thesis” (a fortiori collective actions like “conquering Rome”, “assault the Tsar’s residence”) involve a multitude of different bodily movements that make up the physical aspect of the total action. Here the ascription of an intention of the total action requires either the performance of the total action or at least the performance of some intermediary actions required to perform or complete the total action when the time has come. (It is this non-basic type of action, especially those which also require some sort of calculation or planning, that is relevant in historical explanations.) But again, when the agent does not display behavior that can be seen as the realization of the goal or its parts, or does not perform or complete the total action when she has the opportunity to do so, then we would take back the ascription of the respective intention.3 At this point, one might wonder if I want to make the dubious claims that the intention to do x is the same as doing x or that we only may ascribe intentions after the action was done (‘ex post actu’ as von Wright 1971: 117 suggested)? No, for we may of course ascribe an intention to an agent without her performing the action—but only when the action is, as it were, already in the “background”, as e.g. when the agent is obviously hindered in executing her intention. But even then she has to show some behavioral signs of her intention, e.g. becoming upset or struggling when hindered to perform X, finding another way to do X when the original means is not available anymore, etc.—i.e. signs which are typical in situations in which the execution of intentions faces obstacles. (These signs are what is usually referred to as “trying”. “Trying” is no form of inner experience and not intelligibly applicable to the successful performance of basic actions. When I raise my arm under normal, nonpreventing circumstances, it is not appropriate to speak of that I  first tried to raise my arm).4 When the ascription of intentions to long-term, mediate, and stepwise (i.e. non-basic) actions is considered, like going to the opera tonight or writing a doctoral thesis, then the agent on the one hand needs to display a kind of behavior which does not contradict

224  Gunnar Schumann the intention ascription. In order to be sensibly ascribed the intention to go to the opera in London tonight she must not board a plane to China the same day (von Wright 1971: 111). On the other hand the agent must also perform preparatory actions that will enable her to perform or complete the intended action when the time to act has come. Otherwise we would doubt the sincerity of her intention expressions or assume that she changed her plans or her means-end-beliefs. So, intention ascriptions require the action itself, or preparatory steps undertaken to perform the whole action, or the struggle against hindrances of the action such that the ascription of the intention to do X is dependent on some description of behavior which refers to the intended action. From such cases of nonbasic actions or cases of hindrances the expressions “to have the goal” and “to have the intention” get their sense and meaning. They do not denote the possession of (Cartesian or neuronal) entities, as it is commonly misunderstood by causalists, which are taken to be causes that effect the action. To think of an intention as a something (be it a mental or physical object or state or process) is to take the expression “having an intention” too literal and to be led astray by its surface grammar. The expression is taken to have the same linguistic role as e.g. “having a car”. But in fact it indicates the possession of an entity as little as “having a lot of work to do” or “having fun”. In non-basic actions the third-person-ascription of intentions is also possible on the basis of the verbal avowal of intentions by the agent (cf. Schurz 2011: 164). But this basis for intention ascription requires that we have no doubts about the agent’s sincerity. Mostly and normally, we do not have doubts about the other’s sincerity of intention expression. But this is often because we know the other agents to be sincere by having sufficiently often done what they announced to be doing or because we simply presume their sincerity unless proven insincere. (And, it is easier for us to believe the verbal expression of an agent’s intention when we can be sure that the intended action conforms to the agent’s general wishes or does not contradict it, but we are more hesitant in our fullbreathed intention ascription when we know that it does contradict the usual wishes of the agent.) In the end, the only touch-stone or criterion of the sincerity of an agent’s avowal is that she performs the action she said she intended when she has the opportunity to do so.5 The point I am trying to make here is not an epistemological one. It is not that we could not know of a goal/intention of an agent unless she acts accordingly—and that the goal/intention was already in her mind (or brain). The point I am trying to make is a semantical one: the very concept of an intention/goal implies that its ascription is only meaningful when there is an according behavior, i.e. the performance of the action itself or preparatory steps towards its performance. Intentional action and intention ascription are conceptually interwoven; the publicly observable expressive behavior is part of the meaning of mental concepts “having a

An Anti-causal Theory of Action  225 goal” and “intention”. That the point really is one of the semantics and not of the epistemology of intentions is shown by the fact that in the case in which an agent with an alleged intention does not perform the required action (although she has the opportunity to perform it), we would not only refrain from ascribing the intention or leave the matter undecided (or leave it as an open question until decided by a future neuroscientific scan)—we would say that the agent in fact lacks the intention to do x. Davidson (1963: 696) and Stoutland (1970) replied to the Logical Connection Argument that the fact that reasons and actions can be described in a way that their relation is logical does not rule out that both are also related causally. But the point of Melden’s version of the Logical Connection Argument was that the only way to describe an agent’s intention is via the action the intention is an intention to. Now Davidson replies that there are very well independent means to identify an agent’s intention, e.g. feelings or actions that are not rationalized by these intentions (Davidson 1963: 696). In other words, there be not one single test for intentions but several ones, such that intentions can be ascribed to an agent without invoking the action the intention is an intention to. But I think that this is wrong. (1) I think it is telling that Davidson does not give an example here for such an independent identification of an intention. It is hard to think of one. Admittedly, one can think of several ways of how wants or desires can be identified and described independently of the action that satisfies the desire. (I can very well have the desire or want to eat a cookie without doing so, although I am capable of eating it and I  am not prevented from doing it.) Desires and wants can be expressed by bodily reactions (sweating, accelerated pulse, intense breathing, shaking); nonverbal behavior (looking at the cookies, walking around the cookies); or verbal behavior (“May I have some cookies, please?”, “I want cookies!”). But my argument makes explicit use of the notion of intention, not desires, wants, or wishes. One can wish or desire to eat cookies without intending to eat some (for, e.g., it is too embarrassing to have the cookies or because one is on a diet) and one can intend to do X without wishing or desiring to do X (like going to work on a Monday morning). Either Davidson is led astray by his own terminology of “desire” or he intentionally picked it to make his causalism look more plausible at this point. (2) Unlike desires, intentions are not felt and do not have to be accompanied by feelings (as already mentioned above). I need not feel anything particular in order to be said to have an intention. In fact nothing specific must be in my mind at all in order to be said to have an intention. Also, if an intention was a feeling, it would be a passive experience or affect which is incompatible with the idea of intentional action. (3) Admittedly, Intentions may be identified without the actual occurrence of the action and agents can be said to have an intention prior to their respective action. But, as already argued for, this applies only to intentions to non-basic actions or to cases of prevented

226  Gunnar Schumann basic actions. And, even ascriptions of intentions to non-basic actions require some intentional action(s) that are rationalized by them, like preparatory actions (to bring oneself in a position to be able to perform/finish the action when the time has come), to smooth out hindrances of the action, to overcome obstacles and resistances by other agents. The most common way by which we identify intentions of agents is probably that they announce the action. But any utterance of the form “I will do X/I have the intention do X” must be sincere in order to count as a proper expression of an intention. And “sincerity” just means that the agent performs the action she announced when she has the opportunity to do so. Now, the logical connection between an intention and the action consists in that whoever forms the intention to do X will have to do X when she has the opportunity to do it. It is not that she has to do X in the sense that she has no other choice. Her body will not set itself in motion. When she forms the intention to switch on the lights at 6 o’clock, it is not enough to just wait until it is 6 o’clock and then—look!—her arm goes up and operates the switch (Wittgenstein 1984: § 627; Hacker 1996: 582). It is in the sense of the normative “must” that she must act. This means when she does not perform the intended act, she cannot be said to have the supposed means-end-belief or the intention. But when she does have the intention (and a respective means-end-belief) and does not act when the time has come, she commits an error—the error of not being coherent. First-person expressions of intentions are thus no predictions of our future behavior; rather, they express standards according to which our future behavior can be measured as the successful or unsuccessful execution of what we committed ourselves to do. “I intend to do X” is not a statement about a cause, about what will most likely make me do X, in fact it is an utterance with normative force. If someone is constantly telling us that he is intending to do X, then we may say that he will have to do it someday. But if the relation between an intention and an action was a causal relation, then that normative expectation would be out of place. Intentions can thus be understood as norms, as self-commitments of our future behavior. Goals and purposes are judged by the agent to be something good in the sense that they are to be pursued/achieved from the standpoint of the agent. In this sense, expressions of intentions in the firstperson use can be understood as having prescriptive force.6 This is an important point because it helps to understand the nature of a practical syllogism correctly. Practical syllogisms are not theoretical syllogisms in the sense that the derived action can be predicted. Nonetheless the concluding action follows with necessity from the premises. The function of a practical syllogism is to show—given the premises—what action must or should be done by the agent in order to be called “rational”. From the intention to do X and the belief that Y is a necessary means to do X it follows with necessity that Y is to be done (from the standpoint

An Anti-causal Theory of Action  227 of the agent). A practical syllogism displays normative thinking, for its volitional premise and its conclusion have a prescriptive linguistical role (in the first-person use). The volitional premise of a practical syllogism thus designates no cause and could not. The intention, which can be said to be a reason for the action, is related to the explicandum in a normative way: It shows why the resulting action was to be done or rationally appropriate to do given a suitable belief premise; it does not show that the action was to be expected as in a causal prediction. It could be objected that it is only in the first-person use of a practical syllogism that it might have a committing role. Since we are concerned with the explanation of human actions, we are to take the standpoint of an observer. So it seems that the previous considerations about the linguistic role of intention ascriptions in first-person use might be beside the point. In explanations of action, we certainly describe an agent as having an intention and so the conclusion of the explanation will be a descriptive statement. But this objection would miss the point. It is true that in explaining human actions we take the standpoint of an observer; even so, our concept of intention is such that the first-person use of “to have the intention to do X” is prescriptive (in regular cases) whereas its third person use is descriptive. Nonetheless, both uses are part of one and the same concept (it is not that “intention” means something different in “I have the intention to do X” and “He has the intention to do X” in the same way that the word “bank” in “I will meet you at the bank” could either mean that I meet you at the river or at the credit institution). We have grasped the concept of intention (or of a goal) only when we know that it expresses a commitment in the first-person use and describes a highly specific form of publicly observable behavior of an agent in the thirdperson use.7 Regarding other’s actions, I  already mentioned that we recognize an item of behavior as an intentional action when we can classify it in familiar patterns of everyday human life. So, the need for an explanation of an agent’s action does often not occur. Only when we are to some extent surprised or baffled by an agent’s behavior or cannot immediately see the purpose behind her actions we may feel the need for an explanation. Then we may either ask the agent for his reasons and/or investigate the context of the action. We have to widen the perspective on the action to identify the reasons, i.e. the intentions and beliefs of the agent. Why did this man on the busy street raise his arm, smile at us, and say something to us? We do not know him. Does he want to sell us something? He does not seem to have any products with him. Does he want to cadge money from us? No, he does not behave like that. Finally, we realize that he was greeting another person right behind us, for now they walk to each other and shake hands and the man does not notice us anymore.

228  Gunnar Schumann Such contextual investigations are also called for when we wonder why an agent performed a certain action. The fact that we can differentiate between possible reasons for an agent to act and the one or the ones for which he really acted, such that we may speak of the reasons that “effected” the action (as Davidson misleadingly did), does not speak for the causal character of these reasons. That a certain reason was in fact the reason for which an agent acted is determined by the context of the action, not by discovering some causal antecedent of the action. If it truly was patriotism (as he said) or just the high pay that Smith volunteered for the army depends on a variety of factors of the context of his action, especially what he did on similar occasions in which he had to choose between patriotic and greedy actions, maybe even how he will act in the future (cf. Ryle 2000). Often, the question, for which reasons an agent acted, can be decided from the third-person-perspective. And if the criteria available speak for Smith’s patriotism, then his patriotism and with it, his intention to serve his country, is expressed in his volunteering. However, if even after longer examination of the context of the action no exact identification of an intention is possible, then there simply is no answer to the question what the real reason for Smith’s action was—for there is nothing that could count as a proper answer (Hacker 2010: 225), except maybe Smith’s saying what his intention was (but only if he has proven to be sincere before).

