From individual to collective intentionality: new essays 9780199936502, 0199936501

Citation preview

From Individual to Collective Intentionality

From Individual to Collective Intentionality New Essays

Edited by Sara Rachel Chant, Frank Hindriks , and

Gerhard Preyer

1

1 Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford New York Auckland  Cape Town  Dar es Salaam  Hong Kong  Karachi Kuala Lumpur Madrid Melbourne Mexico City Nairobi New Delhi Shanghai Taipei Toronto With offices in Argentina Austria Brazil Chile Czech Republic France Greece Guatemala Hungary Italy Japan Poland Portugal Singapore South Korea Switzerland Thailand Turkey Ukraine Vietnam Oxford is a registered trademark of Oxford University Press in the UK and certain other countries. Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016

© Oxford University Press 2014 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by license, or under terms agreed with the appropriate reproduction rights organization. Inquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above. You must not circulate this work in any other form and you must impose this same condition on any acquirer. Library of Congress Cataloging-in-Publication Data From individual to collective intentionality : new essays : / edited by Sara Rachel Chant, Frank Hindriks, and Gerhard Preyer.   pages cm ISBN 978–0–19–993650–2 (hardcover : alk. paper)  1.  Intentionality (Philosophy)  2.  Intentionality (Philosophy)—Social aspects.  I.  Chant, Sara Rachel, editor of compilation. B105.I56F76 2014 128’.2—dc23 2013031120

1 3 5 7 9 8 6 4 2 Printed in the United States of America on acid-free paper

CONTENTS

Contributors  vii Introduction  1 Beyond the Big Four and the Big Five  Sara Rachel Chant, Frank Hindriks, and Gerhard Preyer PART ONE  Collective Attitudes and Actions 1. A  Dynamic Theory of Shared Intention and the Phenomenology of Joint Action  13 Deborah Tollefsen 2. Collective Goals Analyzed  34 Kaarlo Miller and Raimo Tuomela 3. Group Belief and Acceptance  61 Frederick F. Schmitt 4. Against Group Cognitive States  97 Robert D. Rupert 5. The Ontology of Collective Action  112 Kirk Ludwig PART TWO  Collective Rationality 6. Indispensability, the Discursive Dilemma, and Groups with Minds of Their Own  137 Abraham Sesshu Roth 7. Do Groups Have Scientific Knowledge?  163 Melinda Bonnie Fagan 8. Collective Rationality’s Roots  187 Paul Weirich 9. Structural Rationality and Collective Intentions  207 Julian Nida-Rümelin Index  223

Contributors

Melinda Bonnie Fagan, Rice University Sara Rachel Chant, University of Missouri Frank Hindriks, University of Groningen Kaarlo Miller, University of Helsinki Kirk Ludwig, Indiana University–Bloomington Julian Nida-Rumelin, Ludwig Maximilian University of Munich Gerhard Preyer, Goethe-University Frankfurt am Main Abe Roth, Ohio State University Robert D. Rupert, University of Colorado–Boulder and University of Edinburgh Frederick F. Schmitt, Indiana University–Bloomington Deborah Tollefsen, University of Memphis Raimo Tuomela, University of Helsinki Paul Weirich, University of Missouri–Columbia

From Individual to Collective Intentionality

Introduction Beyond the Big Four and the Big Five S a r a Rac h e l C h a nt, F r a n k H i n d r i k s , a n d G e r h a r d   P r ey e r

Many of the things we do, we do together with other people. Think of attending a meeting, carpooling, and playing tennis. So how does doing something together differ from doing something on your own? Presumably, this question cannot be answered in terms of external behavior only. There need not be any outward difference between my raising an arm and my arm’s rising. Similarly, there need not be any behavioral difference between two people who happen to walk adjacent to each other and the same people going for a walk together. Most answers to the question of how my raising my arm differ from my arm’s rising refer to the intention involved in the former but not in the latter. In response to the question about how going for a walk together differs from merely walking adjacent to each other, an answer that has become increasingly popular in the past two to three decades also refers to an intention: the collective intention to go for a walk together. Collective Intentionality: The Big Four What is a collective intention? Are there any collective intentions? If so, are there any other collective attitudes, such as beliefs and desires? Questions such as these are answered by a theory of collective intentionality. The four most influential theories of collective intentionality are those of Michael Bratman, Raimo Tuomela, John Searle, and Margaret Gilbert. We refer to these theories

2

I ntroduction

as well as to their protagonists as “the Big Four of collective intentionality.” As each of them features in some of the contributions below, we start by briefly introducing the Big Four. Bratman (1992) provides an account of what he calls “shared cooperative activity” that is individualist: it invokes only individual attitudes and the relations between them. A shared cooperative action is distinct from mere uncoordinated actions of unrelated individuals in that the individuals who engage in it exhibit attitudes that bear particular relations to each other. Those relations are not restricted to epistemic relations, such as common knowledge. They extend to semantic relations: the relevant intentions refer to a “we”: the canonical specification of a shared intention, on Bratman’s view, is “I intend that we J.” And they include causal relations: insofar as shared actions are concerned, an individual possesses an intention due to the fact that the relevant other individuals do so as well. Tuomela (1984, 2002) argues that what he calls “joint intentions” comprise sets of “we-intentions,” which are the participatory intentions held by the members of a group (see also Tuomela and Miller 1988). A “we-intention” to perform a joint action involves the intention to perform that action, a belief that a sufficient number of other individuals will participate in its performance, and a belief that others believe that there is an opportunity to perform the action. This set of beliefs at least in part explains why each individual intends to participate in the joint action. Tuomela has developed accounts of stronger and weaker varieties of collective attitudes. I-mode collective intentionality requires only private commitments. The more holist we-mode collective intentionality involves an irreducible collective commitment. The notion of a joint commitment forms the cornerstone of Gilbert’s (1989, 1996) theory of collective attitudes. On her view, a joint commitment cannot be reduced to individual attitudes. A joint commitment exists when two or more individuals have openly expressed their willingness to be jointly committed as a body. Gilbert uses the phrase “as a body” to signal that a whole is formed in the process that cannot be reduced to the individual members. She also draws attention to this by referring to the parties to a joint commitment as “a plural subject.” A joint commitment entails a social (i.e. nonmoral) obligation to uphold the relevant collective attitude. Joint intentions come with a right of each party to rebuke anyone who fails to do her part of the collective action. Thus, Gilbert regards collective attitudes as intrinsically normative. Searle (1990, 1995, 2010) resists any reduction of collective attitudes to individual attitudes. He maintains that no set of individual attitudes adds up to a collective attitude no matter how interdependent they are. On this view, a collective intention is a mental state that is possessed by an individual in a collective mode due to which it is irreducibly collective. Individuals who participate

Beyond the Big Four and the Big Five

3

in a particular collective action derive their individual intentions from the collective intention. Searle’s (1990, 415) account used to be solipsist in that a single individual could have a collective intention without bearing any particular relations to other individuals. Now he insists on each individual having beliefs about the derived intentions of the other participants (Searle 2010, 55–56). Each of these theories is discussed in one or more of the chapters in this book. See Deborah Tollefsen (2004) for a more detailed overview of the Big Four. David Schweikard and Bernhard Schmid (2013) discuss the notion of collective intentionality both from a historic and systematic perspective. They also comment on the uses to which the notion is put insofar as collective reasoning, collective responsibility, social institutions, and human sociality are concerned. Almost all contributors draw on the work of at least one of the Big Four. A lot of them also draw on insights from a wide range of disciplines, including cognitive science, dynamical systems theory, economics, and psychology. Some are critical in that they reject a particular theory of collective intentionality, if not the very idea. Others are more constructive and develop existing theories, or propose new ones. Part I:  Collective Attitudes and Actions The essays in this collection fall naturally into two clusters. The first is concerned with collective intentionality in general and the analysis of collective attitudes as well as of collective actions more specifically. The second addresses collective rationality. A core concern that many of the chapters in the first part share is whether collective attitudes can be reduced to individual attitudes, or eliminated altogether. Related to this, a central question is: if particular collective attitudes cannot be reduced, how are they to be analyzed? What is striking is that most of the contributions do not take this to be an all or nothing proposition. Many contributors recognize that an argument concerning, for instance, cognitive states might not transfer to motivational states. Perhaps only some attitudes can be attributed to collectives and not others. In the first chapter of this volume, Deborah Tollefsen introduces a new theory of collective intentions that focuses on the dynamic process of maintaining coordination throughout the performance of a collective action. Tollefsen draws on empirical findings from cognitive science concerning the way in which individuals maintain control over their bodily actions over time. Her ambition is to explicate the dynamics of collective action in considerable detail in a way that is less cognitively demanding than rival theories are. She argues that a dynamic theory of collective intention serves to improve our understanding of the phenomenology of collective action and the experience of joint control.

4

I ntroduction

Raimo Tuomela and Kaarlo Miller present an analysis of collective goals that extends the ideas that in particular Tuomela has developed over the past three decades. Tuomela and Miller are concerned with understanding the apparent variety of collective goals and set out to explain what makes a goal collective. They argue that collective goals are collectively accepted and that a collective goal is such that it is satisfied for one member of a collective only if it is satisfied for all. Tuomela and Miller prefix their analysis with an illustration of the uses to which they can be put. They argue that conceptions of collective goals are important not only to a number of social sciences but also for political philosophy. More specifically, they point out a number of similarities between the notion of a collective goal and Rousseau’s conception of a general will. Frederick Schmitt criticizes the widely held view that group beliefs are not proper beliefs, but are acceptances instead. Whereas belief is commonly taken to be evidence-based and involuntary, acceptance is voluntary and based on pragmatic considerations. Schmitt argues against this that acceptance is involuntary at least some of the time. He goes on to argue that it is quite possible for a group attitude to be sensitive to evidence and aim at truth. To be sure, interests of the group or its members can subvert its doxastic attitudes, but the same holds for individuals. Hence, the role that interests play is not a good reason for doubting the existence of proper group beliefs. A further point Schmitt makes is that acceptance is parasitic on belief. This entails the idea that it is not possible for an agent to accept anything without having any beliefs. All of these arguments clear the way for the idea that, in addition to acceptances, groups can have proper beliefs. The chapter by Robert Rupert examines the question of whether cognitive states should be attributed to groups. Rupert investigates the explanatory power that cognitive states have insofar as collective actions are concerned. They seem to have such power in folk explanations. Rupert, however, argues that there are a number of disanalogies. First, none of us have first-person access to group attitudes. Second, group actions can be explained more parsimoniously in terms of individual attitudes. Rupert goes on to argue that cognitive science has not provided any evidence in favor of group attitudes. Furthermore, group behavior in animals can be explained in terms of individual cognitive processes. All of these considerations are broadly epistemic. And none of them support the idea that groups can have cognitive states. Kirk Ludwig argues that there are no group agents. In contrast to Rupert, he approaches the issue from a semantic rather than an epistemic perspective. Rupert grants that we typically speak as if there were group agents, as when we attribute actions to corporations or blame a group for a collective action. Ludwig argues, however, that the claims we make do not logically imply the existence of group agents. Ludwig’s argument revolves around a semantics that

Beyond the Big Four and the Big Five

5

he provides for the relevant kinds of claims that do not involve any quantification over group agents. A further implication of the proposed semantics is that there can be unintentional collective actions such that none of the individuals involved have any intention to perform their individual actions as part of a collective action. Interestingly, this puts Ludwig’s theory at odds with those of the Big Four who made joint intentions and/or individual intentions the cornerstone(s) of their theories. Social Ontology: The Big Five Theories of collective intentionality are part of social ontology. The ontology of the social realm includes a wide range of topics including social groups, social norms, and social institutions. A topic that currently draws a lot of attention is collective agency. The idea that is at stake in this debate is that it might be possible for a group to function as an agent in its own right. Peter French (1984) argued that the existence of such group agents has to be recognized if we are to make sense of the responsibility that organizations bear (see also Copp 2006, 2007 and Pettit 2007). More recently, Philip Pettit (2003) has argued that groups can have minds of their own. Together with Christian List, he has made this claim more precise by arguing that the beliefs of certain groups cannot be reduced to the beliefs of the individual members (List and Pettit 2011). The underlying idea is that if a group is to preserve its rationality, it will at some point have to adopt beliefs that are in some way inconsistent with those of its members. Even if this may sound surprising at first, this phenomenon can be encountered in collective settings ranging from small committees to big companies and governments. Consider the jury of a singing competition. The jury, made up of three people, uses three criteria when selecting finalists: vocal quality, technical skill, and musical talent. When a particular singer is up for evaluation, it can happen that his average is high on all three dimensions, even though none of the jury members give him high scores on all of them. This might mean that, even though none of the jury members would have accepted him as a finalist individually, they accept him collectively. List and Pettit (2006) conclude that, rather than supervening on individual beliefs per se, the beliefs of group agents supervene on those beliefs in combination with the collective decision mechanism that transforms them into a collective judgment. By assumption, all three jury members are individually rational. What the example reveals is that, if the jury as such is to form consistent beliefs and be collectively rational, it may have to take decisions that none of the members support individually. In such a situation, List and Pettit argue, there is a discrepancy between individual and collective rationality. In this respect, there

6

I ntroduction

are many parallels between their views and Kenneth Arrow’s (1950) work on preferences, voting paradoxes, and social welfare (List and Pettit 2004). In order for us to take the group seriously as an agent, Pettit (2003) proposes, the group has to “collectivize reason” by applying requirements of rationality such as consistency at the collective level. When this is done, the group acts as a rational and intentional agent and does in fact constitute such an agent. Due to his contributions to the literature on group agency as well as collective responsibility and sociality more generally, Pettit can—in combination with the Big Four of collective intentionality—be regarded as one of the Big Five of social ontology. Part II:  Collective Rationality The chapters in Part II of this volume reflect the significance of Pettit’s work on collective rationality, as well as that of others. What is distinctive about these chapters is that each of them in its own way challenges the idea that there is a straightforward dichotomy between individual and collective level rationality. As is discussed in more detail shortly, Abe Roth, Paul Weirich, and Julian Nida-Rümelin each show that new perspectives on the interplay between individual and collective rationality shed a different light on the alleged discontinuities between these levels. Melinda Bonnie Fagan asks whether scientific rationality is to be located at the level of individual scientists or at the collective level. She starts by examining scientific knowledge as a candidate for irreducible group belief. Beliefs are frequently attributed to the scientific community. Think, for example, of the claim that the physics community now believes that the Higgs boson has been observed. Note that in this example, the belief is attributed to a large, diverse, and geographically separated community of physicists. Fagan considers various arguments for and against the proposition that groups have irreducible scientific knowledge. But ultimately, she concludes that the crucial issue is not who possesses such knowledge, but how the knowledge is produced. Her contribution takes a significant step toward understanding the dynamic process by which individuals both rely upon and change the body of established scientific knowledge. She concludes that the traditional individualistic view of scientific knowledge is vindicated, whereas production of scientific knowledge is an irreducibly collective process. The chapter by Abe Roth concerns the rationality of group agents. Roth challenges the List and Pettit argument on which group agents have to collectivize reason and apply criteria such as consistency at the group level rather than at the individual level. He argues that, in order for a group’s rationality to truly override the rational concerns of the individuals, it must not be rational for the individuals to implement the group judgment. If this were the case, the group’s

Beyond the Big Four and the Big Five

7

decision and subsequent action would be straightforwardly rational from the perspective of the individual members. For example, if the members of a committee were to realize that they were on the cusp of a collectively irrational vote, and the members of the committee were each to be penalized for such a vote, then there is no conflict between individual- and group-level decisions. It would be straightforwardly rational for the individuals to vote in such a way as to avoid those penalties and thereby to safeguard the rationality of the group. In this way, Roth raises the stakes in the debate about what Pettit has called “group minds.” Paul Weirich argues that group actions are rational if the individual acts that constitute them are rational. He observes that individuals sometimes have conflicting goals and desires when they make decisions. Weirich goes on to acknowledge that some of the requirements that individuals can face align well with group efficiency, but that satisfying them can conflict with desires that the individuals may have qua individuals. The reverse is also possible. Weirich denies, however, that such cases reveal genuine conflicts between individual and collective rationality. Collective rationality does not demand individual efficiency in conditions that are not ideal for joint action. Furthermore, it is not irrational for individuals to make a trade-off between competing goals. In line with this, he argues that in the context of judgment aggregation, individuals can tolerate occasional inconsistencies without this entailing irrationality. Weirich concludes that individual rationality entails collective rationality. And finally, Julian Nida-Rümelin argues that cooperation is to be explained in terms of collective intentionality and that cooperation thus explained is rational. He proposes a new conception of rationality that he calls “structural rationality.” Roughly, structural rationality consists in being guided by patterns of behavior that the agent approves of or accepts. Nida-Rümelin argues that, as it respects the Ramsey postulates of the utility theorem, structural rationality is a conservative extension of rational choice theory. He goes on to argue that cooperation requires the availability of a collective action that is acceptable to all cooperators. Cooperation is in part explained—and this is where collective intentionality comes in—in terms of a consensus concerning a particular action as acceptable to all. The upshot is that cooperation is to be explained partly in terms of a normative belief about action opportunities. All of the contributions to the second part of the volume break new ground in their quest of charting individual and collective level processes, as well as the interplay between them. And most of them engage with one or more of the Big Five. Fagan resists the claim, which has been defended among others by Gilbert (2000), that scientific knowledge is irreducibly collective. At the same time, however, she acknowledges that collective processes are crucial insofar as the production of scientific knowledge is concerned.

8

I ntroduction

Roth is skeptical of Pettit’s (2001) claim that collective rationality sometimes requires compromising the rationality of individual members. He points out that individuals might have a stake in how collective decision processes develop. This insight takes the sting out of charges of irrationality targeted at individuals. Weirich agrees with Roth that individuals can have multiple goals, some of which pertain to collectives. He takes a further step, however, and argues that it is simply impossible for individual and collective rationality to diverge. Nida-Rümelin is also critical of alleged divergences between individual and collective rationality. He proposes a conception of structural rationality on which cooperation is often rational both individually and collectively. These chapters on collective rationality make abundantly clear that it is no longer an option for people who contribute to social ontology to juxtapose analyses of individual and collective level phenomena and claim that there is some discrepancy. Both the conception of rationality employed and the interplay between the two levels require careful scrutiny. It may well be that the alleged discrepancies dissolve in the face of an in-depth exploration of these two issues.

References Arrow, K. J. 1950. “A Difficulty in the Concept of Social Welfare.” Journal of Political Economy 58(4): 328–46. Bratman, M. E. 1992. “Shared Cooperative Activity.” Philosophical Review 101(2): 327–41. Copp, D. 2006. “On the Agency of Certain Collective Entities:  An Argument from ‘Normative Autonomy.’ ” Midwest Studies in Philosophy 30(1): 194–221. ———. 2007. “The Collective Moral Autonomy Thesis.” Journal of Social Philosophy 38: 369–88. French, P. A. 1984. Collective and Corporate Responsibility. New  York:  Columbia University Press. Gilbert, M. 1989. On Social Facts. Princeton: Princeton University Press. ———. 1996. Living Together:  Rationality, Sociality, and Obligation. Lanham, MD: Rowman & Littlefield. ———. 2000. “Collective Belief and Scientific Change.” In: Sociality and Responsibility: New Essays in Plural Subject Theory, 37–49. Lanham, MD: Rowman & Littlefield. List, C., and P. Pettit. 2004. “Aggregating Sets of Judgments: Two Impossibility Results Compared 1.” Synthese 140(1): 207–35. ———. 2006. “Group Agency and Supervenience,” Southern Journal of Philosophy 44: 85–105. ———. 2011. Group Agency. Oxford: Oxford University Press. Pettit, P. 2001. “Collective Intentions.” In Intention in Law and Philosophy, edited by N. Naffine, R. Owens, and J. Williams, 241–254. Dartmouth: Ashgate Publishing.

Beyond the Big Four and the Big Five

9

———. 2003. “Groups with Minds of Their Own.” In Socializing Metaphysics. The Nature of Social Reality, edited by F. F. Schmitt, 167–94. Lanham, MD: Rowman & Littlefield. ———. 2007. “Responsibility Incorporated.” Ethics 117(2): 171–201. Schweikard, D. P., and H. B. Schmid. Summer 2013 ed. “Collective Intentionality.” In The Stanford Encyclopedia of Philosophy, edited by E. N. Zalta. http://plato.stanford. edu/archives/sum2013/entries/collective-intentionality/. Searle, J. R. 1990. “Collective Intentions and Actions.” In Intentions in Communication, 401–416. Cambridge, MA: MIT Press, Bradford Books. ———. 1995. The Construction of Social Reality. New York: Free Press. ———. 2010. Making the Social World:  The Structure of Human Civilization. New York: Oxford University Press. Tollefsen, D. 2004. “ Collective Intentionality.” In Internet Encyclopedia of Philosophy. http://www.iep.utm.edu Tuomela, R. 1984. A Theory of Social Action. D. Reidel Publishing Co. (rep. Wien: Springer). ———. 2002. The Philosophy of Social Practices: A Collective Acceptance View. Cambridge, UK: Cambridge University Press. Tuomela, R., and K. Miller. 1988. “We-intentions.” Philosophical Studies 53(3): 367–89.