IV.  Historical Explanations Regarding actions of contemporary agents we are in the epistemically comfortable position that we might ask the agents or observe the action and its context first-hand. Regarding past actions, it will be harder to reconstruct the action’s context because of the decreasing quantity and quality of historical sources. Historical agents only rarely stated explicitly what their intentions were. But they displayed actions, knowledge of which has come down to us today from historical sources, either as material or verbal records. Historians comb through the bequeathed source data, to get a picture as precise as possible concerning the context of past actions and to reconstruct from it the intentions, goals, purposes, and beliefs of the historical agents. In cases in which the reconstruction of the context of the action is not possible, historians will have to abstain from ascriptions of reasons, purposes, and intentions of the agents, and therefore from explanation. The explanation of past actions admittedly has special problems, foremost those which have to do with our knowledge of the past action’s context, but those are no problems sui generis, such that it can be said with some qualifications, that historical explanations work like everyday explanations of contemporary’s actions: As the crossing of a street by a contemporary agent under certain contextual conditions is the expression of her intention to go to work, so Caesar’s

An Anti-causal Theory of Action  229 crossing of the Rubicon is the expression of his intention to seize power in Rome under the historically transmitted circumstances of his action— not: the intention to seize power caused Caesar to cross the Rubicon. The data material which has come down to us contemporaries is the interpretandum from which the historian derives goals and intentions of historical agents and with which further actions of the historical agents can be interpreted. So, to explain an action by a reason is not to refer to a previous event which is a Humean cause of it, but to embed the action in a context such that it can be understood what the agent went after. Often, it will thus be understood what the action was (Melden 1961: ch. IX; von Wright 1971: ch. III.8). The task of the historian then is to identify the intentions, purposes, and goals, sometimes the means-end-beliefs of historical agents to make their bequeathed actions intelligible to us. (Of course this only concerns non-basic actions—but historians commonly only deal with those.)8 In particular, this is his task when some actions of historical agents are not immediately understandable; when we cannot see what they were for. Such is the case, e.g. when we are told that the prima facie unintelligible action of the Roman emperor Caligula appointing his favorite horse to senator was done as a means to humiliate the senators of Rome as a part of Caligula’s wider intention to consolidate the still somewhat shaky sovereignty of the emperor against the senate (Winterling 2012). We understand what the action was in fact: The appointing of the horse was in fact a humiliation of the senators. The task to make historical actions intelligible to us can also involve reconstructing means-end-beliefs of historical agents which are not commonly held today, such as e.g. when historical artifacts such as plague masks are explained by historians of medicine by the people’s belief that they could protect themselves from infection by breathing through beaked masks filled with odorous essences. Historians love details—not because their genuine method consists in describing the particularity of their subjects as has been wrongly held by early pluralists, but because the reconstruction of the intentions, goals, and purposes of historical agents requires good knowledge of the details of the situation and the context of the historical agent’s actions. For only then can they decide if there was an overarching intention in the performance of a particular action and if so, what it was and so for means-ends-beliefs. This also sometimes requires pointing out several occasions in which the historical agent acted in a similar way—just to gain evidence for a plausible reconstruction of the character or principles of action of an historical agent. So, there can be some regularity involved in the investigative work of the historian, but not as laws or probabilities but as evidence to identify certain intentions of historical agents. Making past actions intelligible to us by reconstructing the intentions is also applicable to collective actions. Collective actions are not to be explained by assuming plural subjects, for there is no single collective

230  Gunnar Schumann agent in analogy to individual agents. Collective actions are not to be grasped as being the resultant of individual intentions of a special kind (Searle’s “we-intentions”), for the idea that members of an acting collective each have an intention only because they can be said to share an intention is to stick to the wrong picture of intentions as mental or neuronal entities. But an intention to do something is not an entity at all; to have an intention is not to possess anything at all. It is, as I argued, a way of behavior that can be understood in its context as being bound by a self-commitment. A collective intention is the behavior of the individual members of the group that can be understood in its context as being bound by a self-commitment. So the cracking of a safe by Jim, the waiting of John in the getaway car, and Dick’s keeping the hostages in check are all single actions which in sum form an action that can be understood as the robbing of a bank—what all three agents intend to do. They share one and the same intention, namely: to rob the bank, and everybody’s single action can be seen as a part of performing the intended total action. In this sense the performance of a collective action can be grasped as the performance of a single long-term action. In the latter case the total action is divided into partial actions which are performed consecutively by an individual agent, whereas in the former case the total action is divided into partial actions which are performed simultaneously (or consecutively) by different agents. The context of the performances of the actions makes it clear that the partial actions are parts of the performance of one collectively intended action. To that context may belong the fact that the shared intention was expressed together (a mutual oath or promise or contract)9 or verbally expressed by one and more or less explicitly affirmed by the others (as it is common in political demonstrations and rallies with a crowd and a speaker [“We stand up for XYZ!”]). The context may even only consist in joint behavior, as when Peter helps me push my broken car without me having asked him. Here our nonverbal behavior shows our common intention. Our joint pushing of the car expresses our intention—not, as e.g. Bratman would have it: the shared intention effects our pushing together.10 The contemporary debate about collective intentions is flawed, for most philosophers assume that intentions are entities possessed; that is, they take an expression like “We have the intention . . .” too literally, as if it would mean “We possess some thing . . .” and then they are perplexed by the question, how reified individual mental states make up for a collectively shared one. Sometimes historians explicitly speak of “causes” of past human actions. But this is only because in everyday language, of which the historian makes use, we do not usually differentiate between reasons and causes in a strict way. Both terms can be used interchangeably in several contexts. We speak of the causes of the French Revolution or of the Great War, although in fact human actions are concerned here, and sometimes we speak of the reasons for a physical happening, e.g. an explosion. Nonetheless, there is a conceptual distinction to be made here: The

An Anti-causal Theory of Action  231 spark was the reason of the explosion, i.e. its cause, not the reason for the explosion to happen. We may say that imperialism causes wars, or blatant social inequality causes uprisings, or even that the draught led the dust bowl farmers to leave their home. But strictly speaking, it is not the drought that caused the dust bowl farmers to move to California. The drought did not make the farmers go to California, but the drought changed the conditions for the farmers such that they had to reach their life goals in another way. And those life goals have not been something exotic or alien to us, but are known and shared by nearly all members of our life form (cf. Hart & Honoré 1966: 235f.): safety for oneself and the family, health, wealth, recognition, freedom of expression, a right to self-fulfillment, and so on. Those goals are so natural that they are often not stated explicitly in historical explanations. Often only the prominent change is stated such that the impression of a nomic or causal connection between drought (or war) and migration might occur; that it is the drought that drives people from their homes. When the drought is mentioned as the explanation of the migration, the explanatory work is not provided by a relation of causality or of law or of probability between both events. In fact the drought changed the situation which the agents faced, such that they changed their means-end-beliefs and tried to fulfill their life goals in another way, i.e. by moving to California. The explanation remains a teleological one, although there even might be a regularity between droughts (or wars) and migration. And, sometimes, genuine causal statements are found in historiography, e.g. when a historian says that the drought led to soil erosion or, even more often, when the causal effects of human actions (the effects being intended or not) are concerned. Thus, it is not uncommon that a historian may say that it was human exploitation of the land that led to soil erosion. But proper causal chains as these are not the main subject of the historian, for here the historian relies on the results of the physical sciences. The historian does not usually deal with the explanation of natural happenings. Therefore, there is a good reason that historians usually do not do paleontology, geology, or astronomy. It is not because an event is past that it is of interest to the historian. Historians only deal with past events that can be described in an action language, not in a naturalistic language suitable for physical processes. That historical explanations are teleological explanations does not even forbid predictions of future actions of men. It is usually not the task of the historian to make predictions, but it is not inconceivable that they do. In everyday life, we can predict what an agent will do under theseand-these circumstances given that we know her character and/or her general goals. The same holds for historical agents. But these predictions will not be very physically detailed: No historian will be able to predict the movements of the agents or determine exactly at what time the action will take place. The predictions in social sciences and historiography rest upon knowledge of the agent’s intentions and goals and their

232  Gunnar Schumann beliefs about the situation. Thus, what will be predicted is the action, not the agent’s movements. Also, these predictions have a rather limited range. No predictions of the far future are possible as they are possible in e.g. astronomy. Only insofar as we are justified in believing that the intentions of the agents will be such-and-such and that their beliefs about the situation will be such-and-such will we be able to predict their future actions on a general level.

Notes 1 For comments and corrections on earlier drafts of this paper I am grateful to Alfred Mele, Elena Popa, and Joan Arbery. 2 Although being asleep and being unconscious are not the same state, of course. 3 This claim needs a qualification for actions described by success verbs (like “scoring”, e.g. in football): “.  .  . display behavior related to the goal in a context that makes clear that he tried the action in question . . .”. 4 For a discussion of “trying” as alleged act of will that causes the action, cf. Hacker (1996: 568–575). 5 Or something similar to X as when an agent missed his goal, but lets us see that he tried. 6 That is why Anscombe can identify the crucial difference between expressions of intentions and descriptions in the fact that when actions fail the “mistake is in the action” but not in the language, e.g. when I say: “I will now press Button A” but instead press Button B by mistake. The standard of correctness was my expression of the intention, not some fact in the world. A fact would have been the standard of correctness when my uttered sentence would have had the linguistic role of a description of what I was doing (Anscombe 1963: § 32). 7 It is like to ascribe a behavior as norm-guided. We can describe a behavior as norm-guided and this is in itself a description (truth-apt), but this does not make the norm the agent follows a descriptive statement. 8 Historians usually only deal with longer-term, mediate actions, like political, military, social, economic artistic, religious scientific actions. The very concept of these actions implies that they can only be done for a purpose (cf. Collingwood 1993: ch. V, § 5). 9 So what distinguishes people in the park running for shelter from rain individually and being members of an outdoor ballet performing a collective action just is dependent on the context of the individual’s actions. To the context here belongs that the members of the ballet have agreed on doing a common performance and been instructed. Or that, would one or some people fail to perform the required action the director would shout “cut” or being irritated. But surely it is not the having of a special type of intention (“we-intentions”) tokens of which all members of the collective possess, as Searle (1990) and Gerber (2012) seem to think. 10 Bratman speaks of “mechanisms” (e.g. Bratman 1999: 114).

References Anscombe, G. (1963). Intention, 2nd ed. Oxford: Blackwell. Bratman, M. (1999). Shared Intention. In: M. Bratman, ed., Faces of Intention: Selected Essays on Intention and Agency, 1st ed. Cambridge: Cambridge University Press, pp. 109–129.

An Anti-causal Theory of Action  233 Collingwood, R. (1993). The Subject-Matter of History. In: J. van der Dussen, ed., The Idea of History, 2nd ed. Oxford: Clarendon, ch. V. Epilegomena, § 5. Davidson, D. (1963). Actions, Reasons and Causes. Journal of Philosophy, 60(23), pp. 685–700 (reprinted in: D. Davidson, D. (2001). Essays on Actions and Events, 2nd ed. Oxford: Clarendon Press). Gerber, D. (2012). Analytische Metaphysik der Geschichte: Handlungen, Geschichten und ihre Erklärung. Frankfurt: Suhrkamp. Hacker, P. (1993): Wittgenstein. Meaning and Mind. An Analytic Commentary on the Philosophical Investigations. Vol 3, Part 1, Oxford: Wiley-Blackwell. Hacker, P. (1996). Wittgenstein. Mind and Will. Analytical Commentary on the Philosophical Investigations. Vol. 4, Part 1. Oxford: Wiley-Blackwell. Hacker, P. (2010). Human Nature: The Categorical Framework. Oxford: Blackwell. Hart, H. and Honoré, A. (1966). Causal Judgement in History and in the Law. In: W. Dray, ed., Philosophical Analysis and History, 1st ed. New York: Harper & Row, pp. 213–237. Hume, D. (2008). Enquiry into Human Understanding. Oxford: Oxford University Press. Melden, A. (1961). Free Action. London: Routledge and Kegan Paul. Mele, A. (2000). Goal-Directed Action: Teleological Explanations, Causal Theories, and Deviance. Philosophical Perspectives, 14, pp. 279–300. Ryle, G. (2000). The Concept of Mind, 2nd ed. Chicago: University of Chicago Press. Schroeder, S. (2001). Are Reasons Causes? A Wittgensteinian Response to Davidson. In: S. Schroeder, ed., Wittgenstein and Contemporary Philosophy of Mind, 1st ed. Basingstoke: Palgrave Macmillan, pp. 150–170. Schurz, G. (2004). Normic Laws, Nonmonotonic Reasoning, and the Unity of Science. In: S. Rahman, ed., Logic, Epistemology, and the Unity of Science, 1st ed. Dordrecht: Kluwer, pp. 181–211. Schurz, G. (2011). Erklären und Verstehen: Tradition, Transformation und Aktualität einer klassischen Kontroverse. In: F. Jaeger and J. Straub, eds., Handbuch der Kulturwissenschaften. Stuttgart: J.B. Metzler, pp. 156–174. Searle, J. (1990). Collective Intentions and Actions. In: P. Cohen, J. Morgan and M. Pollack, eds., Intentions in Communication, Cambridge: MIT Press, pp. 401–415. Stoutland, F. (1970). The Logical Connection Argument. American Philosophical Quarterly. Monograph Series, 4, pp. 117–129. Stueber, K. (2012). Understanding versus Explanation: How to Think about the Distinction between the Human and the Natural Sciences. Inquiry, 55(1), pp. 17–32. Von Wright, G. (1971). Explanation and Understanding. London: Ithaca. Von Wright, G. (1985). Of Human Freedom. In: S. McMurrin, ed., The Tanner Lectures on Human Values. Vol. VI, 1st ed. Salt Lake City: University of Utah Press, pp. 107–170. Winterling, A. (2012). Caligula. Eine Biographie. München: C. H. Beck. Wittgenstein, L. (1984). Tractatus logico-philosophicus. Tagebücher 1914–1916. Philosophische Untersuchungen. Werkausgabe. Vol. 1. Frankfurt: Suhrkamp.

13 Meanings and Mechanisms An Actor-Centered Approach to Historical Explanation Daniel Little

In this chapter I will argue for three central ideas. First, historical and social events are amenable to causal explanation. Second, the best way of understanding social and historical causation is through discovery of underlying causal mechanisms and powers: real concrete social arrangements through which one social event or condition produces another event or condition. And third—and most important—I will argue for a particular view of the substrate of social causation: socially embedded actors who act out of an expression of their values, goals, emotions, beliefs, and mental frameworks in the settings of a range of institutional and structural circumstances. This view of historical explanation results in a nice resolution of the key riddle within the philosophy of history, concerning whether historical knowledge is causal or interpretive. With an actor-centered understanding of the nature of social causation, we are enabled to maintain the validity of causal explanations of social and historical outcomes while at the same time recognizing the meaningfulness and intentionality of human action. In order to explain historical outcomes we must both interpret actors and identify causal connections among institutions, value systems, and individual actors.

Social Causation Causal realism is the view that causal processes and powers exist in the world, and that it is an important task for science to uncover them. The view is associated with a wide range of realist thinkers, from Richard Boyd (1984) to Roy Bhaskar (1975, 1989) to Peter Hedström (2005). Causal realism is a defensible position when it comes to the social world. There are real causal relations among social factors (structures, institutions, groups, norms, and salient social characteristics like race or gender). Social structures, entities, and forces exist in the world; they are composed of socially constituted human beings; and they have causal powers and properties that derive from the circumstances of action of the individuals that make them up. We can give a rigorous interpretation to claims like “agrarian reform was impeded in sixteenth-century

Meanings and Mechanisms  235 France by the system of overlapping jurisdictions involved in the French feudal state” or “the extension of rail networks in North America caused changes in patterns of habitation”. Such an interpretation, in the first example, depends upon discovery of the specifics of the legal and institutional arrangements through which power was exercised, and a demonstration of how those arrangements led to the specified forms of behavior. It is therefore perfectly appropriate to conclude that “X caused Y” has the same meaning, whether in the domain of the natural world or in application to social and historical events. The same complex of causal ideas find application in both domains: necessary and sufficient conditions, counterfactual inferences, the availability of underlying causal mechanisms, and the applicability of the idea of causal powers. What differs in the two realms is simply the nature of the substrate through which causal impulse is conveyed; in the natural world it is the workings of the physical properties of natural entities, whereas in the social realm it is the workings of human agency and the institutions that empower and constrain human actors.

The Case for Social Causation Why are we justified in believing that there is such a thing as social causation? This question can be answered on several different levels. First, the concept of causation is intimately intertwined with ideas about necessary and sufficient conditions and the counterfactual intuitions we have about the social world and that are commonly invoked in interpreting history and the social world. Causal judgments are all but inextricable from our ordinary experience of the social world. And second, we have a good answer to the question, what facts about the social world suffice to give rise to real causal mechanisms and powers? The answer is in common among a broad swath of social scientists, from classical Marxism to the new institutionalism to pragmatist sociology of knowledge: The underlying ontology of social causation is the actions and states of mind of the myriad socially situated individuals who constitute the social world. Consider first our use of counterfactual judgments and judgments of necessary and sufficient conditions in the social world. When we examine a sequence of events A, B, C, O we often have evidence for believing things like these: 1. If A had not occurred then O would not have occurred. 2. The fact that B occurred in the context of A and C guaranteed that O would occur. 3. The occurrence of C enhances the likelihood of the occurrence of O. 4. There are identifiable underlying processes conveying the state of the world in which A, B, and C occur to the state of the world where O occurs.

236  Daniel Little These are thoroughly causal ideas about how the world works. The first statement means that A was necessary for the occurrence of O; the second statement means that B in its context was sufficient for the occurrence of O; the third statement means that C is causally relevant to the occurrence of O; and the final statement means that there are underlying causal mechanisms that explain the necessary and sufficient status of various events. When causal theorists discuss these concepts they usually give examples from the natural world—the occurrence of an accident on an icy road, the occurrence of an industrial explosion, or contraction of a contagious disease. But the same kinds of linkages can be discerned in the social world as well. 1. If Hutu extremists had not taken over radio broadcast facilities, then broad-based massacres of Tutsi individuals and families would not have occurred; that is, radio broadcasts were a necessary condition in the circumstances for the occurrence of Rwandan genocide. 2. Given X, Y, Z, the financial collapse of Lehman Brothers made it likely that world financial crisis would occur; that is, collapse of Lehman Brothers was sufficient in the circumstances to bring about global financial crisis. 3. The rapid influx of MENA refugees into Western European democracies enhanced the likelihood of the rise in popularity of extremist nationalist political movements. These examples show that the same causal structure can be discovered among events and conditions in the social world as the natural world. We can observe and validate relations of necessity, sufficiency, and causal relevance among social events and processes, and we can identify some of the mechanisms that underlie these relations. Causal order in the social world can be observed, investigated, and documented.

Causal Substrate and Causal Mechanisms Here is a more basic reason for being confident that social causes exist. Causal judgments rest upon assumptions about how things work—what the governing processes and powers are that make up the medium of events in question and provide the connective structure between cause and effect. There is a substrate for any particular domain of causation, and the substrate embodies specific features of activity and causal connectedness. It is this underlying causal potency that gives rise to the reality of causal powers attached to things. Features of the substrate constitute the “necessity” or “push-force” that exists between cause and effect. And crucially, we can provide good answers to the question, what is the substrate of social causation and social causal powers?

Meanings and Mechanisms  237 In the case of the world of physics and chemistry, the necessity associated with causation is natural necessity. “Given the fundamental characteristics of natural entities and forces, the occurrence of A brings about the occurrence of B”. In the case of the social world, we need another phrase, but neither “social necessity” nor “historical necessity” exactly captures the idea. We might introduce a new phrase to capture the idea along the lines of “agentic necessity”: “Given the characteristics of beliefs and desires of a set of actors within a specified institutional context, the occurrence of A brings about the occurrence of B”. (Both natural necessity and agentic necessity must accommodate the availability of probabilistic causal powers.) Social causes have a different ontology than natural causes. They are the result of constrained and motivated social actions by concrete social actors, and these actors are not subject to anything analogous to laws of nature. So the idea of natural necessity does not help in the case of social causes. If we wanted to provide a counterpart notion of agentic necessity, it might go something like this: • Given a social environment populated with actors something like this and embodying rules and institutions something like that, change A brings about outcome B through the actions of these ordinary actors. So, for example, the advent of collective agriculture in China in 1958 led to a precipitous decline in agricultural productivity. The institutions and incentives of farming and consumption changed; within the institutions of collective agriculture, ordinary farmers found the choice of becoming a free rider or an easy rider an appealing one. It is readily observed that this is a substantially weaker foundation for stable causal powers of social structures and entities than we have in the natural world. The constituents of social processes—individuals—change over time and place. And the workings of the same institutions and systems of practices and rules will be significantly different if they are populated by actors with significantly different dispositions. (This is one of the central postulates of the idea of “methodological localism” that I have argued for elsewhere: Individuals are socially constituted and socially situated [Little 2006]. It also converges with the idea of an institutional logic that a number of scholars have developed [Thornton, Ocasio and Lounsbury 2012].) Thus there is a reasonably well understood substrate underlying causation in the social world (socially constituted and situated individuals doing things within specific rules and practices) and this substrate does in fact convey a change at one end of a causal process [A—a change in the rules of supervision in an organization, let us say] to a change in the

238  Daniel Little outcome [B—less petty corruption within the organization], through a series of events that are systemic enough to allow us to see the forcing circumstances and processes of the transition from A to B. This formulation focuses attention on what was referred to as “agentic necessity” above. It leads the historian to attempt to uncover the specifics of motivation and institutional setting through which individuals in the given circumstances are led to behave in ways that lead to the generation of B.

The Collingwood View There is a line of thought in the philosophy of history, best expressed by R. G. Collingwood, that opposes the idea that historical inquiry can or should provide causal explanations of historical events. W. H. Walsh provides a particularly clear and sympathetic formulation of this perspective in his short work, Philosophy of History (1968). Drawing upon Collingwood, Walsh maintains that historical knowledge has to do entirely with sense-making. History concerns conscious human action, and the work of the historian is to understand the meaning and purpose of the actions of historical actors. “[The historian] aims [. . .] at a reconstruction of the past which is both intelligent and intelligible” (1968: 32). “What every historian seeks for is not a bare recital of unconnected facts, but a smooth narrative in which every event falls as it were into its natural place and belongs to an intelligible whole” (33). And Walsh introduces a novel cognitive operation as the core intellectual work of the historian, colligation. Here is how he defines this concept: “to locate a historical event in a larger historical process in terms of which it makes sense [. . .]. This process of reasoning serves to establish the ‘inner connections between certain historical events’ ” (23–24). This means that the tools of interpretation of meanings and reasons are crucial for the historian—much as the hermeneutic philosophers in the German tradition had argued. Collingwood and Walsh are not mistaken that history is made by conscious actors, and they are right in maintaining that interpretation of action is a crucial part of historical research. However, they are mistaken in imagining that this fact rules out the possibility of historical causation. Their error arises from an excessively positivist understanding of science and causation, and their implicit assumption that science and causation require strict laws and regularities as their foundation. Here is Walsh’s summary description of “science”: We may sum up the results of this brief attempt to bring out the main features of the common conception of science and scientific knowledge as follows. We apply the term “science” to knowledge which (i) is methodically arrived at and systematically related; (ii) consists of, or at least includes, a body of general truths; (iii) enables us to

Meanings and Mechanisms  239 make successful predictions and so to control the future course of events, in some measure at least; (iv) is objective, in the sense that it is such as every unprejudiced observer ought to accept if the evidence were put before him, whatever his personal predilections or private circumstances. (1968: 36) The naive positivism underlying this account is evident: Laws, predictions, and objectivity are the essential components of scientific knowledge, according to Walsh. Post-positivist philosophy of science has shown that none of these characteristics is essential to the production and presentation of scientific knowledge. The approach taken in this essay is intended to demonstrate that there is no contradiction between an actor-centered approach to social change and a causal interpretation of historical events. And by “cause” here I mean the standard interpretation: The failure of Lehman Brothers caused the global financial crisis, and the substrate of this causal relation was the vast number of individuals within multiple institutions deciding to act in ways that aggregated to financial crisis. Conscious actors are the substrate of social and historical causation; and features of agency in turn underlie the creation of causal powers associated with various kinds of social arrangements and institutions.

Social Mechanisms and Powers Carl Hempel argued that historical explanations ought to have the same logical structure as any other scientific explanation: a derivation of the thing to be explained from a set of general laws and specific statements (Hempel 1965). This view of historical explanation has not been acceptable to most philosophers of history for at least 50  years. Are there plausible intuitions about the ways the social world works that stand as credible alternatives to Hempel’s covering law model? There are. A particularly strong alternative links explanation to causation, and goes on to understand causation in terms of the real causal mechanisms and powers of various entities and structures. Rom Harré’s work explored this approach earliest (Harré and Madden1975), and Roy Bhaskar’s theories of critical realism push these intuitions further (Bhaskar 1975, 1989). The discovery of social mechanisms and powers often requires the formulation of mid-level theories and models of these mechanisms and processes—for example, the theory of free-riders. By mid-level theory I mean essentially the same thing that Robert Merton (1967) meant to convey when he introduced the term: an account of the real social processes that take place above the level of isolated individual action but below the level of full theories of whole social systems. Marx’s theory of capitalism illustrates the latter; Jevons’s theory of the individual consumer as a utility maximizer illustrates the former. Coase’s theory of transaction