Part one

Collective Attitudes and Actions

1

A Dynamic Theory of Shared Intention and the Phenomenology of Joint Action D e b o r a h To l l e fs e n

1.1 Introduction Individual intentions shape and inform individual actions. But we do not always act alone, and it is coordination with others that raises interesting issues regarding the intentional structure of joint action. Indeed, many prominent philosophers of action theory, including John Searle (1990, 1995), Michael Bratman (1993, 2004, 2006), Margaret Gilbert (1989, 1996, 2003), and Raimo Tuomela (2000, 2002, 2005)  have argued that joint action requires joint or shared intentions. The accounts of shared intention on offer typically involve the postulation of a set of individual intentions (or attitudes of a certain sort) that have a common content and are interrelated in specific ways. Although these accounts have done a great deal to reveal the nature of joint agency, they only skim the surface of a complex phenomenon. This chapter attempts to dig deeper. It has two modest aims. First, I lay out a framework for a theory of shared intention that provides us with a richer understanding of the types of joint actions out there in the world and the types of agents engaged in joint action. This framework also helps us to understand how joint action unfolds over time and how joint actions are controlled and monitored by participants. Because this framework is modeled after Elisabeth Pacherie’s (2006) dynamic theory of intention, I will call this framework the dynamic theory of shared intention. Second, I want to make salient a feature of joint agency that has, for the most part, been overlooked in the literature on joint action—what I call the phenomenology of joint

14

C ollective Attitudes and A ctions

agency.1 The experiential aspect of doing things with others plays a role in the control and monitoring of joint actions, or so I shall argue. A few clarifications before we begin:  In this chapter I  am focusing on the notion of shared intention. This refers to the ways in which individual participants in a joint action have intentions with a common content and how those intentions are interrelated. Shared intention also goes by the name of joint intention and we-intention in the literature. By group intention, I  mean the intentions that are appropriately attributed to a group rather than to the individuals that comprise the group. Many believe that there are no group intentions or any other group attitudes, such as belief. I have argued elsewhere to the contrary (2002). But because my focus in this chapter is on shared intention, I will not discuss the attribution of intentional states to groups.2 Also, I will not be arguing for the existence of shared intention in this chapter. John Searle (1990, 1995), Michael Bratman (1993, 1999), Margaret Gilbert (1989, , 1996, 2003), and Raimo Tuomela (2000, 2002, 2005) have spent a good part of the past three decades establishing the need to move beyond simple summative accounts of shared intention. On a simple summative account, A and B share an intention to x, just in case A has the intention to x and B has the intention to x. It has been well established that joint action requires that individuals share intentions in a more robust sense than present in summative accounts. If anyone remains skeptical as to the need to posit more robust shared intentions, I direct them to the vast literature on shared intention. 1.2  Why Do We Need a Dynamic Theory of Shared Intention? Although I will not be arguing for the need to acknowledge shared intention, I  will argue for the need to move beyond the theories now on offer. I  have argued elsewhere (2005) that standard accounts of shared intention provide analyses that do not extend easily to the case of joint action among young children and nonhuman animals. Aside from Searle’s theory of collective intentionality, current accounts of collective intentionality arguably presuppose that the participants have a fairly sophisticated theory of mind. Consider, for instance, Michael Bratman’s account of shared intention: (1) a. I intend that we J. b. You intend that we J. (2) I intend that J in accordance with and because of 1a and 1b, and meshing subplans of 1a and 1b; you intend likewise. (3) 1 and 2 are common knowledge between us (Bratman, 1993)

Dynamic Theory of Shared Intention

15

To fulfill these conditions requires that the participants be mutually responsive to the intentions of other participants and that the complex of attitudes that constitutes shared intention be formed under conditions of common knowledge. But how can a participant be mutually responsive to the intentions of other participants if she lacks the ability to attribute intentions to others, as young children (prior to the age of 4)  and animals do?3 And to know what another knows, as some theories of common knowledge require, seems to presuppose an ability to attribute attitudes, specifically beliefs, but young children and animals are thought to lack this meta-representational ability.4 If shared intention is required for joint agency, then it appears that young children and animals cannot engage in joint action. But prima facie they do engage in joint action. They walk together, move objects together, and play with others. These are paradigmatic forms of joint action, and there is no reason to rule them out as genuine forms of joint action. Indeed, to do so begs the question regarding the nature of joint agency.5 The existence of joint action in animals and young children does not, however, mean that the current accounts of shared intention and joint action are false—they may well be true for certain types of adult human action. What it means is that we need to think more carefully about the types of intention and the types of perspective-taking that might be in play during certain forms of joint agency. It has also been noted by Christopher Kutz (2002) that the standard accounts of shared intention do not help us to understand joint actions that occur “on the fly” or spontaneously without prior planning.6 Most accounts of shared intention are modeled on the idea of conscious and planned action. Thus, they do not help us to understand the nature of joint actions that occur without reflection or planning, but that nonetheless involve intentions. Finally, the current accounts of shared intention (and this includes Searle’s account) do not provide us with an understanding of how joint actions unfold over time and how individual agents monitor both the joint action and their own actions in order to fulfill the shared intention. That is, current accounts have focused almost exclusively on the planning of joint actions rather than the execution of joint actions. They focus on the project (joint action) rather than the process (joint acting). Bratman’s introduction of shared values (2004, 2006) has gone some way toward explaining the stability of joint actions over time, but I think there is more to be said about how subplans are executed in a joint action.7 In particular, I think we need a more naturalized and dynamic theory of shared intention—one that takes into account both the planning and execution of joint actions.

16

C ollective Attitudes and A ctions

1.3 A Dynamic Theory of Intention In “Towards a Dynamic Theory of Intentions” (2006), Elisabeth Pacherie argues that a full understanding of the agency of individuals requires that we identify several forms of intentions, their different functional roles, their distinctive content, and the dynamic transitions among these forms of intention. She is not the first to identify different levels of intention. John Searle (1983), G.  E. M.  Anscombe (2000), and others, have developed the idea of intention-in-action, for instance, and there is a great deal of work in psychology, cognitive science, and cognitive neuropsychology on the unconscious processes that guide bodily action.8 But Pacherie’s discussion of the different levels of intention and their dynamic interaction is a particularly nice example of how philosophy and cognitive science can provide a more detailed account of human agency and, by extension, I shall argue, a more detailed account of joint agency. The first level of intention involves future-directed intentions (also called prior intentions). Future-directed intentions terminate practical reasoning about ends, prompt practical reasoning about means, and help to coordinate actions between persons and within a single agent across time. They are formed prior to the onset of the action and represent the action as a whole unit. The content of these intentions specifies an action type rather than a token. Their general content is required because the satisfaction of a future-directed intention depends upon unpredictable features of the situation.9 Future directed intentions are also subject to normative constraints. As Michael Bratman (1987) has argued, these types of intentions are subject to means-end coherence and must be consistent with other beliefs and attitudes the agent has. The second level of intention involves present-directed intentions (also called intentions-in-action). When I spontaneously get up from my desk and go to the refrigerator for a drink, I have not made a prior intention to do so, but I do so intentionally (or with the intention of getting a drink). My intention is present in my action. These intentions are typically assigned the functions of initiating the intended action, sustaining it until completion, guiding its unfolding, and monitoring its effects. Present-directed intentions often inherit their action plan from a future-directed intention. As Pacherie describes it, the present-directed intention “anchors” the future-directed intention in the context of the action. It provides the specifics of the plan.10 Present-directed intentions ensure the rational control of the ongoing action where this involves both a “tracking control” and “collateral control.” Tracking control enables the agent to monitor her own action, keep track of its progress, and allows for adjustments in order to increase the chance of success. Collateral control involves control of the possible side effects produced while performing

Dynamic Theory of Shared Intention

17

the action. The agent will either adjust to undesirable side effects or, in some cases, abort the action. Because present-directed intentions are responsible for the monitoring and guidance of actions, they apply to a context of action that the agent is usually consciously perceiving and conceptualizing. There is also a lower level of guidance and control of an ongoing action that is much finer grained and responsible for the precision of the action and the smoothness of the execution. This lower level is subpersonal and unconscious. This is the third level of intention, and it involves motor intentions. The role of motor intentions is to transform the perceptual information an agent receives from the environment into sensory motor information. This sensory motor information allows for the monitoring and control of the precise movements of the body and its interaction with the environment. There is a vast literature on motor cognition,11 and I do not have space here to discuss all of the empirical literature regarding the content and role of motor representations. Here are a few details regarding motor cognition and its relation to perception that are relevant to my discussion of joint action. Although there remains controversy regarding the exact nature of perceptual and motor representations, one of the most prominent theories on offer (Jeannerod 1997, 2006; Jacob and Jeannerod 2003) suggests that there are two visual systems—one that is dedicated to the identification and recognition of objects and another that is dedicated to the identification and recognition of action. The vision-for-action system extracts visual information about a situation and the objects in the context and uses this information to build motor representations that aid us in acting on the environment. So, for instance, when I look at a table and see a cup that I would like to grab, the visual system for action will represent the cup in terms of the type of movements needed to reach for it and its shape and size in terms of the type of grip that it affords. The vision-for-action system builds upon an environment’s affordances and uses internal models of action or motor programs (Gibson 1979). The goal of the action and appropriate motor programs will determine the way the cup will be represented to me by narrowing the possible affordances. As Pacherie describes it: A given situation usually affords more than just one possibility for action and can therefore be pragmatically organized in many different ways. These affordances along with the goal of the actor “prepotentiate” motor programs. P-intentions can govern the selection of motor programs in that the goal can be inherited from a p-intention. If my p-intention is to get a cup of coffee, the m-intention will inherit this goal and determine a motor program for action that allows me to grasp the mug in an appropriate manner. If my p-intention were to throw the cup at an intruder, the m-intention might

18

C ollective Attitudes and A ctions

involve a representation slightly different and the motor program for throwing the cup based on its affordance as a large hard object rather than one for drinking from. (2006, 13) Thus, present-directed intentions have a downward dynamic that informs the motor intentions. When motor intentions are governed by present directed intentions, there will be finer control of the action. Motor programs come in two forms:  forward and inverse (Wolpert and Kawato, 1998). Forward models represent the causal flow of a process in a system and use it to predict the next state of the system. Inverse models work backwards; given a certain goal and the current state of the system, the program works out steps that would allow one to achieve the goal. Inverse models compute the means toward a given goal, and forward models compute the consequences of implementing these means. The control of movement typically depends upon the use of both forward and inverse models. The sensory motor system makes comparisons of various representations of desired, predicted, and actual states of affairs and uses the results of this comparison for the regulation of bodily movements. Pacherie calls her theory of intention dynamic because it highlights the ways in which these levels interact in order to monitor and control human action. Information from higher levels is transformed into representations at the lower levels of intention, and likewise there is a feedback to higher levels. For instance, when information regarding collateral effects comes from motor representations, it makes its way to the level of present-directed intention, and changes in present-directed intention will have feedback effects on various other motor representations. But not every action will be controlled and monitored by all three types of intentions. Some decisions are made without prior planning, and so there is no future-directed intention involved. And the existence of routine, reflexive, or automatic actions suggests that in some cases, present-directed intentions are not necessary in order to initiate the action. Novel or difficult actions are typically much more closely controlled by present-directed intentions, but habitual actions that require very little reflection could simply involve motor intentions where motor programs are put in place and carried out unconsciously.12 1.4 A Dynamic Theory of Shared Intention There is a great deal more to be said about these various types of intention, but this brief overview will provide a basis for introducing a more complex framework for understanding shared intention and its role in guiding joint agency. It is my contention that in order to accommodate a wide variety of types of