240  Daniel Little costs is a good example of a mid-level theory ( 1988): general enough to apply across a wide range of institutional settings, but modest enough in its claim of comprehensiveness to admit of careful empirical investigation. Significantly, the theory of transaction costs has spawned major new developments in the new institutionalism in sociology ( Brinton and Nee 1998). The discovery of social mechanisms and powers often requires the formulation of mid-level theories and models of these mechanisms and processes—for example, the theory of free-riders. Here the idea is that causation is not to be understood along Humean lines, as no more than constant conjunction. (This is where the neopositivist insistence on general laws originates.) Instead, the idea of a real causal power is taken as a starting point. Things have the capacity to bring about changes of specific circumstances, in virtue of their inner constitution (or what Harré is content to call their essences). Nancy Cartwright’s ideas about causation and general laws fall in the same general vicinity ( 1983, 1999), though she is not a critical realist. But her critique of Hume and the search for regularities and her preference for capacities is similar. When causal realism is brought to the social and historical sciences, it brings the idea that there are structures, entities, and forces in the social world that really exist and that derive ontologically from a substrate of activity that give substance to their causal powers. (That is, I endorse the view that social structures and entities are ultimately constituted by the actions and thoughts of the individuals who make them up; but that there is no imperative to reduce the properties of the social entity to a compound of properties of the individuals.) In the case of the social world, that substrate is the socially constituted, socially situated actor, or what I call the premise of methodological localism ( Little 2006). This ontology is helpful for setting a program of inquiry for social scientists and historians. Instead of looking for general laws of a given domain, the researcher is encouraged to discover the particular causal properties and powers of specific kinds of things. This emphasis on the particular and the local is particularly well suited to the challenges of historical and social research. Nancy Cartwright doubts the validity of searching for even exact laws of physics. This doubt is all the more reasonable in the case of social phenomena. It is pointless to look for general laws of bureaucracy, the military, or colonialism. What is more promising, however, is to examine particular configurations of institutions and settings, and to attempt to determine their causal powers in the setting of a group of social actors. Significantly, this is the approach taken by Charles Tilly and others within the field of the study of contentious politics ( McAdam, Tarrow, Tilly 2001). Suppose we are interested in France’s collapse in the Franco-Prussian War. We might expend significant research work on discerning the organizational and command structure of the French Army in the 1850s and

Meanings and Mechanisms  241 1860s. We might look in detail at Napoleon III’s state apparatus, including its international relations bureau. And we might gather information on the structure, capacity, and organization of the French rail system. Then we might offer an explanation of a number of events that occurred in 1870 as the result of the causal properties of those historically embodied organizations and institutions. The real performance properties of the rail system under a range of initial conditions can be worked out. The conditions presented by the rapid mobilization required by suddenly looming war can be investigated. The logistical collapse that ensued can be explained as the result of the specific causal properties of that complex system. There are no general laws of these phenomena; but it is straightforward to identify causal powers and mechanisms involved in their workings. (Michael Howard’s history of the Franco-Prussian War gives careful attention to these sorts of factors; Howard 1961.) The point here is a fundamental one. The covering law model depends on a metaphysics that gives primacy to laws of nature. The frameworks of critical realism and its cousins depend on a view of the world as consisting of things and processes with real causal powers. This intellectual framework is applicable to the social world as well as to the natural world. And it provides a strong intellectual basis for postulating and investigating social causal mechanisms. Any conception of causal powers requires that we have an idea of the nature of the substrate of causation in various areas. And the social metaphysics of actor-centered sociology provide a strong candidate for such a framework in the case of social causation.

Actor-Centered History The approach taken here to historical explanation can be described as “actor-centered” ( Little 2014). The basic idea is that social phenomena are constituted by the actions of individuals, oriented by their own subjectivities and mental frameworks. We need to ground our hypotheses about social and historical causation in theories of the pathways through which actors embody those causal processes. Actors in relation to each other constitute the “substrate” of social causation. Actors make up the microfoundations of social causes and processes. Actors constitute the causal necessity of social mechanisms. It is recognized, of course, that the subjectivity of the actor does not come full-blown into his or her mind at adulthood; rather, we recognize that individuals are “socialized”; their thought processes and mental frameworks are developed through myriad social relationships and institutions. So the actor is a socially constituted individual; that is, individuals acquire beliefs, norms, aversions, desires, and practices through the social environment in which they develop. This in turn implies that agency depends upon a previous iteration of institutions and norms through which current actors acquired their mental

242  Daniel Little frameworks. This is not offered as a strong version of social determinism, but as the much more reasonable and limited observation that individuals acquire major aspects of their mental lives through immersion in a range of social environments. If we take the approach to social explanation that demands that we understand how complex social processes and assemblages are generated by the actions and thoughts of individuals, then it is logical that we need to develop a theory of the actor. In chemistry we need to have a theory of the atom; and in the social world we need to have a theory of the dynamics and capacities of the conscious, deliberative actor. It is clear that human beings bring specific frameworks of thought, ideas, emotions, and valuations to their social lives, and these frameworks affect both how they interpret the social realities they confront and the ways that they respond to what they experience. Human beings have “frames” of cognition and valuation that guide their experiences and actions. The idea of a practical-mental frame is therefore a compelling one, and it should be a possible subject for empirical sociological investigation. It is worthwhile addressing this topic, because it would appear that we don’t yet have a particularly good vocabulary for formulating questions about agency. There is a range of approaches to the social sciences that fall under the umbrella of actor-centered theories. One important fissure among these theories is that between “thin” and “thick” theories of the actor— theories which provide less or more detail about the mental frameworks and beliefs of the actors being described. The extremes of the two types of theories range from pure rational choice theory to social psychology and ethnography. The two types of theories have complementary strengths and weaknesses. Thin theories, including especially rational choice theory and game theory, make use of a particularly sparse theory of the actor’s decision framework. This approach provides a basis for representing the motives and decisions of actors that can be readily incorporated into powerful techniques of simulation and calculation. Thick theories, including pragmatist and ethnomethodological theories, offer a basis for investigating particular social settings of action in detail, and they provide an in-depth basis for explaining and understanding the choices, judgments, and behavior of the individuals they study. Let us consider an example of historical research that depends heavily on a thick theory of the actor. Robert Darnton attempts to tease out some of the distinguishing elements of 17th-century French rural and urban culture—through folklore, through documented collective behavior, or through obscure documents authored by police inspectors and bourgeois observers. He is a “realist” about mentalités; and he recognizes as well the plasticity and variability of mentalités over time, space, and group. (“I do not believe there is such a thing as a typical peasant or a representative bourgeois”; Darnton 1984: 6.) And he is more interested in

Meanings and Mechanisms  243 the singular revealing incident than in the large structural narrative of change; he demonstrates that careful historical interpretation of a single puzzling event can result in greater illumination about a historical period than from more sweeping descriptions and narratives. Darnton does not accept the notion that “good” social history must be quantitative or generalizable. Rather, he sees the task of a cultural social historian as one of uncovering the threads of voice and action that permit us to reconstruct some dimensions of “French peasant worldview” and to see how startlingly different that worldview is from the modern view. Our distance from the French peasant is great—conceptually as well as materially. So the challenges of uncovering these features of agency and mentality based on very limited historical data are great. In the title essay of the volume Darnton goes into a single incident in detail: the autobiographical account of Nicolas Contat, a printer’s apprentice (later journeyman), in which Contat describes an episode of cat killing by the apprentices and journeymen in the shop. Darnton relates the incident to its cultural and social context—the symbolic role that cats had in festivals in the countryside, contemporary attitudes towards violence to animals, the sexual innuendo represented by killing the mistress’s cat, the changing material relations between master and worker in the 18th-century trades. Darnton offers a “thick description” of this incident, allowing the reader to come to a relatively full interpretation of the significance of the various elements of the story. At the same time, he sheds light on the background mentalité and social practices of workers and masters. So the essay is a paradigm of interpretative cultural history. Darnton’s work in this book is valuable for the philosophy of history in several ways. First, it exemplifies a different model of historical knowledge: not a series of events, not a cliometric analysis of society and class, but an interpretation of moments and mentalités in a fashion designed to shed light on the larger historical moment. It is an effort to make historical understanding “ethnographic”. Second, it possesses its own form of rigor and objectivity. The facts matter to the interpretations that Darnton offers—the facts of the multiple versions of folk stories, the facts of what we know about the changing circumstances in the printing trades, the facts of peasant hunger at several periods in the 17th and 18th centuries. Third, it has the potential for shedding deeper light on French popular action than we are likely to gain from a traditional “rational actor” or class-conflict approaches. The motives that Darnton discerns among the printers are sometimes goal-directed; but sometimes emotional, and sometimes related to the simple recklessness of young men in constraining circumstances. Finally, Darnton’s work here provides some specific insights into questions about the historical study of “mentalités”. Darnton shows that it is possible to make significant headway in the project of figuring out how distant and illiterate people thought about the world around them, the

244  Daniel Little social relations in which they found themselves, the natural world, and much else. The documents available to us in the archives have a richness that speaks to these ways of thinking the world; it is therefore a valuable task for the historian to engage in piecing together the details of daily life and experience that the documents reveal and conceal. Darnton’s task is an ethnographic one; but unlike Clifford Geertz, he does not have the option of immersing himself in ordinary life activities with these longdead French people. So the genius of his work in these essays is his ability to make use of historical artifacts and documents to tease out the underlying details of mentality that were likely in place among these individuals, and that help to make sense of their actions.

Meanings and Mechanisms It is plain that there are two large categories of factors that are fundamental to understanding social processes—meanings and mechanisms. Meanings are crucial because historical events are constituted by conscious human actors, whose actions have meaning to themselves and to observers. Mechanisms are crucial because we want to discover the causally relevant events that produced a given historical outcome, whether the success of the Chinese Communist Revolution or the meltdown of the Chernobyl nuclear reactor. Here I have given a preponderance of attention to the importance of social causal mechanisms within historical and social explanation. We explain a social outcome when we identify the social mechanisms that brought it about. It is crucial to bear in mind always, however, that there is a complementary dimension to social life and social process—the pervasive fact that people act within frames of meanings and interpretations that they bring to their social relationships and their social worlds. Human action is meaningful action, and we cannot make sense of action without attributing meanings, intentions, and frameworks of understanding and desire to the individuals who constitute a social encounter. What are meanings? The substrate to causal order in the social world is the fact of a population of socially constituted, socially situated actors whose beliefs and actions constitute social entities. Human beings are cognitive, affective, conscious, and intentional beings. They represent the world around them in terms of language, and they create meaningful representations for themselves of the actions and circumstances of other individuals. This is the fundamental ontology of the social world; it is the heart of the philosophical doctrine of ontological individualism ( Elster 1989, Hedström 2005). Actors are different from billiard balls and they are different from electrons or quarks. Actors are conscious, intentional, representational systems. They have the capability of forming beliefs and representations of the world around them, including their relations with other human beings. Their motivations are affected

Meanings and Mechanisms  245 by the norms, prejudices, and mental frameworks through which they interpret the world. Their actions are the result of their subjective mental systems of representation. This is the essential point that is emphasized by the hermeneutic tradition of the philosophy of society: Human action is meaningful, in the sense that human beings form representations and interpretations of the world in which they act. They operate on the basis of mental frameworks of representation and interpretation. It is an important component of social science and ordinary life to attempt to reconstruct the nature of the beliefs and representations that are held by individuals in various settings. This is what Robert Darnton attempts in The Great Cat Massacre. This is not a new insight, of course; it was fundamental to the hermeneutic approach to social life, including the influential thinking of Wilhelm Dilthey (1989, 1996), and it is the core of the philosophy of history expressed by Collingwood and Walsh. But the classical hermeneutic approach tended to under-value the importance of causation and mechanisms in the social world; whereas subsequent research in the philosophy of social science has made it clear that both mechanisms and meanings are inseparably embedded within the social world. It is in fact misleading to portray mechanisms and meanings as complementary “dimensions” of social change. Rather, we might say that social mechanisms depend upon meanings, for the simple reason that social mechanisms depend upon actions, and actions presuppose meanings. This is the thrust of the emphasis here on “actor-centered” approaches to sociology. The actor-centered perspective takes seriously the meanings, values, and cognitive and practical frameworks that individuals bring to their interactions in the social world, and it urges social scientists and historians to improve upon their current theories of the actor. Institutions and organizations are often invoked as causal factors or mechanisms in the production of important social outcomes. But institutions always work by influencing the behavior of the individual actors whom they touch; so either explicitly or implicitly we need to have a theory of the actor’s mental frameworks if we are to understand the causal power of institutions to influence outcomes. Think for a moment about how meanings and intentional actions give rise to a common social mechanism, hate-based nationalist mobilization in Western Europe and the United States. A few strident leaders formulate a message of hate against a group—currently, Syrian and North African immigrants in various European countries; they find means of gaining access to national media (through provocative demonstrations); and they extend their influence from the tiny percentage of racist extremists ex ante to a sizable percentage of the more moderate population. How does this work? Why do ordinary non-racist citizens fall prey to the hateful messages of the extreme right? Presumably a convincing answer will depend on the specifics of the communications strategies and messages

246  Daniel Little conveyed by the nationalist party, interlocking with an astute reading of the fears and suppressed prejudices of the majority population. In other words, the mechanism of racist mobilization depends on a substratum of political emotion and belief that can be adroitly manipulated by the racist group and its leaders. Philosophers sometimes distinguish meanings and causes as subjective and objective respectively. (This is implied in the fundamental distinction drawn by Georg Henrik von Wright [1971] between hermeneutic and causalist approaches to a domain.) But this is not a useful way of thinking about the two categories. Meanings are often fully objective—in the sense that we can investigate them empirically and they can be demonstrated to have stable and enduring effects in the world. And social causes have an element of subjectivity built into them, for the simple reason that social causes always invoke the subjective states of mind of the actors who make them up. It is not even accurate to say that meanings exist solely within the actor, whereas causes exist outside the actor. The meanings that Weber ( 1930) identified in the notion of the Protestant Ethic are indeed embodied in a population of individuals (inward); but they are pervasive and influential on those same individuals (outward). So the Protestant Ethic is both an inner state of mind and an external and coercive set of values and beliefs. Meanings and mechanisms intersect in a unique way. Consider a humdrum piece of urban sociology: the fact that African-American persons in the United States demonstrate higher rates of unemployment and lower levels of wages and wealth accumulation. Sociologists believe that several mechanisms lead to this outcome that is evident in population statistics. One particular mechanism is the role that implicit bias plays in hiring decisions, leading to differential rates of employment for black and white applicants. We may hypothesize that many hiring managers possess frameworks of assumptions about race, qualification, and work habits that discriminate between white and black applicants. This is a feature of meaning, widely distributed throughout the population. It becomes the causal power underlying many mechanisms of discrimination that exist—in employment, in healthcare, and in housing. Or in other words: The social mechanism of implicit racial discrimination, leading to marked gaps between white and black individuals in employment, health, and residence, is powered by a fact about widespread mental frameworks that convey ideas and assumptions about race that lead to differential behavior.

Conclusion The riddle of history is not so difficult after all. Historical events have causes, which may be structures, institutions, ideologies, normative systems, or social practices. And the social entities that occur in causal

Meanings and Mechanisms  247 statements in historical explanations have microfoundations at the level of historically situated actors. Individuals have mental frameworks, decision rules, cognitive systems, emotions, and memories; and these meaningful mental states are both socially conditioned and socially powerful. Individuals in concrete social contexts contribute to the creation of the causal powers possessed by the social arrangements they inhabit. Meanings and mechanisms make history.

References Bhaskar, R. (1975). A Realist Theory of Science. Leeds: Leeds Books. Bhaskar, R. (1989). The Possibility of Naturalism: A Philosophical Critique of the Human Sciences, 2nd ed. London: Harvester Wheatsheaf. Boyd, R. (1984). The Current Status of Scientific Realism. In: J. Leplin, ed., Scientific Realism, 1st ed. Berkeley: University of California Press, pp. 195–222. Brinton, M. and Nee, V. (1998). New Institutionalism in Sociology. New York: Russell Sage Foundation. Cartwright, N. (1983). How the Laws of Physics Lie. Oxford: Oxford University Press. Cartwright, N. (1999). The Dappled World: A Study of the Boundaries of Science. Cambridge: Cambridge University Press. Coase, R. (1988). The Firm, the Market, and the Law. Chicago: University of Chicago Press. Darnton, R. (1984). The Great Cat Massacre and Other Episodes in French Cultural History. New York: Basic Books. Dilthey, W. (1989). Introduction to the Human Sciences. Princeton: Princeton University Press. Dilthey, W. (1996). Hermeneutics and the Study of History. Princeton: Princeton University Press. Elster, J. (1989). Nuts and Bolts for the Social Sciences. Cambridge: Cambridge University Press. Harré, R. and Madden, E. (1975). Causal Powers: A Theory of Natural Necessity. Oxford: Basil Blackwell. Hedström, P. (2005). Dissecting the Social: On the Principles of Analytical Sociology. Cambridge: Cambridge University Press. Hempel, C. (1965). The Function of General Laws in History. In: C. Hempel, ed., Aspects of Scientific Explanation: And Other Essays in the Philosophy of Science, 1st ed. New York: Free Press, pp. 231–243. Howard, M. (1961). The Franco-Prussian War: The German Invasion of France. London: R. Hart-Davis. Little, D. (2006). Levels of the Social. In: S. Turner and M. Risjord, eds., Philosophy of Anthropology and Sociology, 1st ed. Amsterdam: North-Holland, pp. 343–371. Little, D. (2014). Actor-Centered Sociology and the New Pragmatism. In: J. Zahle and F. Collin, eds., Rethinking the Individualism-Holism Debate: Essays in the Philosophy of Social Science, 1st ed. Dordrecht: Springer, pp. 55–75. McAdam, D., Tarrow, S. and Tilly, C. (2001). Dynamics of Contention. New York: Cambridge University Press.

248  Daniel Little Merton, R. (1967). On Theoretical Sociology: Five Essays, Old and New. New York: Free Press. Thornton, P., Ocasio, W. and Lounsbury, M., eds. (2012). The Institutional Logics Perspective: A New Approach to Culture, Structure, and Process. Oxford: Oxford University Press. Von Wright, G. (1971). Explanation and Understanding. Ithaca: Cornell University Press. Walsh, W. (1968). Philosophy of History: An Introduction, 3rd ed. New York: Harper & Row. Weber, M. (1930). The Protestant Ethic and the Spirit of Capitalism. London: Allen & Unwin.

14 The Origins of Historiographic Causation Aviezer Tucker

Introduction Even a furtive philosophical reading of historiography cannot miss the causal propositions that are strewn ubiquitously throughout most historiographic texts. Historians write that events or processes caused others without giving much thought to conceptual or epistemic considerations and theories of causation. If historians become acquainted with the main philosophical theories of causation, regulatory, conditional, interventionist, counterfactual, Bayesian nets and so on, they are befuddled. They cannot recognize much of what philosophers write as describing what they think, practice, and write and they certainly do not think of themselves as working with regulatory generalizations or testing historiographic counterfactuals. Either the philosophical theories of causation are not quite applicable to historiography, or historians do not quite understand what they are doing and writing, or as I argue in this essay, both. Much of the debate in the philosophy of historiography about causation became trapped in 1940s philosophical amber, the covering law model, and equally fossilized alternative explanatory models. From the perspective of contemporary philosophy of causation and explanation, shooting down the covering law model and its alternatives is as easy as shooting down World War II spitfire and Messerschmitt warplanes with contemporary anti-aircraft missiles (Tucker 2004: 185–207). Why bother?! My purpose in this article is to move the discussion to the philosophical present and redirect it. I show that historiography is distinguished epistemically in its essential inference of descriptions of past events, processes, and causal relations. The epistemology of historiography is of inference rather than observation or intuition, memory or fiction. The past is lost, over and done with, and cannot be restored, reconstructed, or observed. Absolutely everything we know and can possibly know about the past, including causal connections between past events or processes must be inferred from information that the past transmitted to the present and was received. Statements of historiographic causation are derived from

250  Aviezer Tucker information in the present that was transmitted from past events, processes, and causes and effects. Let us call the entities in the past that transmitted information origins. Let us call the entities in the present that receive that information receivers. Origins transmit information to receivers. Information preserved in receivers may be used to infer its origins. Origin is an epistemic relational concept. As much as a cause can only be identified in relation to its effects and there are no causes without effects, origin can only be identified in relation to its receivers and there are no origins without receivers. Origins transmit encoded information signals to receivers. When you write an email, you send encoded electronic information to a server that receives it. In older days when the live performance of an orchestra was broadcast on the wireless, it sent encoded information via radio waves to receiving radios. Likewise, historical events and process send information signals to receivers that store them until historians decode them and call them historical sources. There are many different types of signals, transmission channels, and forms of encoding: Background radiation traveled from the origin of the universe to scientific instruments today. Species transmit information about their properties and ancestry via DNA through reproduction to descendant species. My writing this sentence now transmits information from my thoughts via electronic and linguistic encoding to the computer’s hard disk and thence to readers who receive and decode the information. During transmission, the information goes through a period of latency when it is not expressed. Latency can be very short or very long, from the age of the universe in the case of background radiation to the brief moment between the utterance of words by one person and their comprehension by another. Processes of encoding, information transmission, and decoding have levels of reliability or fidelity that measure the ratio of preserved information at the decoding end of the process of transmission to the information encoded and transmitted at its origin. Information signals are mixed with varying levels of noise and have different levels of equivocation, loss of signal. For example, ceteris paribus the transmission of information from a historical origin event to a decoding historian is more reliable, has a higher degree of fidelity, when it is encoded in a written form immediately after the event by an eye witness and preserved in an archive than when it is transmitted through oral traditions that inevitably mix signal with noise and have high rate of equivocation. The scope of possible inferences of origins from receivers depends on the ratio of preservation of information in receivers to entropy, disorder. Some information is lost during transmission (equivocation), and noise that does not carry information is mixed with the signal. For example, “prehistorical” societies caused through social evolution contemporary societies since we are all the descendants of “prehistoric” people and people have always lived in societies. But contemporary societies did not

The Origins of Historiographic Causation  251 preserve much information about their “pre-historic” origins. This prevents us from knowing much about “prehistory”, the period before the invention of writing, and what we know is from archaeological findings and genetic analysis, not from encoded information preserved by oral traditions or some other mean of social transmission of information. Information about interesting topics like the systems of beliefs and governance and the languages of humans for almost all the quarter million years or so life span of our species was lost due to equivocation and mixture with noise. Knowledge of history “depend[s] on the development of systems to record events and hence accumulate and transmit information about the past. No records, no history, so history is actually synonymous with the information age, since prehistory is that age in human development that precedes the availability of recording systems” (Floridi 2010: 3). Though societies are the effects of their pasts, they cannot transmit information reliably across centuries unless there is a written form of language that can preserve information reliably. The meaning of origins as a philosophical concept is linked to the contested meaning of information. In Shannon’s classical information theory, information means reduction in uncertainty. It measures the quantity of information received as proportional to the level of uncertainty prior to reception of the information, the prior probability of the information signal; the less certainty there was prior to receiving the signal, the less probable was the signal prior to being received, the more information is transmitted. Shannon codified his understanding of information in the Inverse Relation Probability Principle: The more improbable the signal, the more information it is transmitting. The more probable or possible a signal is, the less information it is carrying. For example, “the boy who cried wolf” was uninformative because of the very high prior probability that he would cry wolf (Floridi 2010: 42).1 The extensions of some origins and receivers and some causes and effects overlap. This extensional partial coincidence can obscure the conceptual or metaphysical and epistemic distinctions between origins and causes. It may account for the dearth of philosophic discussions of origins in comparison with causes. Clarifying what origins are, and how we come to know them requires a clear analysis of the conceptual and epistemic distinctions between origins and causes. There have been many conflicting theories of causation that disagreed about what causes are, how causes are inferred, and how causal propositions are justified. My purpose is not to argue for or against particular theories of causation, but to distinguish origins from causes and show that the inference of origins is the primary inferential pursuit of historians, while the inference of causes is secondary or derivative of it. For this purpose, I distinguish origins from different theoretical concepts of causes, to highlight the conceptual and epistemic distinctiveness of origins. I divide the comparison between origins and causes into metaphysical and epistemic parts that

252  Aviezer Tucker analyze respectively what origins are and how we come to know them vs. what causes are and how we come to know them. Finally, I show how the inference of origins can infer chains of token causes and effects. The philosophy of historiography may focus then on the primary inference of origins rather than on the derived inference of causes.

Metaphysical Conceptual Distinctions Transmission Origins transmit coded information to receivers. Extracting the information requires decoding. Between the transmission and the decoding, the information is latent, unexpressed. Historical events transmit information to documents that receive and store it. Transmission of information is unnecessary for causation. For example, the “perfect crime” is an effect that does not preserve information about its cause, the perpetrator. The perfect criminal causes the crime but s/he either does not transmit information about herself or by the time it reaches a receiver it has been subjected to so much equivocation (e.g. wiping out of finger prints) or mixed with so much noise that it cannot be decoded. Philosophers have debated whether there is simultaneous causation. There have been internally coherent theories of simultaneous causation since Kant (Rosenberg 1998). Simultaneous causation may be consistent with some of the main theories of causation ( Buzzoni 2014). If there is simultaneous causation, it distinguishes causes from origins because origins cannot be simultaneous with their receivers because information cannot travel faster than the speed of light. A  period of latency, when information is transmitted between encoding and decoding, must intervene between origins and receivers. Conditionals Some theories of causation reduce it to deterministic or stochastic, indicative or subjunctive conditionals. The relation between origins and receivers is also conditional. The difference is that causes may be conditions for the existence or occurrence of their effects, while origins are conditions for the information encoded in their receivers, for content rather than existence. When effects and receivers have the same extensions, the existence of effects-receivers may depend on causes, but the information they contain is conditional on their origins. For example, a decision of Thomas Babington Macaulay (later the Whig historian of England) caused the English language to become the official language of India. The origin of the language that became the official language of India is 19th-century British English. The discovery of the Rosetta Stone in 1799 by Bouchard caused Jean-François Champollion’s decoding of the