Dynamic Theory of Shared Intention

19

joint action and types of agents engaged in joint action, we need to acknowledge three levels of shared intentions: shared future-directed intentions, shared present-directed intentions, and shared motor intentions. Shared future-directed intentions, like their nonshared counterparts, will serve as terminators of practical reasoning about the ends needed to complete the joint action and as prompters of practical reasoning about means and plans, and will help to coordinate individual actions and resolve conflicts between participants in the joint action. They are formed consciously prior to the joint action. Most of the accounts of shared intention in the philosophical literature offer us a theory of shared future-directed intentions. They offer us an intentional structure that is in place prior to action and that provides rational (and in some theories, normative) guidance for joint actions. I remain agnostic regarding the exact form these intentions take. It seems to me that shared future-directed intentions will take a variety of forms, depending on the type of group performing the action. In some cases, Gilbert’s plural subject theory will be appropriate; in other cases Bratman’s will be fitting, and so on. But these types of shared intention cannot be all there is to the story. If we consider the paradigm cases of joint action in the literature—painting a house together, executing a basketball play—future-directed shared intentions do not seem to help us understand how the joint action is able to unfold over time and how individual actions of which the joint action is composed are coordinated in such a way to fulfill the shared intention. They specify a type of action—“I intend that we paint the house”—rather than a token action. Thus, they don’t provide us a way for understanding how the actions of individuals are coordinated in order to fulfill the general goal specified in the content of “I intend that we J.” Nor do they provide an understanding of joint actions that occur “on the fly”—without prior planning, commitments, or agreements. Thus, we need to acknowledge another level of shared intention:  the level of shared present-directed intention. Why must we acknowledge such a level? Won’t individual present-directed intentions, those governing the individual actions that comprise the joint action, be enough? Recall that the role of present-directed intentions is to initiate the intended action, sustain it until completion, guide its unfolding, and monitor its effects. Present-directed intentions often inherit their action plan from a future directed intention. The general content of the future-directed intention is transformed into a specific action plan. If the shared future-directed intention is going to do its work in guiding the joint action, it too must be transformed into a specific action plan; but the plan cannot be just about the agent’s own actions. It must also make reference to actions of other participants. There will, of course, be specific action plans initiated by individuals that will govern individual action; but if there is no present-directed intention that is shared in the sense of making reference to

20

C ollective Attitudes and A ctions

content of the intentions of others’ present-directed intentions, then the individual intentions will not be coordinated in a way that ensures that satisfaction of the shared future-directed intention. Suppose we share, in the Bratmanian sense, the future-directed intention of moving a table from the downstairs living room to the upstairs den. I intend that we move the table upstairs; you intend that we move the table upstairs. My intention is formed by way of your intention and vice versa, and all of this is done under conditions of common knowledge. This complex of attitudes will guide and shape our individual future- and present-directed intentions. Because of this shared intention, I will form present-directed intentions (and perhaps future-directed intentions) involving the side of the table that I lift, but these intentions will need to be guided not just by the prior shared intention that we move the table but also by shared present-directed intentions to move the table together in specific ways that will help us negotiate the stairwell and other aspects of the present environment. Present-directed intentions provide both collateral and tracking control. In the case of joint action, we must track not only our own actions but also the joint action itself and our relation to it. It is the tracking of the joint action that informs me to adjust my own individual actions. If we are failing to lift the table high enough to negotiate the stairs, you or I will adjust our individual actions in order to fulfill our future-directed intention. But this adjustment will only be made if I am also tracking the joint action as a whole and not just my individual agency. Likewise, in order to identify collateral effects of the joint action, which will then inform individual present-directed intentions, there needs to be a shared present-directed intention that provides a representation of the joint action as a whole and not simply my part of the action. What will shared present-directed intentions look like? I have argued elsewhere (2005) that we can understand shared present-directed intentions along Bratmanian lines—as complexes of individual present-directed intentions that share a common content.13 However, their content will be more specific. Further, these intentions will be formed under conditions of joint attention rather than common knowledge; participants are mutually aware (a purely perceptual notion) of the intention in actions of other participants. The existence of shared present-directed intentions makes sense of the way in which young children and animals can engage in joint action. It is generally accepted that young children and many primates, though they lack a robust theory of mind, have an understanding of intentional agency. Reading intention does not involve reading minds; it involves interpreting action in a context. Likewise, because joint attention does not require knowledge of other minds but an awareness of what another is seeing, a capacity both animals and young children share, there is no concern that shared present-directed intentions

Dynamic Theory of Shared Intention

21

involve a robust notion of common knowledge. Shared present-directed intentions involve the sharing of a perceptual space and a sharing of its affordances. The presence of shared present-directed intentions also makes sense of joint actions that occur spontaneously and without preplanning. Suppose you and I simply see that the table needs to be moved out of the way, and the upstairs den offers the only alternative spot. We might simultaneously pick up ends of the table and begin to move the table without prior planning. In this case, there is no shared future-directed intention, but there are shared present-directed intentions that control and monitor our individual action and the joint action as it unfolds over time. Not all forms of joint agency will require shared present-directed intentions. When, for instance, two individuals share a future-directed intention to meet at the coffee shop, their individual actions need not be governed by shared present-directed intentions, because getting to the coffee shop need not require the fine-grained coordination of their individual actions. All that is required is the shared future-directed intention that informs and guides their individual present-directed intentions. The need for shared present-directed intentions arises when fulfilling the future-directed shared intention involves the coordination of bodily movements across space and time. And it is in these cases that we will also need an understanding of the motor representations that make joint action possible. Thus, we must also acknowledge a level of shared motor intentions. There is a variety of new empirical findings that suggests that in cases of joint action where participants need to coordinate their bodies, they do so on the basis of shared motor representations (Sebanz, Bekkering, and Knoblich 2006; Obhi and Sebanz 2011; Knoblich and Jordan, 2003). Specifically, they share task representations in that they integrate others’ action capabilities into their representation of the action and their own action planning. They also use feedback about others’ actions in order to anticipate the control of their own individual actions (Knoblich and Jordan 2003; Richardson, Marsh, and Baron 2005, 2007). In addition to shared-task representations, motor resonance may play a crucial role in the coordination of bodily movements. A multitude of studies have shown that during observation of action, a corresponding representation in the observer’s action system is activated (Gallese, 2001, 2009a, 2009b, 2009c; Iacoboni et al. 2005). Although there is a great deal of controversy regarding the role of mirror neurons in theory of mind, such motor resonance might be the neural basis for a more sophisticated theory of mind reading. In the context of joint action, the motor resonance system might, therefore, play a role in the formation of shared future- and present-direction intentions. In order to form shared future- and present-direction intentions, I need to, in some cases, predict the actions of others. My motor resonance system may help me predict what

22

C ollective Attitudes and A ctions

you will do with your end of the table in order that I might be able to coordinate my actions at my end of the table (Chersi 2011, Pacherie and Dokic 2006). As with the other levels of intention, motor intentions will often be informed by intentions formed at the higher level. Present-directed intentions will inform the motor intentions and offer a way of narrowing the motor programs (task representations) relevant to achieving the shared intention or goal. In addition, because joint action does not always require imitative action (doing what another does) but complementary action, there needs to be a suppression of the resonance system. Although there is no clear consensus on the mechanisms that inhibit motor resonance, shared intention—either present- or future-directed—could play a role in suppressing mimicry and facilitating the performance of actions that complement the other participant. Again, these levels of shared intention need not all be present in every joint action; just as individual actions may be on the fly, so too we will have cases of joint action on the fly, and so there will be no need to posit the existence of shared future-directed intentions. We might even imagine cases of joint action at the level of adults that are so automated and reflexive that they are governed completely by motor intentions—perhaps certain moves in a skilled ballroom dance represent such a case. 1.5 The Phenomenology of Joint Agency With the dynamic theory of shared intention in mind, I now want to turn to the phenomenology of joint agency and its role in controlling and monitoring joint action. The experience an individual has of his own agency has been of interest to those working on free will and determinism for some time now (Frankfurt 1988) but it is not until recently, with the renaissance of interest in consciousness and phenomenology in general, that philosophers of mind have explored the issue (Horgan, Tienson, and Graham 2003; Bayne and Levy 2006; Pacherie 2006, 2007). As many have noted, the sense of agency we experience as human agents is really made up of a number of distinct experiences including the sense of initiating an action, the sense of mental causation (the feeling that one’s intentions cause one’s actions), and the sense of authorship. I will focus my discussion on what has been identified as the sense of control. I do so not only to make the discussion more tractable but also because the issue of control and its relation to intention arise in a number of different ways in discussions of shared intention and joint agency. The sense of control itself is not a simple phenomenon. The phrase “sense of control” is often used to refer to two different experiences. On the one hand, it

Dynamic Theory of Shared Intention

23

may be used to identify the sense of control one has over an action, where an agent feels that he is guiding and determining the course of events—not only his own body but also aspects and elements that we would include in a description of the event (e.g., in riding a bike for instance one might have a sense of control over one’s legs and arms but also over the bicycle). On the other hand the phrase “sense of control” may be used to identify the level of effort one has to exert in order to initiate, control, and maintain an action in the face of perturbations (e.g.,when the bike hits a rock in the road and one needs to gain control of the trajectory of the bike). Where does the experience of agency come from? Because future-directed intentions are formed prior to the action, they cannot contribute to the sense of agency one has as they engage in an action. They do, however, provide the agent with a sense of personal continuity and a temporally extended sense of agency. Present-directed intentions and motor intentions both guide and control action and occur simultaneously with the action, and so they are the most likely candidates for contributing to the experience of agency, and specifically to the sense of control. As we have seen, motor control involves motor representations in the form of models or programs that produce action anticipation and correction. The motor control system uses both forward and inverse models to predict the effects of certain motor commands. The motor system also makes comparisons between predicted effects, desired effects, and actual effects and detects mismatches between them. Although these mechanisms are typically unavailable to consciousness, Pacherie (2007) suggests that they may nevertheless underlie the experience of acting in its most basic form. When the discrepancy between predicted and actual sensory consequences of the movements becomes too large to be automatically corrected, it becomes accessible to consciousness, and present-directed intentions must be formed in order to make corrections. This discrepancy gives rise to a feeling of a loss of control or of effortfulness. The sensory motor signals used for motor control, which influence the dynamics of intentions at higher levels, contribute to the experience of control we may have as corrections are made. As Pacherie describes it:

This is also why the representational content of the experience of acting may appear so thin. This is especially so when one is engaged in. . . “minimal” actions, actions that are performed routinely automatically, impulsively or unthinkingly. These actions unfold with little or no conscious control by present directed intentions. Their phenomenology may therefore involve nothing more than the faint phenomenal echo arising from coherent sensory motor flow. (2007, 15)

24

C ollective Attitudes and A ctions

What role does this sense of control play? The sense of control or lack of control over our own actions operates as a feedback mechanism in the control and monitoring of actions at both the level of motor intentions and present-directed intentions. It is because I begin to feel out of control that I monitor more closely for collateral effects and, as a result, form present-directed intentions that influence the motor system. As my motor system adjusts I may feel, as a result of this change, a sense of control or a lack of effortfulness. Likewise, a feeling of effort provides feedback that alters both my present-directed intentions and the motor system. I may form prior intentions or future-directed intentions with the aim of changing my actions in the future based on the experience of control or the experience of a lack of control. If our experience of our own agency involves a sense of control or, in some cases, an experience of a lack of control, it is natural to ask whether joint agency has a corresponding phenomenology and to what extent this phenomenology is different from our experience of our own agency. In what follows, I explore this issue in more detail. We have seen that the experience of agency involves many different experiences. I do not have space here to explore the possibility that there is an analog to each of these experiences when one is engaging in joint agency. It is likely that the experiences we have when engaging in joint agency are a function, in part, of the experiences we are having of our own agency. But I think there is more to the experience of joint agency than this, and this becomes obvious when we focus on the sense of control. Philosophers of mind have identified the various aspects of the phenomenology of agency by engaging in the method of contrasts in which various cases of agency are contrasted with other cases in an effort to make the phenomenon more salient. Such a methodology might be called descriptive phenomenology.14 I will follow this methodology and propose a few cases for contrast in order to reveal the type of experience I want to identify. There is a small four-legged maple coffee table that must be moved from the downstairs to an upstairs bedroom that serves as a family room. The table weighs approximately 50 pounds. You place your hands on each side and lift the table, making sure the weight of the table is evenly distributed and you walk to the stairwell. Each step requires that you maintain your grip on the table and monitor the table’s movements as well as your own. You cannot see your feet, so your steps must be made in a way that monitors for collateral interference. You may have to adjust the level of the table, raising it above your knees to around your chest to avoid hitting it on the stairs. And as you reach the top of the stairs, you may have to adjust the way you hold it, turning it, perhaps, to fit through a doorway. The task is more complex than I make it sound here in terms of the subpersonal mechanisms involved, and my description may sound too complex

Dynamic Theory of Shared Intention

25

in the sense that normally, we would perform such a task without any reflective awareness of our steps, positioning of arms, and so on. Now consider the sense of control and how this sense of control might alter and change at various stages of the task—the experience, for instance, of bumping into the railing along the stairway and, for a moment, losing control of the table and your own equilibrium; or the sense of the table leaning forward in your procession and hence pulling you along as it falls to the floor. If the table is cumbersome and heavy, the sense of effortfulness will be acute. Now consider the experience of lifting and moving the table up the stairs with another person. You position yourselves on each end of the table and lift at the same time. Sometimes lifting together will be coordinated by a synchronizing function (at the count of 3); other times, the joint action can be initiated by simply saying “OK, lift!” or a nod of the head. One person might lift first, and then the other initiates lifting the other side. The feeling of the weight of the table will be different—for one, it won’t be as heavy. But there is also a feeling that things are not completely under your control. The end of the table at which your partner stands is not completely under your control—nor is it completely out of your control. Your own movements at your end of the table influence the movements of the other participant. The movements of each participant will shift the weight of the table and subsequently change the experience of control each feels for the part of the table they control. Now compare the experience of moving the table up the stairs alone and carrying the table up the stairs jointly. When I carry the table by myself, I can control the distribution of weight. If it feels as if it is falling forward, I can adjust my arms and my body to reconcile the situation. When I carry the table with another, the distribution of the weight of the table is not entirely in my control (nor is it entirely out of my control). We jointly control the way in which the table is positioned, and when I feel that you are slacking, for instance, I have to not only adjust my own body but also get you to adjust yours. I might feel a loss of control in precisely the same way the object hitting the banister made me feel a loss of control. But when we compensate and adjust our bodies so that we are coordinated in a way that distributes the weight of the table evenly or at least evenly enough for us to function, I do not have the sense that I alone am in control of the movement of the table; rather I feel as though we have gained control of the situation (at least in cases when I am not doing all the work!). Again, when I lift the table with others, it isn’t as though someone else is controlling my action. Although this might happen when your partner pushes you up the stairs and you lose control of the speed of the action, in a successful joint action, there is a sense in which control is distributed among us—there is an experience of joint control. In successful cases of joint action, we have a distinct experience that might be captured by the following phrase: “We are in control.”

26

C ollective Attitudes and A ctions

So we can contrast the feeling of control one has when they carry the table themselves with the feeling of being pulled by the table as it falls to the ground, and again, with the sense of control one feels when the table is jointly carried by you and a partner and the fluctuation of control one might feel in their own action as well as the distribution of control experienced as you successfully carry the table up the stairs with another. The difference in the experience marks the distinction between the experience of control and the experience of joint control. Consider another set of cases: dancing alone; dancing in concert with someone else, as in the case of a line dance; and dancing together, as in the case of ballroom dancing.15 When dancing alone, I have control over my movements. I have an experience of being in control and perhaps of losing control, at times. The sense of effortfulness may become (at least it does for me) more acute when trying to dance in concert with others or in ballroom dancing. But during line dancing, I do not have a distributed sense of control. I do not have a sense that we control our actions; rather, you control your action, I control mine, and we control our own actions while keeping in mind the actions of others. My action possibilities are narrowed by the space and by the actual dance, but the action is not yet joint, and I do not experience it as such. We are engaged in many individual actions that happen to be coordinated or synchronized. But when I dance the tango with a partner, the experience is unique. Dancing the tango is a type of action that is essentially joint. I cannot dance the tango alone. And until our bodies are thoroughly coordinated in the ways that the dance requires, we cannot be said to be dancing the tango at all. What experience of control does one have in these cases? There is, of course, the experience of control one has over their own movements within the joint action, though if I am being led, I will feel less control (or rather, I will be in control of the loss of control) than if I were to dance alone. But there is a sense in which we jointly control the tempo of the dance and the path the dance will take. I experience our movement and not just my bodily movement. If the dance is successful, our movements will be seamless. Our experience will be equally seamless. That is, our experience of the action will not be easily parsed into the experience of control I have over my own body and the experience you have of control over your body. We will have a sense of joint control—a sense that we are dancing. Where does this experience of joint control come from, and what role does it play? Pacherie has suggested that experience of control at the level of individual action arises in part from information at the level of motor intentions. Motor control at the level of motor intentions is automatic, and typically, subjects are not aware of the nature of adjustments and corrections made at this level. It is only when the discrepancy between predicted and actual sensory consequences of the movements becomes too large to be automatically corrected that it becomes accessible to consciousness. The sensory motor signals

Dynamic Theory of Shared Intention

27

used for motor control could then contribute to the experience of control and, thus, the experience of joint control. So, when there was a prediction based on the shared future-directed intention to move the table upstairs or the shared present intention, and this prediction did not match the actual sensory consequences of our movements, the experience of joint control (or the lack of it) would become salient. At the level of shared present-directed intentions, the sense that we are in control would rely on the perceived match or mismatch between the predicted perceptual effects, corresponding to the situated-shared goal, and the actual perceptual effects of our joint agency, while at the level of shared future-directed intentions, this would depend on whether the joint action is carried out successfully. At both levels, the better the match, the more feeling of control and in the case of joint action, the more feeling of joint control. In the case of individual action, the sense of control has significant effects on the action as it develops over time. Collateral control, for instance, is often influenced by the experience I have of being in control of the action. When I carry the table up the stairs alone and feel my control waning because I hit the wall, this experience may lead to adjustments in motor programs or in present-directed intentions— take the elevator. It may even result in the formation of different future-directed intentions. In the future, I will look for another way to take this table upstairs. Likewise, the sense of joint control will have the function of providing feedback to participants and will allow them to alter both shared present-directed intentions and shared motor intentions as well as their own individual intentions, both present- and future-directed. The feeling of a loss of our control over the movement of the table may lead us to alter our shared present-directed intentions regarding the way in which the table will be carried by us, and our individual present-directed intention regarding our part of the joint action. My discussion of the experience of joint control has focused on a case of joint action that requires just two participants and the coordination of bodily movements. There are many other forms of joint action that include other participants and where there is no need for the close monitoring of other’s bodily movements or the coordination in a particular space and time of bodies. These may be cases in which all that is needed are shared future-directed intentions. But this is not to say that there will not be experiences of joint agency in these cases. I  use the example of carrying a table because reflection on it makes salient the experience of joint control I want to highlight. In other forms of joint agency, the experience of joint control may not be present, but other experiences associated with the phenomenology of joint agency may be. Consider the experiences one might have as the member of a successful athletic team—the feeling that we did it. This seems to be a paradigmatic experience of authorship, but in this case, it is not an experience of one’s own authorship but of joint authorship.16

28

C ollective Attitudes and A ctions

1.6 Conclusion There is much more to be said regarding the dynamic theory of shared intention I propose here and the phenomenology of joint agency. My aim has been to lay out a framework and to highlight a phenomenon that I think needs to be included in the current discussion of joint agency. Joint agency is quite complex, and just as individual actions require for their understanding the specification of both personal and subpersonal mechanisms, joint agency, too, will require a look at the personal and subpersonal mechanisms responsible for their production. I have argued that we need to acknowledge shared intentions at various levels and acknowledge their role in the monitoring and guiding of joint action as it unfolds over time. Doing so not only provides a more psychologically informed account of joint action, but it also makes sense of joint agency among young children and animals and joint agency that occurs spontaneously. I have also argued that identifying the cognitive aspects of joint agency will not be enough. We also have to acknowledge the phenomenology of joint agency. Part of what allows us to guide and monitor our joint agency and alter our individual agency so as to contribute to the joint action as it unfolds over time is our experience of joint control—one aspect of the phenomenology of joint agency. I hope my focus on this phenomenon will foster more discussion regarding the qualitative aspects of doing things with others.17, 18

Notes 1. For those familiar with the phenomenological tradition, my use of the term “phenomenology” may seem inappropriate. I use the term in the way that some analytic philosophers of mind have adopted in recent years to identify both a methodological approach of describing experience and the nature of such experience. I am aware that this is not the sense in which Heidegger, Husserl, and other thinkers of the phenomenological tradition used the term, and I don’t intend here to enter a debate as to whether the allocation of the term is inappropriate or philosophically suspect. 2. Hans Bernhard Schmid (2009) makes a useful distinction between collective agent models of joint action and teamwork models of joint action. My focus in this chapter is on teamwork models of joint action. I agree with Schmid that teamwork models are fundamental. Indeed, it is precisely because individuals share intentions that the actions of structured groups such as corporations are interpretable from the intentional stance. 3. I refer to this as the mutual responsiveness problem (2005). Although some research suggests that children as young as 18 months have the capacity to read intentions (Tomasello et al. 2005; Tomasello and Carpenter 2007; Krachun et al. 2010), there is still controversy concerning the onset of a theory of mind in children and even more debate regarding non-human animal’s theory of mind. I don’t have space here to enter this debate fully. My point, in (2005) and here, is that reading intentions is not the

Dynamic Theory of Shared Intention

29

same as being mutually responsive to the intention of the other as is required by Bratman’s theory. Being mutually responsive to the intentions of others requires not just a reading of the intention but also a tracking of the intention as it unfolds and a monitoring of it in comparison with one’s own intention. It seems clear to me that young children have a rudimentary theory of mind, but the current theories of joint action, developed with the adult human in mind, presuppose a robust theory of mind—one that requires the recognition, responsiveness, and attribution of mental states (not simply an understanding of intentional or goal directed behavior) to another and oneself. There is no need to attribute such sophistication to infants and young children in order to explain how they engage in joint actions. The same goes for animals. 4. I call this the “common knowledge” problem in my 2005 article. 5. In my 2005 article, I suggest that if one understands common knowledge as something like “common ground” or “perceptual openendness,” then this resolves the common knowledge problem; and if young children have an ability to read intentions-in-action, this might help to resolve the mutual responsiveness problem. In effect, I suggest that the intentions present in cases of joint action among children involve shared present directed intentions and a mutual awareness of them rather than shared prior intentions. So, I offer a revision of the Bratman analysis that accommodates joint action in young children. I say more about this revision when I discuss shared intentions in action below. 6. Kutz also argues that standard accounts of shared intention are egalitarian and cannot make sense of shared intention in a group that is hierarchically structured. Because my own theory can accommodate Kutz’s theory as well as others, I will not enter this debate here. I remain agnostic as to the exact content and form of shared intention. I suspect that shared intention will be realized in a variety of different ways, depending on the nature of the group and the nature of the agents within the group. 7. One might respond on behalf of Bratman and others that their aim was to provide a philosophical analysis of joint action and not an account of the mechanisms that bring it about. The philosophical literature on joint action is concerned with giving an analysis of the concept of joint agency rather than an explanation of how joint action is executed. This rejoinder sees execution of action as something the cognitive scientist or psychologist—not the philosopher—should be concerned about. Therefore, the fact that philosophical theories do not explain how joint actions unfold over time is not a mark against them as this is not their aim. This response rests on a controversial distinction between philosophy and cognitive science and the nature of explanation in both. I don’t have space here to challenge this distinction. If one would like to read me as offering a cognitive science of shared intention and action that augments existing philosophical theories, I would not be adverse to that. 8. See, for instance, the work of Marc Jeannerod (1997, 2006) and Jacob and Jeannerod (2003). 9. They are, as Bratman (1987) describes it, partial plans to be filled in as the action unfolds. 10. “The temporal anchoring, the decision to start acting now is but one aspect of this process. Once the agent has established a perceptual information-link to the