The Origins of Historiographic Causation  253 information it stored, but it was not its origin. The origin was the event when the ancient author composed it 2,000 years earlier. Simple reduction of causation to sufficient or necessary conditions has been recognized as too narrow and too wide: Too narrow because most causes we mention in everyday life or in historiography are neither sufficient nor necessary, and too wide because some sufficient and/or necessary conditions are not considered causes. For example, the presence of oxygen in the atmosphere has been necessary for all the events in human history, yet historians do not mention it as a cause. These considerations led to Mackie’s classical interpretation of causation by means of conditional statements: Causes are Insufficient but Non-redundant parts of Unnecessary but Sufficient conditions of an outcome (INUS for short) (Mackie 1974). In Mackie’s formulation, a cause is insufficient on its own to bring about the effect, but it is indispensable for the set of conditions that suffice to produce the outcome. This allows us to say, for example, that the economic depression of the thirties caused Nazism without saying that economic depressions invariably cause totalitarian regimes. Instead, the economic depression was a part of the set of conditions that sufficed for the rise of Nazism, and a part that could not have been replaced by another condition. Still, the INUS formula is too wide because it does not provide criteria for distinguishing conditions from causes. INUS conditions categorize as causes factors that historians consider only as conditions, non-redundant parts of unnecessary but sufficient conditions of their outcome, such as breathable atmosphere. Conditional theories of causation have had to propose pragmatic, interventionist, or objective and subjective anthropomorphic criteria for distinguishing causes from other conditions. Collingwood (1998) suggested that the concept of causation and the distinction between causes and conditions are anthropomorphic; reflecting practical human interests rather than an objective relation. Hart and Honoré (1985) argued that causes are distinct of conditions in being either “abnormal” in a context, or voluntary. William Dray (1980, 1989) argued that historians typically use value laden normative judgments to consider whose behavior was not normative, abnormal, condemnable, and therefore responsible as a cause rather than condition for historical events such as wars. Dray studied historiographic causal explanations of the American Civil War and the First World War to show how historians from opposing sides, through agreeing on the conditions, assigned different causes because they wanted to assign responsibility for the war to the other side. The criteria that distinguish causes from conditions affect the assumption of “background conditions”, or “comparison situations” that together with the causes bring about the effect. For example, when historians ask what caused Gutenberg’s invention of printing, they may assume as background conditions the conditions in contemporary 15thcentury Italy, or the conditions in the Holy Roman Empire during the

254  Aviezer Tucker 14th century and ask, respectively, which conditions were present in the Holy Roman Empire, but not Italy, and caused the invention of printing, and which conditions appeared for the first time during the 15th century to cause the invention of printing. Hitchcock (1996) argued that causal propositions include in addition to causes and effects also the assumption of effects that did not happen. Schaffer (2005) synthesized all the above to theorize that causal claims have four components, causes, effects, and their respective comparison classes. The choices of comparisons and alternatives are value laden. The distinction between origins and other conditions of decoded information in receivers is much simpler. The information in receivers is conditional on their origins. The existence of the receivers may be conditional on other conditions such as causes. Historical primary sources preserve information from their origins. They may be conditional on many other causes and conditions such as their discovery in the archive and the previous depositing and preserving of the documents there. Receivers may have multiple origins. For example, historians of ideas infer multiple origins of ideas in texts that preserved the ideational information they transmitted. It is possible then to distinguish more from less important origins according to the proportion of information in the receivers that originated with them. For example, Russell’s philosophy was more of an origin of the philosophy of Wittgenstein than NeoKantianism. The Cambridge position that Russell organized with Keynes for his protégé was a cause or condition of Wittgenstein’s philosophy, but not an origin. Counterfactual Theories Some of the information that origins transmit is often lost en route to the receivers. Unless they preserve, or not, one and only one bit of transmitted information, receivers preserve degrees of transmitted information. In other words, receivers are analog, the degree to which they preserve information signals sent by their origins is continuous and gradable. For example, the Lithuanian language preserves more information about its Indo-European origin than English. Ceteris paribus, diaries of leaders preserve more information about their motives, world views, and plans than minutes of their meetings. Unlike origins, causes do not “transmit” more or less of something gradable to their effects like Moliere’s virtus dormitiva (that parodically explains why opium causes sleep in The Imaginative Invalid). Causes usually either cause the occurrence or instantiation of the effects, or not. Ordinary causal statements such as “the assassination of Archduke Ferdinand in Sarajevo in 1914 caused the outbreak of the First World War”; or “Louis XVI’s failed attempt to escape caused his imprisonment and execution” do not allow for analog grading but for binary “digital” effects.

The Origins of Historiographic Causation  255 When effects take multiple values, it is in comparison to effects of the same type. It is not a measure of receiving something to different degrees from their causes. For example, the adverse demographic effects of the First World War were more severe in France than in the United States and Belarus suffered more destruction in the Second World War than any other country. These valuations result from comparisons between effects of the same type, percentage of dead young men and destruction of infrastructure and property respectively, not the result of different degrees of transfer of something gradable from the causes. There may be ceteris paribus correlations between measured values of causes and measured values of effects. For example, different degrees of valued monetary and fiscal economic policies, at least in theory, affect to different degrees the values of their economic effects. However, these valued degrees of effects measure correlations between changes in the causes and changes in the effects, not the transmission of something to the effects that preserve it to different degrees and can then infer it backward. For example, decline in inflation may be the effect of a central bank’s raising of interest rates; it may also be the effect of many other economic variables. The brute fact of declining inflation does not indicate by itself its cause let alone the degree to which it is affected by each cause or causes. By contrast, an article about the decision of a Central Bank to raise interest rates preserves to a measurable degree the information transmitted by the Central Bank. Articles in different newspapers may preserve that information or mix it with noise to different measurable degrees. These differences between “analog” receivers and mostly “digital” effects distinguish origins from causes in the framework of counterfactual theories. David Lewis’ (1973) theory of causation reduces it to counterfactual dependence: ceteris paribus if the cause event was the case, the distinct effect event would be the case, and vice-versa, had the cause event not been the case, nor would have the effect event. Later, Lewis revised his theory to accommodate “chancy”, probabilistic, causation ( 1986): Had the cause not occurred, the probability of the effect would have been lower than it was, and vice-versa. “Symmetrical” overdetermination of effects by causes that happen simultaneously, and preemption of causes that would have brought about the same effects by other causes that happened first, emerged as some of the main challenges to this theory. For example, Caesar’s death on the Ides of March was overdetermined by the assassins. Many discoveries and inventions were made by independent scientists and inventors. Had the first discoverer or inventor been eliminated from history, the second one would have caused the same type of historical effect. Yet, we say that the battle of Waterloo caused the ultimate defeat of Napoleon, even though had the Prussians come too late to win the battle, the weakness of Napoleon’s army after the retreat from Moscow implies that it would have been defeated soon thereafter.

256  Aviezer Tucker Lewis revised his theory in response to these challenges by stipulating that a causal chain must connect the cause with its effect, a stipulation that preempted causes cannot satisfy. A detailed, “fragile” description of effects can distinguish causes that brought about the fragile effect from those that would not. “Alternations” in the preempting cause influence the effects while alterations in the preempted cause would not (Lewis 2004). For example, James Watt invented-caused the Steam Engine though had he been eliminated from history, somebody else would have. Yet, to make an alternation, had Watt decided to paint his first engine deep purple, the first steam engine would have been deep purple. Defenders of the counterfactual theory of causation proposed to “preempt” the challenges of preemption by stipulating fine-grained and precise descriptions of effects (Paul 2000, Coady 2004). These criteria are too broad because they cannot distinguish causes from anything else that influences the properties of effects, like factors that delay them. For example, the Italian invasion of Greece affected the German invasion of the Soviet Union by delaying it, yet it was not a cause of Operation Barbarossa. Further, arguably, the “fragile” description of effects does not agree with entrenched pragmatic senses of causes and effects: We do not care what the unique pattern of stab wounds on Caesar’s body was, only that he was assassinated by members of the Roman elite in a futile attempt to save Republican Rome. Nor are we interested in the color of Watt’s steam engine, only in understanding what caused the invention of a machine that produced ten horsepower that were sufficient to kickstart the industrial revolution. When we explain these effects, we look respectively for political and military and technological and economic causes and effects. Since causation depends on counterfactuals according to Lewis, justifying and determining causal hypotheses requires confirming historical counterfactuals. Lewis suggested that when alternative causal counterfactual models compete, we choose the one that is most “similar” to the actual world or to its history. Such truth conditions, as many critics noted, are vague and their evaluation is intuitive. Competing counterfactuals may claim greater similarity to the actual world from different perspectives; since what we consider “similar” depends on contexts and pragmatic considerations. Consequently, the identification of causes according to Lewis’ criteria may be underdetermined. A counterfactual analysis of origins would not be as challenged by overdetermination, preemption, or truth conditions because analog receivers preserve information transmitted from their origins to varying degrees. The more information rich are the receivers, the more sensitive they are to their origins and the fewer origins can overdetermine them or preempt each other. To take an example where the extensions of cause and effect partly coincide with origins and receivers, consider the death of Caesar. As an effect, death is digital. A person can be either dead or

The Origins of Historiographic Causation  257 alive (“dying” means still alive, though not for long). Caesar was surrounded by his killers and few stabbed him at a time successively, so some assassins preempted others, and altogether they overdetermined his death. This challenges simple counterfactual accounts of causation. Yet, if Caesar’s body is a receiver of information about the origins of the stab wounds, rather than an effect in political history, each wound received information about the individual dagger and the angle and thrusting strength of the stabbing. The unique pattern of stab wounds would not be overdetermined or preempted. Likewise, genomes are rich in information about their ancestors and so are not usually overdetermined and their origins usually do not preempt each other. In the historical sciences like phylogeny, human history, and historical linguistics, it is not common for different and independent origins to transmit exactly the same rich information to receivers. The truth conditions of counterfactuals whose precedent is an origin are easier to determine than those whose precedent is a cause because it is easy to measure and compare degrees of information richness of actual and counterfactual receivers. Valued degrees of information can measure similarity between competing counterfactuals and the actual world precisely, explicitly, and unintuitively. For example, Kant and Sartre acknowledged Hume as an origin for their philosophy. But if we study Kant and Sartre’s texts as receivers of information, we find few traces of Hume in Sartre, while Kant’s philosophy is a response to the Humean challenge. Consequently, we can determine counterfactually that Hume was a major origin of Kant without whom there probably would have been no Kantian transcendental philosophy, but only a minor origin for Sartre’s philosophy, without whom Sartre could have written pretty much the same existentialist philosophy. Similarly, we can measure the degree of dependence of languages on their origins. Though there are some English origins of some Italian words and phrases, its main origin is Latin. Without English, Italian would have looked very similar to its present form, without Latin there could be no Italian. Markov Conditions Causation may be modeled as a Bayesian network that connects types of events (Pearl 2000). Nodes in such Bayesian networks are connected by causal-conditional, arrows. These arrows trickle “down”, from causes to effects but cannot form a “circle”, whereby effects loop back to affect causes. The distinction between “parent” causes and their “descendant” effects is unambiguous. Independence in a Bayesian network means the satisfaction of the Markov conditions: “A variable represented by a node in the Bayesian network is independent of all variables represented by its non-descendent nodes in the Bayesian network, conditional on all variables represented by its parents nodes” ( Bovens & Hartmann 2003: 69).

258  Aviezer Tucker Receivers are tokens. They often do not satisfy the screening conditions that the Markov conditions stipulate because receivers can and often do exchange information. To use the genealogical terminological analogies, receivers can be each other’s parents, descendants, and siblings in the absence of natural prohibition on informational incest. For example, species, even phylogenetically remote ones, have exchanged DNA, languages have exchanged vocabulary, and witnesses whether to history or crime may and sometimes do exchange information (Tucker 2016). When the extensions of effects and receivers coincide, causal “parents” may conditionalize the occurrence but not the information of the effectsreceivers “descendants” by determining the expression or suppression of the information, but not its content. For example, rules that regulate restricted access to documents in archives can cause the suppression or expression of documentary receivers of historical information, but they cannot determine their contents. Process Theories Process theories of causation, advocated in the philosophy of science by Wesley Salmon (1984, 1998), in the philosophy of the social sciences by Jon Elster (1989), and in the philosophy of historiography by Maurice Mandelbaum (1977), developed and revised Hume’s contiguity clause by considering causes and effects to be parts of a continuous process. A process is not assembled from causal chains; vice versa, causes and effects are abstracted from conceptually primary processes. By being derived from processes, causation can be free of regularities, conditionals, or counterfactuals. Philosophers debated the properties of the process from which causal chains are abstracted. Some reduced it to a mechanism, others compared it to a space-time “worm”, while still others distinguished it by the transmission of a “mark”. Identical information may be transmitted through different channels using different forms of encoding. Transmitted information is an emergent property in relation to mechanisms. For example, the same information may be transmitted orally face to face or on the phone, in sign language, on paper, electronically, or by drumming in Morse code, to mention but a few possible mechanisms. Information can be transmitted without a known mechanism. In Darwin’s evolutionary biology species transmit information to successive species without any mechanism. His theory was accepted by scientists long before scientists began to understand the genetic mechanism for transmission of information. The systematic similarities in grammatical structure and vocabulary among the Indo-European languages are too numerous, systematic, and reliable to have arisen independently without a common origin. Yet, it is unknown where, when, and how the languages split. It is impossible to determine whether there was a single cause, a proto-Indo-European language that