30

C ollective Attitudes and A ctions

situation of action, she must insure that the action plan is implemented in that situation. This means that she must effect a transformation of the purely descriptive contents of the action plan inherited from the future directed intention into indexical contents anchored on the situation” (Pacherie 2006, 6). 11. See, for instance, Jeannerod (1997, 2006), Gibson (1979), Goodale (1998), Milner and Goodale (1995). 12. One might object that there if are no present- or future-directed intentions involved, then there is no “action,” but mere behavior. If someone throws a ball at my head and I raise my arm to catch it, this seems a routine and reflexive action, one that may have involved no intention in action or prior planning, but it seems quite different from a mere behavior. It still seems that it is something that I do. We would not describe it as an accident or something that happened to me rather than something that I did. The attribution of action to individuals seems to suggest that we are more flexible with our notion of agency than most action theories acknowledge, and that whether something counts as an action or not depends on more than the way it was causally initiated, but also on the context in which it takes place and holistic considerations involving the agent’s others dispositional states. This suggests we ought to abandon a simply causal theory of action (both at the individual and the joint level), but I do not have space here to pursue this line of argument. 13. In Searle (1990), we-intentions are described as intentions in action. Searle’s account of shared intentions then is not modeled on the idea of prior intentions in the way that Bratman’s is. However, because Searle’s account does not require that we-intentions be had by other participants, and does not require that the content of the we-intention be somehow shared with other participants, it is difficult to see how his we-intentions-in-action would be able to play the role intentions in action are supposed to play. In effect, his we-intentions are simply too solipsistic to play a role in coordinating bodies in space and time. 14. Thanks to Lucas Olsen for suggesting this phrase. 15. I thank Susanna Siegel for suggesting these cases to me. 16. The phenomenology of joint agency, although foreign to philosophical discussions, has not been entirely overlooked in other disciplines. It has, in fact, received a great deal of discussion in the social science literature on group dynamics, though it does not go by the name of “phenomenology.” Consider Janis’s (1972) discussion of groupthink and social cohesion. Attempts to define social cohesion are almost entirely phenomenological in nature. They focus on the experience of group members—the sense of togetherness or we-ness. When agency is highly unified, there is a sense that “we did it,” and this has significant impact on the participants’ sense of their own agency—in some cases it produces detrimental effects, such as groupthink and what social psychologists call de-individuation, the sense that one has lost his own individuality. Contemporary social psychologists are also interested in the qualitative aspects of doing things with others, specifically, the concept of “entrainment”—feelings of interpersonal harmony. See, for instance, Goodman et al. 2005. 17. At the time of writing and submitting this paper for the volume Pacherie had not addressed the issue of how her account of intention could be extended to joint

Dynamic Theory of Shared Intention

31

action nor had she addressed the phenomenology of group agency. I am pleased to see that Pacherie has now turned to these issues. Pacherie (2012) does something similar to what I suggest here—extends her theory of levels of intention to shared intention and discusses the phenomenology of joint agency. She does not, however, focus on the sense of control (sense of joint control) but rather on the contrast between the sense of self in agency and the sense of joint agency and the ways in which different types of joint action can alter each of these. Her discussion of the various levels of shared intention is a welcome contribution to the joint action literature, and the discussion of the phenomenology of joint agency is certain to spark a great deal more interest in this phenomenon. 18. Versions of this chapter were presented at Southern Society of Philosophy and Psychology, March 20–22, 2008, New Orleans, LA; the American Philosophy Association, Central Division Meeting, symposium on Shared Intention, Chicago, April 18–21, 2007; and Illinois State University, March 23, 2007. I am grateful for the very helpful comments and criticisms I received from those audiences. I would also like to thank Rick Dale, Shaun Gallagher, Lucas Olsen, and two anonymous reviewers for their insightful comments and suggestions.

References Anscombe, G. E. M., 2000. Intention. Boston: Harvard University Press. Bayne, T., and N. Levy. 2006. “The Feeling of Doing: Deconstructing the Phenomenology of Agency.” In Disorders of Volition, edited by N. Sebanz and W. Prinz, 49–68. Cambridge: MIT Press. Bratman, M. 1987. Intention, Plans, and Practical Reason. Cambridge:  Harvard University Press. ———. 1993. “Shared Intention.” Ethics 104(1): 97–113. ———. 1999. Faces of Intention:  Selected Essays on Intention and Agency. Cambridge: Cambridge University Press. ———. 2004. “Shared Valuing and Frameworks for Practical Reasoning.” In Reason and Value: Themes From the Moral Philosophy of Joseph Raz, edited by S. Scheffler and M. Smith, 1–28. Oxford: Oxford University Press. ———. 2006. “Dynamics of Sociality.” Midwest Studies in Philosophy: Shared Intentions and Collective Responsibility 30(1): 1–15. Chersi, F. 2011. “Neural Mechanisms and Models Underlying Joint Action.” Experimental Brain Research 211(3–4): 643–53. Frankfurt, H. 1988. The Importance of What We Care About. Cambridge: Cambridge University Press. Gallese, V. 2001. “The ‘Shared’ Manifold Hypothesis: From Mirror Neurons to Empathy.” Journal of Consciousness Studies 8: 33–50. ———. 2009a. “Mirror Neurons.” In The Oxford Companion to Consciousness, edited by T. Baynes, A. Cleeremans, and P. Wilken, 445–46. Oxford: Oxford University Press. ———. 2009b. “Mirror Neurons, Embodied Simulation, and the Neural Basis of Social Identification.” Psychoanalytic Dialogues 19: 519–36. ———. 2009c. “We-ness. Embodied Simulation and Psychoanalysis. Reply to Commen­ taries.” Psychoanalytic Dialogues 19: 580–84.

32

C ollective Attitudes and A ctions

Gilbert, M. 1989. On Social Facts. London: Routledge. ———. 1996. “Concerning Sociality: The Plural Subject as Paradigm.” In The Mark of the Social, edited by J. Greenwood, ch. 1. Lanham, MD: Rowman & Littlefield. ———. 2003. “The Structure of the Social Atom: Joint Commitment and the Foundation of Human Social Behavior.” In Socializing Metaphysics, edited by F. Schmitt, ch. 2. Lanham, MD: Rowman & Littlefield. Gibson, J. 1979. The Ecological Approach to Visual Perception. Boston: Houghton Mifflin. Goodale, M. 1998. “Visuomotor Control: Where Does Vision End and Action Begin?” Current Biology, 8(14):, pp. R489–R491. Goodman, J. R. L., R. W. Isenhower, K. L. Marsh, R. C. Schmidt, and M. J. Richardson. 2005. “The Interpersonal Phase Entrainment of Rocking Chair Movements.” In Studies in Perception and Action VIII, edited by H. Heft and K. L. Marsh, 49–53. Mahah, NJ: Erlbaum. Horgan, T., J. Tienson, and G. Graham. 2003. “The Phenomenology of First-Person Agency. In The Metaphysics of Mind and Action, edited by S. Walter and H. D. Heckman, 323–40. Exeter, UK: Imprint Academic. Iacoboni, M., I. Molnar-Szakacs, V. Gallese, G. Buccino, J. C. Mazziotta, and G. Rizzolatti. 2005. “Grasping the Intentions of Others With One’s Own Mirror Neuron System.” PLoS Biology: 3(3):. e79. Jacob, P., and M. Jeannerod. 2003. Ways of Seeing:  The Scope and Limits of Visual Cognition. Oxford: Oxford University Press. Janis, I. 1972. Victims of Groupthink. Boston: Houghton Mifflin. Jeannerod, M. 1997. The Cognitive Neuroscience of Action. Oxford: Blackwell. ———. 2006. Motor Cognition:  What Actions Tell the Self. Oxford, UK:  Oxford University Press. Knoblich, G., and J. S. Jordan. 2003. “Action Coordination in Groups and Individuals:  Learning Anticipatory Control.” Journal of Experimental Psychology, Learning Memory, and Cognition 29: 1006–16. Krachun, C., M. Carpenter, J. Call, and M. Tomasello. 2010. “A New Change-of-Contents False Belief Test:  Children and Chimpanzees Compared.” International Journal of Comparative Psychology 23: 145–65. Kutz, C. 2002. “Acting Together.” Philosophy and Phenomenological Research 61(1): 1–31. Milner D., and M. A. Goodale. 1995. The Visual Brain in Action. Oxford:  Oxford University Press. Obhi, S. S., and N. Sebanz. 2011. “Moving Together:  Towards Understanding the Mechanisms of Joint Action.” Experimental Brain Research 211: 329–36. Pacherie, E. 2006. “Towards a Dynamic Theory of Intentions.” In Does Consciousness Cause Behavior? An Investigation of the Nature of Volition, edited by S. Pockett, W. P. Banks, and S. Gallagher, 145–67. Cambridge, MA: MIT Press. ———. 2007. “Sense of Control and Sense of Agency.” Psyche 13(1): 1–30 (special issue on the phenomenology of agency). ———. 2012. “The Phenomenology of Joint Action:  Self-Agency vs. Joint-Agency.” In Joint Attention:  New Developments, edited by A. Seemann, 343–89. Cambridge, MA: MIT Press. Pacherie, E., and J. Dokic. 2006. “From Mirror Neurons to Joint Actions.” Journal of Cognitive Systems Research 7: 101–12.

Dynamic Theory of Shared Intention

33

Richardson, M. J., K. L. Marsh, and R. M. Baron. 2007. “Judging and Actualizing Intrapersonal and Interpersonal Affordances.” Journal of Experimental Psychology: Human Perception and Performance. 33, 845-859. Richardson, M. J., K. L. Marsh, and R. C. Schmidt. 2005. “Effects of Visual and Verbal Interaction on Unintentional Interpersonal Coordination.” Journal of Experimental Psychology: Human Perception and Performance 31(1): 62–79. Schmid, H. B. 2009. Plural Action:  Essays in Philosophy and Social Science. Verlag: Springer. Searle, J. 1983. Intention: An Essay in the Philosophy of Mind. Cambridge, UK: Cambridge University Press. ———. 1990. “Collective Intentions and Actions.” In Intentions in Communication, edited by P. Cohen, J. Morgan, and M. E. Pollack, ch. 19. Cambridge, MA: Bradford Books, MIT Press. ———. 1995. The Construction of Social Reality. New York Free Press. Sebanz, N., H. Bekkering, and G. Knoblich. 2006. “Joint Action:  Bodies and Minds Moving Together.” Trends in Cognitive Sciences 10(2): 70–76. Smeets, J., and E. Brenner. 2001. “Action Beyond Our Grasp.” Trends in Cognitive Sciences 5(7): 287. Tollefsen, D. 2002. “Organizations as True Believers.” Journal of Social Philosophy 33(3): 395–411. Tollefsen, D. 2005. “Let’s Pretend! Joint Action and Young Children.” Philosophy of the Social Sciences 35: 75–97. Tomasello, M., and M. Carpenter. 2007. “Shared Intentionality.” Developmental Science 10: 121–25. Tomasello, M., M. Carpenter, J. Call, T. Behne, and H. Moll. 2005. “Understanding and Sharing Intentions: The Origins of Cultural Cognition.” Behavioral and Brain Sciences 28(5): 675–91. Tuomela, R. 2000. Cooperation:  A  Philosophical Study. Dordrecht:  Kluwer Academic Publishers. ———. 2002. The Philosophy of Social Practices. Cambridge, UK:  Cambridge University Press. ———. 2005. “We-Intention Revisited.” Philosophical Studies 12(3): 327–69. Wolpert D. M., and M. Kawato. 1998. “Multiple Paired Forward and Inverse Models for Motor Control.” Neural Networks 11: 1317–29.