The Origins of Historiographic Causation  259 was spoken perhaps in Anatolia by the first farmers, perhaps by Aryans in Northern India, or perhaps on the steppe of Eurasia, or perhaps somewhere else; and if so, when. Alternatively, there could have been several unrelated languages spoken by peoples who came to reside next to each other and over centuries and millennia affected each other’s languages until they gradually converged before splitting. Polynesians grew and ate potatoes before the arrival of European explorers. The Polynesian word for potato resembles the word used by the native, pre-Columbian, languages of the Andes where the potato originated. The Polynesian would have to travel thousands of miles in open sea without a compass to reach South America and then, even more improbably, find their way back to the Polynesian islands in the Pacific, again without a compass in open sea, with their newly acquired potatoes and the word they had learned to refer to them. While the South American origin of the Polynesian word for potato is beyond doubt, there is no known process, mechanism, or causal chain that connected South American with Polynesian potatoes. The same considerations hold for the interpretation of a process of information transmission as a “worm” in space-time, representing the four-dimensional extension of the process that is not necessarily mechanical. It is often impossible to trace a continuous transmission of information from origins to receivers, and even when it is possible, it can be redundant. Sometimes, as in the case of the Indo-European languages and the Polynesian potato, there is insufficient evidence to infer how space-time worms from the origins of the Indo-European languages and the Polynesian potato could have possibly looked like. On other occasions, for pragmatic reasons, it does not matter. Decoded rich information of low prior probability from two or three independent and coherent receivers is sufficient for inference of knowledge, without considering the process of information transmission (Tucker 2016). When we receive coherent information from independent experts we do not investigate how they received the information, only if they truly cohere (Tucker 2014). Salmon (1984, 1998) suggested that genuine processes involve the transmission of a mark along the process. Much of Salmon’s efforts and the ensuing debates concentrated on distinguishing “genuine” processes that transmit “genuine” marks, from pseudo-processes that only seem to transmit marks, but string together similar events that seem to succeed each other but are not threaded together by genuine mark transmissions or causation. The classical examples were of correlated types of effects that preserve marks from a common type of cause but are not causally connected to each other. For example, the correlation between barometers and storms results from a common origin, change in barometric pressure. Token events as well may seem to transmit a mark but are not causally connected. For example, historical “waves” like democratization or monotheism may have different tokens of a common type of

260  Aviezer Tucker cause, or they may have no recent type of common cause either (Tucker 2015: 16–17). Origins transmit special types of “marks”, information signals, and their transmission does not have to be contiguous. Through ontologically simple, distinguishing signals from noise may be epistemically challenging. It requires separating information signals that converge on receivers and tracing them back to their respective origins to peel off noise from signal. Causal Regularities and Information Preservation Philosophers since David Hume have been concerned with distinguishing causes and their effects from mere sequences of events. Regularity theories of causation claim that the assertion that a token event caused another token event must be justified by a relevant type-type causal regularity. Metaphysical accounts of causation that stipulate a covering law or looser probabilistic or statistical regularity or just a local generalization cannot be tested. For example, Davidson suggested that causal token claims contain a commitment to the existence of some law without describing it. “According to Davidson, in order for Ducasse to claim that a c-change cause an e-change, there must be event-types C’ and C (and event-types E’ and E) such that the change of c from being C’ to being C caused the change of e from being E’ to E, that is, there must be a c-like event (meaning an event-type of the form: has changed from C’ to C) and an e-like event (meaning an event-type of the form: has changed from E’ to E) such that c-like events are followed by e-like events” (Psillos 2009: 147). There is no obvious method of refuting Davidson’s causal theory if it is wrong. Almost no causal propositions in historiography are linked with an explicit or easily extrapolated law. But a Davidsonian account can explain this away by claiming that the alleged implied laws or regularities are unknown, but implicitly present. Woodward distinguished causal explanations of types and tokens, e.g. of the sliding of a block down a plane (type) and the extinction of the dinosaurs (token). Explanations of tokens2 are not deduced from laws: the Deductive Nomological view of the relationship between ordinary and scientific explanation gets matters exactly backwards. All human cultures have produced causal explanations, but the notion of a deductively valid argument and the notion of a law of nature are complex and sophisticated products of a very specific intellectual and scientific tradition. Rather than trying to understand all varieties of causal explanation in terms of these specialized categories, we should instead begin with a more general notion causal explanation, understood in manipulationist terms, and then attempt to

The Origins of Historiographic Causation  261 understand explanations that appeal to explicit chains of deductive reasoning and laws of nature as one specific variety within this genus. (Woodward 2003: 20) Woodward’s alternative to the deductive-nomological model, “invariant generalization”, is also between types ( 2003: 17f.). It is “necessary to appeal to claims about type causal relationships [. . .] to elucidate tokencausal claims. [.  .  .] token or singular causal claims always should be understood as committing us to the truth of some type-level causal generalization” (Woodward 2003: 72). Irrespective of the type of regularity involved in causation, many philosophers find it difficult to imagine causation without a type-type regularity of some kind in the background. As Woodward put it: I work [. . .] within a quite restricted framework, in which information about deterministic type-causal relationships is assumed to be part of our background knowledge, and the only question is what these type-causal relationships and other background information imply about token-causal relationships. (Woodward 2003: 75) Woodward was right to note that the deductive-nomological model as an explication of what people mean or imply when they utter causal statements is highly unlikely prior to the introduction of the concept of exceptionless law of nature in Europe in the scientific revolution of the 17th century (cf. Tucker 2005). But the same culturally specific charge can be leveled at Woodward’s own manipulationist regularity view though it is more general and fits a broader cultural scope. Causal explanations in the historical sciences and parts of sciences like cosmology and geology that cannot be manipulated are standard counterexamples to the manipulationist account (for rebuttals of this argument see Buzzoni 2014). Still, as long as the regulatory account of causation in its manipulationist guise is metaphysical, there can be no obvious way to test it. Information transmissions from token origins to token receivers need not imply or assume type-type background regularities. The preservation of information in receivers renders regularities redundant. When it is certain that the information could have been transmitted from nowhere else but the origins, the regularity that may or may not govern the transmission of information is redundant. For example, the Polynesian potato and the word for it could have only come from South America. The origins of DNA sequences that are found in genomes of phylogenetically remote species could have only come from the genomes of the origin species. Regularities are redundant even if possibly knowable. These information transmissions do not break any known laws of nature, but they are not covered by any either.

262  Aviezer Tucker Type-type general theories or regularities about the transmission of information in time may connect origins with receivers, but they are never sufficient and only sometimes necessary for inferences of origins. Type-type regularities about information transmission are assumed when scientific theories are necessary for decoding information embedded in token receivers. For example, decoding the information embedded in background radiation and tracing it back to the Big Bang requires the assumption of type-type regularities; so is the decoding of genomes to infer their phylogenic origins. But in historiography, decoding information, mostly in documents, requires knowledge of languages and historical conceptual frameworks and some particular background knowledge about norms of information preservation or destruction, e.g. which documents the archival rules considered important enough to be preserved and which were discarded, but nothing like scientific regularities. The Epistemologies of Origins and Causes The epistemologies of causation and origins do not have to depend on metaphysical theories. It is possible to bracket off the metaphysics of causation or treat the concept as a primitive, irreducible, and undefined fundamental concept that can neither be eliminated as “stone age ontology” nor reduced to something else (Miller 1987). The epistemic project is not to analyze the concepts of causes and origins as much as to elucidate the best practices of inferring them. I show next how the distinction between origins and causes leads to different epistemologies and methods of inference. The inference of origins commences with decoding information nested in receivers. Some decoding requires no theoretical background, for example, when we decode the information transmitted via texts, testimonies, or voluntary and involuntary physical gestures. Scientific decoding requires a theoretical background, sometimes simple and commonsensical, as when linguists inferred the origins of language families on the basis of rudimentary theories of language that assumed that grammar changes more slowly than vocabulary and that some parts of the vocabulary change more slowly than others. They discovered that proper names of places, words that refer to fauna and flora, and words that mean immediate family members and body parts are more reliable, preserve more information about their origins, than other parts of language. These simple generalizations allowed philologists to infer common origins of languages from correlations between information preserving words in the more reliable sub-group of their vocabularies (Tucker 2004: 46–91). Other type-type information transmission theories are fully embedded in advanced science, for example, the theories that allowed scientists to decode the information in background radiation to infer the origins of the universe.

The Origins of Historiographic Causation  263 By contrast, inferences of causes from their effects do not require decoding because causes do not necessarily transmit encoded information to effects. If causes are necessary conditions of the effects, it is possible to infer the causes from the existence of the effects. But that does not involve an act of decoding any more than causation entails encoding. If causes and effects are contiguous in one of the senses discussed above, knowledge of the effects and the contiguous process that connected them with their causes may be necessary and sufficient for the inference of causes. For example, when our ancestors hunted large game they could see how the spears they threw flew through the air and pierced the animal and how it bled and died. They could then naturally conclude that their throwing of spears killed the animal, just as farming must have begun when our ancestors observed how seeds developed into mature plants and concluded that the seed is the cause of the mature plant. By contrast, the inference of origins from receivers does not require knowledge of or descriptions of contiguity. As I noted above, it is possible to infer the common origins of the Indo-European languages and the South American origin of the Polynesian potato without a known contiguous process of information transmission from origins to receivers. Sometimes, the inference of origins involves the tracing back of the information genealogies of testimonies, texts, and so on. But gaps in the evidence for the genealogy do not preclude the inference of origins. Theories of causation that assume that regularities mediate between types of causes and types of effects, may interpret the inference of tokens of causes from to be retrodiction whose premises are descriptions of the token effects and type-type regularities. Hempel (1965) thought such retrodiction may be the foundation of knowledge of history. Some type-type regularities are useful for the decoding of receivers and inference of their origins, as noted above. These regularities are exclusively about the transmission of information in time, not the laws of human nature or of social evolution. Still, information regularities between types of origins and types of receivers are not necessary for all inferences of origins from receivers. Origins and receivers are tokens that occur at definite times and places. It is possible to infer properties and descriptions of origins from the preserved information they transmit to receivers. Therefore, the inference of origins from receivers does not have to rely on knowledge of regularities that govern the relations between types of origins and types of receivers. The more information is preserved in the receivers, the more knowledge can be inferred about their origins. When receivers do not preserve sufficient information to determine one and only one origin, they infer a range of possible origins and a probability distribution. For example, if a witness describes a perpetrator, the more properties the witness recalls, the more information rich is the testimony, and the narrower would be the range of possible origins-offenders. When a single origin can be

264  Aviezer Tucker determined, the more information the receivers preserve, the more can be known about it. The more informative is a historical primary source, the more we may infer about the historical events that transmitted the information to the source. There is a simple, linear, relation between the information richness of receivers and the level of informative detail of the descriptions of origins they infer. Regulatory theories of causation entail the opposite conclusion: The more information rich are the descriptions of token effects, the less determined are the retrodictive inferences of their token causes, because the more difficult it becomes to justify the relevant necessary covering regularities. Ceteris paribus the scope of evidence that can be used to justify regularities narrows the more information rich are the descriptions of the types of effects. For empirically justified regularity to cover types of cause-effect sequences, there must be a sufficient scope of empirical cases of token causes and effects of the same type. Arguably, precise, “fine grain”, information rich, descriptions of causes and effects are in inverse ratio to their similarity to other events. The more fine-grained is the description, the more difficult it is to justify type-type regularities about it. Vice versa, the more imprecise the description, the less information it carries, but the easier it is to fit it to a type-type regularity and support a casual proposition or explanation. The golden mean is repetition without exception (Psillos 2009: 144–148). In the social sciences, including Historical Social Science, however, there are tradeoffs between accuracy and scope (Tucker 2004: 151–167). When causes conditionally overdetermine effects and may preempt each other, in the absence of further evidence, the effects underdetermine their causes. Ceteris paribus the more fine-grained is the description of the effect, the fewer possible causes it could have and the less underdetermined would be their inference. However, some causal statements cannot be made “fine-grained” without a cost in underdetermining the counterfactual or regulatory justification of the causal inference. As just noted above the more fine-grained is the description of effects, the more difficult it is to justify or determine the regularity that should describe the conditional relations between causes and effects. When receivers are information rich, the inference of origins from receivers can calibrate the degree of information richness, detail and “fine-graininess”, of the description of the receivers for the purpose of determining their origins. Ceteris paribus, there is positive correlation between the information richness of receivers and the determination of their origins. The ceteris paribus positive correlation between the information richness of receivers and the determination of their origins has an exception when the receiver is a testimony or a set of testimonies that are not independent of each other. The human proclivity and capability to deceive, to mix signals with noise or just broadcast noise, preempts the epistemic advantages of information rich receivers that hold

The Origins of Historiographic Causation  265 ceteris paribus for involuntary types of receivers like genome sequences, languages, and background radiation. Liars and confidence artists are particularly keen on spinning long and complex, information-rich yarns made up of much noise. Only coherent, information-rich testimonies from multiple independent witnesses can and does indeed determine testimonial origins, thus forming the basis for social knowledge (Tucker 2016). Inferences of Historical Causal Chains From Inferences of Origins Origins are not just single events that transmit information to receivers. Origins can also be processes and causal chains that send encoded information to receivers in the present. For example, the expansion of the universe is a process that has been taking place and possibly accelerating for billions of years and is the origin of red shifted signals that reach the present. Receivers can infer origins that are processes and chains of causes and effects. Descriptions of such origins may form the basis for historiography. If a leader wrote in a reliable private diary as a receiver of information that he intended to do so and so for such and such particular reason, the diary preserves the information about the historical causal link between motive and action, cause and effect. Covering regularities, counterfactuals, rational choice models, and emphatic understanding, are all redundant for supporting such historiographic assertions of cause and effect, even if available, and justified. Historical processes like the French Revolution are the origins of numerous information signals that were received and stored by receivers that historians call historical sources and decode to infer descriptions of historical processes. The French Revolution caused the end of the French Monarchy and then the Jacobin Reign of Terror. To justify this statement by using a regularity, historians would have had to consider it a token of the type Revolution or a less general type of revolution and then find or invent a relevant social science regularity. To justify this statement by using a counterfactual, historians would have had to imagine a world without a French Revolution and see if the French monarchy would have ended anyway, a fairly difficult counterfactual to evaluate because of its inconsistency with its ceteris paribus conditions since too many things would have had to change in history to allow the French Revolution not to happen (Tucker 2004: 227–239). However, the French Revolution, the end of the monarchy, and the consequent Reign of Terror as a single causal sequence is the origin of documents that recorded the events that followed July 14, 1789. When decoded, these information-preserving receivers of the information transmitted by the chain of causes and effects between 1789 and 1793, can infer and justify representations of that causal sequence, explain how and why the Revolution radicalized and led to the Jacobin dictatorship and the rule of terror. All this can be inferred without resorting

266  Aviezer Tucker to regularities that connect social types of cause with types of effects or counterfactuals that connect token historical causes and effects. Whether or not there are discoverable regularities about the causes and effects of revolutions, and whether or not historians or social scientists can and do discover and possess them, they are redundant for the inference of historical chains of causes and effects that compose the French Revolution because that causal chain is also the origin of receivers in the present that preserve the information about that historical process. Social sciences can use historiographic descriptions of historical processes or sequences of historical token causes and effects to inspire and illustrate if not quite confirm social theories about social types (Tucker 2012). In sum, the debates about causation in the philosophy of historiography focused on an epistemic derivative, rather than on the basic inference of origins from receivers, an inference that distinguishes the historical sciences from theoretical or experimental sciences (cf. Tucker 2012). The extent to which historians can and do offer causal explanations of events, or for that matter rational explanations of action or understanding of historical minds, depends entirely on whether information from the past reaches receivers in the present that then can be decoded to infer their origins, whether they are causal chains, or rational actions and decision making, or the minds of past agents. Models of historiographic explanation, causation, understanding, rational choice, and so on are derivative and not primary, nor are they mutually exclusive alternatives that historians choose between voluntarily or on the basis of their values, as Rickert (1962) would assert. They are not mutually exclusive since rational choice and mental states can be causes and rationality can be a mental state. They are derived and depend just like propositions about historical causality entirely on the information that reached the present without equivocation and can be separated from noise, so historians can decode it to infer its origins. The limitations on historiographic explanations, whether causal or not, reflect the extent to which information that historical events processes and causal chains transmitted to receivers was preserved rather than subjected to equivocation and mixture with noise, and the ability of historians to decode that information to infer and determine its origins.

Notes 1 An alternative philosophical conceptualization of information is semantic (Dretske 2008). It stipulates that information must be true. This concept of information is too narrow for the historical sciences. Signals that satisfy Shannon’s criteria for information may have no propositional content that can be true or not, or they may have a content that is manifestly false. They may nevertheless transmit signals about their origin. For example, the historiography of ideas traces the transmission of ideas that were false from their origins to receivers, for example ancient cosmologies or Aristotelian notions of nature. These ideas have negative truth values, still they can be used to infer their origins. Comparative Historical Linguistics traces back the origins of families

The Origins of Historiographic Causation  267 of languages from the information that descendant languages received from them. Even intentionally deceptive propositional content may transmit valuable information about its origins. For example, historical forgeries attempt to mislead their readers about their provenance but may preserve information about their actual historical origins. The “Donation of Constantine” is a forged document that does not contain information from the era of Constantine the Great but preserves information about its later origins in the second half of the 8th century. Historians are not concerned exclusively or even primarily with the veracity of the propositional contents of historical sources. Their first question after decoding the signal is where does this signal come from, what is its provenance or origin? 2 Token effects are sometimes referred to in the literature alternatively as “singular” or “immanent”. I treat these terms as synonymous.

References Bovens, L. and Hartmann, S. (2003). Bayesian Epistemology. Oxford: Oxford University Press. Buzzoni, M. (2014). The Agency Theory of Causality, Anthropomorphism, and Simultaneity. International Studies in the Philosophy of Science, 28(4), pp. 375–395. Coady, D. (2004). Preempting Preemption. In: J. Collins, N. Hall and L. Paul, eds., Causation and Counterfactuals, 1st ed. Cambridge: MIT Press, pp. 325–340. Collingwood, R. (1998). An Essay on Metaphysics, 2nd ed. Oxford: Clarendon Press. Dray, W. (1980). Perspectives on History. London: Routledge. Dray, W. (1989). On History and Philosophers of History. Leiden: Brill. Dretske, F. (2008). Epistemology and Information. In: P. Adriaans and J. Van Benthem, eds., Philosophy of Information. Handbook of the Philosophy of Science. Vol. 8, 1st ed. Amsterdam: Elsevier, pp. 29–47. Elster, J. (1989). Nuts and Bolts for the Social Sciences. Cambridge: Cambridge University Press. Floridi, L. (2010). Information: A Very Short Introduction. Oxford: Oxford University Press. Hart, H. and Honoré, T. (1985). Causation in the Law, 2nd ed. Oxford: Clarendon Press. Hempel, G. (1965). Aspects of Scientific Explanation. New York: Free Press. Hitchcock, C. (1996). The Role of Contrast in Causal and Explanatory Claims. Synthese, 107(3), pp. 395–419. Lewis, D. (1973). Counterfactuals. Oxford: Blackwell. Lewis, D. (1986). Philosophical Papers. Vol. II. New York: Oxford University Press. Lewis, D. (2004). Causation as Influence. In: J. Collins, E. Hall and L. Paul, eds., Causation and Counterfactuals, 1st ed. Cambridge: MIT Press, pp. 75–106. Mackie, J. (1974). The Cement of the Universe. Oxford: Clarendon Press. Mandelbaum, M. (1977). The Anatomy of Historical Knowledge. Baltimore: Johns Hopkins University Press. Miller, R. (1987). Fact and Method: Explanation, Confirmation and Reality in the Natural and Social Sciences. Princeton: Princeton University Press. Paul, L. (2000). Aspect Causation. Journal of Philosophy, 97(4), pp. 223–234.

268  Aviezer Tucker Pearl, J. (2000). Causality: Models, Reasoning, and Inference, 2nd ed. Cambridge: Cambridge University Press. Psillos, S. (2009). Causation and Explanation. London: Routledge. Rickert, H. (1962). Science and History: A Critique of Positivist Epistemology. Princeton: D. Van Norstand. Rosenberg, J. (1998). Kant and the Problem of Simultaneous Causation. International Journal of Philosophical Studies, 6(2), pp. 167–188. Salmon, W. (1984). Scientific Explanation and the Causal Structure of the World. Princeton: Princeton University Press. Salmon, W. (1998). Causality and Explanation. New York: Oxford University Press. Schaffer, J. (2005). Contrastive Causation. Philosophical Review, 114(3), 327–358. Tucker, A. (2004). Our Knowledge of the Past: A Philosophy of Historiography. New York: Cambridge University Press. Tucker, A. (2005). Miracles, Historical Testimonies, and Evidence. History and Theory, 44(3), pp. 273–290. Tucker, A. (2012). Sciences of Tokens and Types: The Difference between History and the Social Sciences. In: H. Kincaid, ed., The Oxford Handbook of Philosophy of the Social Sciences, 1st ed. Oxford: Oxford University Press, pp. 274–297. Tucker, A. (2014). Epistemology as a Social Science: Applying the NeumanRubin Method to Explain Expert Beliefs. In: C. Martini and M. Boumans, eds., Experts and Consensus in Social Science, 1st ed. Dordrecht: Springer, pp. 155–170. Tucker, A. (2015). The Legacies of Totalitarianism: A Theoretical Framework. New York: Cambridge University Press. Tucker, A. (2016). The Generation of Knowledge from Multiple Testimonies. Social Epistemology, 30(3), pp. 251–272. Woodward, J. (2003). Making Things Happen: A Theory of Causal Explanation. New York: Oxford University Press.

Author Index

Anscombe, G. Elizabeth M. 4, 13, 122, 130, 154, 156, 222 Collingwood, Robin G. 4, 6, 10, 194 – 199, 204, 232n9, 238 Dancy, Jonathan 110, 127 – 133, 140, 150, 157 Davidson, Donald 15, 18, 20, 46, 53, 61, 63, 66, 74, 79, 82, 86, 97 – 101, 103, 109, 112 – 115, 117, 122, 126, 130, 132, 141, 151, 205, 225, 260 Dray, William 4, 12 – 15, 21, 33, 151, 167, 195, 199, 202 – 207, 253 Hacker, Peter M. S. 19, 147, 122n3, 232n5 Hempel, Carl Gustav 12, 154, 167, 194, 198, 202, 207, 209, 215, 239, 263

Hume, David 7, 14, 80, 92, 112, 125, 132, 141, 155, 157, 168 – 171, 221, 240, 257, 260 Hyman, John 112 – 116, 118, 121, 130, 151 Mele, Alfred 3, 19, 46 – 57, 61 – 67, 100, 104 Neurath, Otto 11 Smith, Michael 127, 150, 156 Stoutland, Frederick R. 16, 225 von Wright, Georg Henrik 15, 18, 130, 168, 212n17, 217, 223, 229 Winch, Peter 11, 15, 17, 110n2 Wittgenstein, Ludwig 4, 13, 19, 112, 120, 153, 212n15, 218, 226

Subject Index

action: action vs. behavior (see behavior); basic vs. non-basic 75n2, 223, 226, 229; collective (joint) actions 4 – 6, 8, 24, 78, 172, 187 – 191, 229; intentional action 2, 45, 52, 59, 61, 65, 79, 116, 171, 218, 223, 225; omissions 217; subsidary actions 54, 219 behavior: in contrast to action 2, 15, 49, 61, 63, 69, 79 – 84, 179, 181, 217 belief-desire-model see practical syllogism causation: Collingwood on 197; counterfactual theory of 22, 168 – 171, 174, 176, 235, 254 – 257; as making-a-difference 132; regularity theory of 12, 16, 18, 22, 112, 132, 167 – 169, 260, 264 Davidson’s challenge 18, 46, 63, 73, 79, 101 – 107, 228 desire 16, 20, 46, 66 – 74, 87, 89, 115 – 118, 172, 182, 219, 225 deviant causal chains 18, 52 – 57, 75n3, 112 – 114, 220 disposition 13, 91, 116 – 120, 136, 140, 144n20, 201 – 204, 206, 221, 237 explanation 87, 108, 167, 179; deductive-nomological (see Hempel, Carl Gustav); dispositional explanation cf. disposition; historical 4, 12, 21, 82 – 85, 141, 150, 167, 175, 194, 198, 201, 228 – 232, 234, 241 – 246

first-person-authority 20, 87, 104, 120, 222 intentions 70, 73, 116, 172, 222 – 227; as dispositions 116 introspection see first-personauthority Logical Connection Argument 14 – 17, 69, 117, 130, 153, 221 – 227 mechanism 8, 21, 173, 190, 213n7, 239, 244, 258 practical knowledge 14, 222 practical syllogism 14, 46, 74, 99 – 102, 135, 150, 171, 184, 192n7, 202, 216, 226 rationalizing principles see Davidson’s challenge reasons: apparent 128, 135, 140, 144n17, 154 – 158; explanatory 97, 100, 104, 109, 126, 143n7, 146, 149; future 134, 138, 141, 199; as mental states 7, 20, 46, 66 – 72, 74, 87, 126 – 130, 150, 171, 218 – 221, 266; motivating 126, 146, 149, 153, 156; as neurally realized 3, 46, 68, 79, 86, 219; normativity of 3, 98, 101, 108, 121, 126, 141, 147, 151, 153, 156, 160n15, 226; objectivism about 127, 130, 154; psychologism about (see reasons, as mental states); real (“effective”) reasons (see Davidson’s challenge) self-deception 11, 93, 181 subconscious see self-deception