2

COLLECTIVE GOALS ANALYZED K a a r l o M i l l e r a n d Ra i m o T uo m e l a

2.1  Introduction: What Are Collective Goals? The notion of collective goal has been used in various senses. Goals, collective or not, can in general be regarded as analogous to wants or desires on the one hand, and to intentions on the other. The basic form of goal ascription is the following:  agent x’s goal is that P (e.g., to see to it that a certain state of affairs will come about). Here, “goal” can sometimes be replaced by “want” and sometimes by “intention.” Intentions characteristically have actions in their contents. Following the mainstream in action theory and artificial intelligence studies (AI), we take intention contents and goal contents to be propositional. Accordingly, the content of an intention is not an action type as such; the content is that the action is performed by the intending agent. Consider, for example, x’s goal that the window will be closed. As such, it can be the want of an agent or, in the case of intentions, the wanted state that the window is closed is also a result of the agent’s intended action. In the intention case, goal satisfaction necessarily involves the agent’s action, that is, his seeing to it that the window is closed.1 In this chapter, we will proceed from collective want goals to goals that are equivalent to full-blown collective intentions. Collective goal is here understood as a kind of umbrella notion, applicable in various fields of political and social philosophy as well as in economics and other social sciences. We emphasize that collective goals must be collectively accepted and will give an analysis of this. Due to this emphasis, our account is related to contractualist ideas in social and political philosophy. Jean-Jacques Rousseau distinguished between private will, the will of all, and general will.

Collective Goals Analyzed

35

General will represents the true interests that everyone has as a member of a civil society. Our account of collective goal has the following features in common with Rousseau’s notion of general will. First, in accordance with Rousseau, we take a collective goal to be shared, at least to a significant degree, by the members, and not only “accidentally” shared, but shared in their “roles” as group members. Second, analogous to Rousseau’s notion of general will that represents the “true interests” of the members as citizens, a collective goal represents “group interest” or group preference. In The Social Contract, Rousseau also writes that every individual can personally have a particular or private will that contradicts or differs from “the general will which he has as a citizen” (Rousseau 1762, book I, ch. 7). Third, in accordance with Rousseau, our notion of collective goal is compatible with the group members having divergent and even contradictory private goals. After Rousseau, in more recent social and political philosophy, for example, John Rawls emphasizes that citizens share collective goals: “in the well-ordered society of justice as fairness citizens do have final ends in common” (Rawls 1993, 202). According to Rawls, the citizens not only share more or less contingent ends, in an ideal society where justice as fairness prevails, but they also “share one very basic political end,”. . . “namely the end of supporting just institutions and of giving one another justice accordingly” (202). Another example is given by Raimo Tuomela’s theory of sociality in which the notion of group ethos is central (Tuomela 2007, ch. 1). As the ethos has the world-to-mind direction of fit, it can be regarded as a group goal and hence a collective goal in a broad sense. Here, we need not presume any kind of general super-goal representing the true interests of members as citizens, as Rousseau requires of the general will, and we need not legislate that the members ought to form and then conform to such a goal. Our account of collective goal does not require that “one very basic political end” is shared in something like a Rawlsian fashion either. Yet we accept that there are hierarchies of goals. Our policy is first to sort out different types of (collective) goal and next, given such a goal, to study what “subgoals,” if any, the members ought to form. We will proceed by giving simple examples in which the constituents of collective goals can be clearly distinguished. The conceptual machinery can then be applied to cover more complex cases. Here we arrive at the commitments and obligations that typically ensue from the collective acceptance of a goal. Besides political and social philosophy, collective goals are central in economics and other social sciences. For example, the achievement of a collective goal often amounts to the satisfaction of what the economists call a public or common good.2 Collective goals have interesting applications in game theory, as well. From the group’s point of view, the obtaining of the collective goal content is to be preferred

36

C ollective Attitudes and A ctions

to its not obtaining: the group should not choose goals it prefers not to obtain. This can be considered a collective rationality requirement, even in cases where it would be rational not to prefer it from an individual point of view. Accordingly, in the case of a collectively rational group, satisfaction of the collective goal entails satisfaction of the group-preferred alternative, all things considered. Thereby, with the introduction of group or team preferences, standard game theory can be extended to cover team reasoning (cf. Bacharach 1999) or we-reasoning (for an argument and discussion, see Hakli, Miller, and Tuomela 2010). Social institutions and roles form a central field of research in sociology and other social sciences. Although we cannot go into any detail here, the following argument can be outlined to demonstrate the relevance of collective goals for this research. First, it can be assumed that all institutional role-actions are actions performed as members of a group. Second, genuine acting as a member of a group is, in one way or another, contributing to the satisfaction of a collective goal. Consequently, also institutional acting (acting in a role) is acting for a collective goal. A link between collective goal and the sociologists’ notions of social institution and role can accordingly be established with the help of the notion of acting as a group member (for this notion, see Tuomela [2007, 22–24]). In section 2.2, we discuss and distinguish various types of collective goals, differing in strength. Presently, it suffices to assume that institutional acting is acting to achieve or uphold some kind of collective goal. It should be noted that the agent can act in his role for various motives; his decisive reasons for contributing to the satisfaction of a collective goal can be private or even egoistic so that the satisfaction of the collective goal may not be why he acts. Yet a basic underlying collectively accepted goal is required also in this case as it sets the limits of how he is to act if he is to satisfy his institutional role. Finally, especially in the case of a foreign culture or society, awareness of the collective goal or goals is also required for an outside observer to understand or interpret “what is going on there.” Accordingly, the notion of collective goal can have a significant methodological role in social and cultural anthropology, and also these areas of research can benefit from philosophical analysis of collective goals. Not only are collective goals relevant for the philosophy of sociality and the social sciences, but they also have an important twofold function in everyday social life: a collectively accepted goal serves to coordinate the entire collective action aiming at the satisfaction of the goal, and it guides the individual agent in performing his part. In this chapter, collective goals are approached along two dimensions. We proceed from mere want-like goals to intended goals. The other and more central distinction is between private and collective goals:  we will also proceed along the private-collective dimension and end up with what will be called

Collective Goals Analyzed

37

full-blown intended collective goals. To this end, we will first present some relevant principles for the classification of goals. We will show how a given specific goal state,3 when various additional predicates are ascribed to it, may yield different goals of increasing complexity. In the literature related to our topic, one can find detailed analyses of collective intentions, beliefs, commitments, acceptances, agreements, obligations and rights, preferences, and interests. An analysis of a general collective pro-attitude covering these notions would be welcome. Consider, for example, our collective goal to get the house painted. This goal involves intentions—typically we-intentions of the participants—and accompanying beliefs and commitments. Normally, in the case of a “we-mode” goal (to be discussed later), collective acceptance or agreement is also involved. This collective acceptance typically brings along obligations and rights for the members, and its content can be taken to represent their common interest or group preference. The purpose of this chapter is to contribute to a general analytic account of collective goals, which, though not reducible to any account of the above notions, at least in some form, involves them all. The focus of this chapter is to find answers to the question: “what makes a goal collective?” Our classifications and analyses serve their purpose if they can give answers to this question. 2.2  Classification of Goals In this chapter, goals are analyzed as pro-attitudes, the content of which describes what the agent wants, and we will concentrate on intended goals. Both action goals and state goals will be discussed, although we employ the latter as paradigm cases, and the actions of agents can be considered as means to achieve the goal states (cf. endnote 3). Our focus is not only on intended goals but also on collective goals, as opposed to private ones. We now give stylized definitions of seven new notions relevant to collective goals4 and first discuss the first three in terms of a concrete example. By applying and combining the distinctions to be made, one can study various goals that can be justifiably called collective in some sense. (a) Goal P of an agent x is shared if and only if there is at least one other agent who has goal P and they believe that they both have it.5 (b) Goal P of an agent x is dividable if and only if there will, or at least can, be parts or shares for at least one other agent to bring about (or sustain) P. (c) Goal P of an agent x has collective content if and only if, necessarily, P is satisfied for x if and only if it is satisfied for other group members sharing P.

38

C ollective Attitudes and A ctions

Suppose that our group, consisting of you and me, collectively accepts the goal of getting the house painted. In that case, we share the goal and believe so.6 There can, but need not, be specific parts for both of us. For example, if your task is to paint the front of the house and mine the back, then our goal is also dividable. Finally, the goal has collective content because, due to our collective acceptance, neither of us can satisfy it for herself only. The properties (a)–(c) can be combined so that, for example, shared goals with collective content will result. Case (a) represents shared goals.7 If a goal is shared in the sense of (a), the goal contents are the same and the agents believe so. (If this latter requirement would be dropped, we obtain goals that are “shared” in ignorance.) These independent “goals of all” are typically private and less applicable in accounting for the collectivity of goals. They correspond to what Rousseau called “the will of all” in contradistinction to general will. Case (c) gives the minimal requirement that we consider sufficient for the collectivity of x’s goal content, and it is necessary in the sense that also the stronger notions presuppose its truth. It should be noted that goals with collective content and collective goals are not equivalent. A goal’s being collective entails its having a collective content in the sense of (c), but not conversely. In case (c), the collectivity of a goal is analyzed in terms of the collectivity of its content, which calls for a more specific analysis of collective content. This we will do in section 2.3 in terms of the satisfaction conditions of the collective goal. Besides having collective content, x’s collective goal often has a collective origin that can be revealed by tracking the “history” of the collective goal. The collective goal might have been the result of, and engendered by, an action of collective decision or acceptance. Before going on with our analysis of collective content, two things should be noted. First, it is not presupposed that what the pro-attitude “is about” will in fact ever exist or be true. This basic idea has taken various forms in the literature and has been applied to the contents of, for example, intentions, expectations, and wants. As having a goal is taken to be a pro-attitude, we also accept the idea for the contents of goals. Second, the content of any pro-attitude, here the content of a collective goal, can at least partly be characterized in terms of its conditions of satisfaction. These conditions describe roughly, what it would be like if the pro-attitude (goal) were satisfied. Accordingly, these conditions characterize the content of the goal. Therefore, it is an appropriate task to specify the necessary and sufficient conditions for satisfying a collective goal content. By tabulating types (a), (b), and (c), we obtain eight new types of goals. The first goal is one that is shared, divisible, and has collective content. This is “the most collective” goal based on these predicates. An agreement-based goal to write a joint paper could be an example of this case. The eighth goal is one that

Collective Goals Analyzed

39

is unshared, not divisible, and has private content. This goal cannot be collective in any sense. A person’s present goal to make a sandwich is an example of this case. Between these two goals, there are six other cases of differing “degrees” of collectivity. We will not discuss these any further here, in part because these are not the only features that account for the collectivity of goals. Instead, let us turn to the most central and developed notion of collective goal: we-mode goal. (d) Goal P of an agent x is in the we-mode relative to group g (or x has P in the we-mode) if and only if (i) x is functioning qua member of g, (ii) goal P has been collectively accepted or at least is based on what the members collectively accept and, consequently, (iii) it satisfies condition (c), (iv) it serves as a group reason for the members, (v) x and the other members are collectively committed to satisfying P, (vi) P satisfies the collectivity condition, and (vii) x intends or wants to participate in the satisfaction of P for g.8 In the stylized we-mode case (d), the satisfaction of the goal P is “for g.” P’s satisfaction being for the group involves that the group is taken to “own” its activities, and thus it is often able to make either itself or some other party the beneficiary. (In some contexts, it is thus appropriate to speak of selfishly versus altruistically functioning groups.) It is important to distinguish the aspect of who “owns” the activity or work involved in the satisfaction of P from the aspect of being a beneficiary. To “own” the satisfaction activity entails that the group can choose whether P comes about due to its action and normally also the right to choose the party that will be the intended beneficiary, be it the group itself, some group member(s), or another group or person. In (d), the “qua-member” relation is applied. We could say that if a person acts because of his we-mode goal, he acts qua group member. Given that there is a part for each agent to perform, the we-mode entails that they intend to participate in satisfying the goal together with the others. The participation can be direct or indirect. In the latter case, the member still participates in some “indirect” way to the group’s seeing to it that the goal will be satisfied, for instance, by taking part in the group’s discussion of who should be directly involved or by standing in reserve for direct participation. And if a participant then acts qua member, he does not independently happen to perform a part, but he performs his part as his part of the joint goal. Because the we-mode also entails that the participants are collectively committed to what they have accepted, it has a guiding and controlling function. If they thus form a goal for the group, the existence of that group goal will be their authoritative group reason for action as group members. There is group authority in the present case primarily because

40

C ollective Attitudes and A ctions

the participants are functioning as a group—and this is out in the open. They must give up part of their “original” authority (that they have in the noncollective case) to the group. Their collective commitment to P elucidates the role of this requirement. So even in most simple we-mode cases, group authority and group reason are involved. Note, too, that mainly because of the authoritative group reason and collective commitment to the group (its main goals, beliefs, and standards, etc.), the group is at least in part responsible for its members’ intentional actions qua group members.9 Basically because of being collectively accepted by the group members qua group members, the we-mode goal (d)  satisfies the three central criteria for we-mode activities: group reason, collectivity condition, and collective commitment.10 Let us apply these three we-mode criteria to our concrete example. Suppose that our group, consisting of you and me, collectively accepts as its goal to get the house painted. When I have goal P in the we-mode, my group reason is not the mere content P of our acceptance, but the reason also involves the fact of our acceptance of it. To put this somewhat more “externalistically,” the reason here can be taken to be P qua collectively accepted to be our goal. In our example, the reason is to get our house painted by us—and this goal spelled out as “the house gets painted with the painting done jointly and intentionally by us” intrinsically requires our collective acceptance of this goal as our goal. Our collectively accepted goal to get the house painted provides me with a reason, indeed a group reason, that I would not have otherwise. In this case, I intend to paint my share at least in part because of this group reason. Accordingly, in the first person case, my group reason is not merely what we accept as our goal but in part is also the very social fact that we accept that content as our goal. So long as the group members are collectively committed to achieving the goal, they take the group reason as authoritative, hence generally preemptive. When collectively accepting the goal, the group members voluntarily give some of their authority to the group. According to the collectivity condition, necessarily, the goal is satisfied for a member if and only if it is satisfied for the other members. Finally, because of being collectively committed to achieving the goal, the group members bind themselves to the goal, and a member cannot unilaterally renounce the commitment. The social fact that we accept P as our goal is an authoritative reason for both of us to intend and act accordingly. We can achieve our goal in various ways. For example, we can paint the house together, I or you can paint it alone, we can hire someone to do it, you can provide the paint and I can pay for the job, and so on. The collectively accepted goal provides a group reason for each of us to do our share as agreed. In addition to the existence of this group reason, there is also collective commitment to act for this reason, viz. we have voluntarily committed ourselves to seeing to it that the house gets painted as agreed. This

Collective Goals Analyzed

41

collective commitment also involves an “achievement goal” for both parties to see to it that mutual awareness of the obtaining of the goal will eventually be achieved or, if it turns out that it cannot be achieved, to see to it that this will become mutually known. According to the collectivity condition, if the goal is satisfied—that is, if the house is painted as agreed—it is necessarily satisfied for both of us. Even though it would have been my own private house that got painted, I cannot satisfy the collective goal for myself only. It is useful to contrast we-mode goals with I-mode goals, as they have different functional roles in thinking and acting concerning how to satisfy the goal as well as typically different commitments to the goal and to other group members. The following I-mode goals can now be distinguished: (e) Goal P of an agent x is in the I-mode (or x has P in the I-mode) if and only if x is privately committed to achieving P and he intends or wants to satisfy P at least in part for himself qua private person This general notion of an I-mode goal is independent of whether x functions purely privately or as a member of a group. The beneficiary of the satisfaction of P can in the present case be the agent x himself, his group, or some other party. We will here distinguish between two central subclasses of the I-mode, the first of which assumes that the agent is not functioning as a member of any group but intends the satisfaction only for himself: (f) Goal P of an agent x is in the private I-mode (or x has P privately) if and only if x is privately committed to satisfying P and he intends or wants to satisfy P only for himself qua private person. (g) Goal P of an agent x is in the pro-group I-mode (or x has P in the pro-group I-mode) relative to group g if and only if x is functioning qua member of group g, x is privately committed to satisfying P, and he intends or wants to satisfy (or participate in the satisfaction of) P for the group members sharing P.11 In the pro-group I-mode case, the group members share their private goal P with the other group members. That is why it is appropriate to say that P is a goal for the individual group members, x, and the others. Given that there is a goal P so that the criteria of the we-mode goal in (d) are satisfied, then P is available for the group, and each member can correctly use locutions like “it is our goal that P.”12 Given that there is this goal so that (e) or (f) is satisfied, agent x can correctly use locutions like “it is my goal that P,” and when (f) is satisfied, x cannot correctly express himself by “it is our goal that P.” Accordingly, every purely privately held goal is also an I-mode goal, but not

42

C ollective Attitudes and A ctions

conversely. On the other hand, a person can have goal P simultaneously both in the I-mode and in the we-mode. However, x’s we-mode goal entails neither that he has nor that he does not have the corresponding I-mode goal. Note that x cannot have goal P simultaneously in the we-mode and in the private I-mode. What are the basic differences between we-mode, case (d), and pro-group I-mode, case (g)? A pro-group I-mode goal need not satisfy any of the three criteria of the we-mode: group reason, collectivity condition, and collective commitment. First, although the agent who has goal P in the pro-group I-mode intends to contribute to the satisfaction of P for the group, she need not intend to act because of the group reason, viz. because of their collective acceptance of goal P. Second, in the absence of collective acceptance, the pro-group I-mode need not satisfy the collectivity condition. Third, the member acting in the pro-group I-mode (e.g., a group “benefactor”) need not be a participant in any collective commitment to achieve P. The difference between the we-mode and pro-group I-mode is evident also in cases where there is no fully binding collective acceptance and collective commitment to a goal. Here, the difference is in reasoning strategies of the two modes. In game-theoretic terms, in the case of a rational group, the goal that is collectively accepted to be realized can be taken to maximize (expected) group utility. It can also be shown that we-mode reasoning in many cases yields better results for the group and for the members than pro-group I-mode reasoning would yield. However, we cannot here go into a discussion of game theory.13 The above cases, (a)–(g), deal with goals from the viewpoint of an individual group member. Let us now apply the notions of we-mode and I-mode to shared and joint goals of group members. When applied to a concrete example, we get the following cases of intended goals: (h) We-mode joint goal to be satisfied by joint action: we jointly intend to paint the house together. (i) We-mode joint goal to be satisfied by single agent action: we jointly intend to see to it that agent x paints the house. (j) I-mode shared goal to be satisfied by I-mode joint action: we share the goal to paint the house together in the I-mode. (k) I-mode shared goal to be satisfied by single agent action: we share the goal to see to it that agent x paints the house. Cases (h) and (i) satisfy the criteria of we-mode (cf. (d) above): group reason, collectivity condition, and collective commitment. These three criteria are satisfied because of collective acceptance. Accordingly, the source of we-mode is in the collective acceptance of the goal, not in joint action to satisfy the goal. It is

Collective Goals Analyzed

43

not essential to we-mode goals that they be satisfied by joint action of the members. We-mode joint goals can be and often are satisfied by single-agent action. The shared I-mode goals (j) and (k) do not involve collective acceptance, and consequently they are only weakly collective. Let us end this section by introducing three importantly different notions of a collective goal, of which the second will be further analyzed and discussed in the next section: (1) collective goal based on shared “we-wants” (or “we-goal”) (2) (intended) collective goal (3) a group’s (intended) goal The first is the weakest notion, in which the central connecting “social glue” is the participants’ mutual belief. It does not even require that the members intend to achieve the goal. Wants are intentional in the “aboutness” sense, but one can want something without intending to achieve it. Thus, some people may have as their goal to attend a certain opera performance, believing that the others in the group also have that goal and also believing that this is mutually believed in the group. This goal can be merely want-based, but it can also become intended. The next notion in strength is the intended collective goal. It involves the participants’ intention to achieve the goal, requiring collective activity. Some people can have as their intended collective goal to see to it collectively that an old historically valuable building is restored. An intended collective goal in this sense is based on the goal-holding agents’ intentions to satisfy the content of the intention by acting together. The “condition of collectivity” (basically condition (c)) applies to intended collective goals (see the discussion in the next section). Our third collective goal notion is group’s intended goal to perform an action. In this case, the goal is not ascribed to individual members but to the group itself. For example, a state can have as its goal to conquer a certain territory. Basically, at the level of the members of the group, this can involve at least two quite different things, indicating two different senses in which a group can intentionally have a goal. First—and this is the “normative” or group-binding sense—the group may have a decision-making system by virtue of which the goals are jointly agreed upon for the group. Second, a group can be said to have an intended goal to achieve something if its members—or a majority of them— share at least a weak we-goal in the sense of our category (1) above.14 How is the group’s goal related to the collective goals of its members? These notions obviously are not independent. This dependence can be explicated in terms of a kind of supervenience relation: every change in the group’s goal (in the full, we-mode sense) is necessarily accompanied by some change in the

44

C ollective Attitudes and A ctions

collective we-mode goals of the members.15 However, the member’s shared we-mode goal need not depend on their private goals. For instance, a group (e.g., a group of extremists) could have a goal P (e.g., to destroy capitalism), and the members could have P as their derived we-mode collective goal but yet not have P as their private goal (or in an extreme case, even have not-P as their private goal). Supervenience with respect to private goals thus fails. However, note that the above point about supervenience concerned we-mode group goals. One may speak of group goals in a weaker, I-mode sense: Group G has the goal P to the extent its members share the “we-goal” to achieve P. A we-goal in the I-mode sense is defined in a stylized way as follows: a member has the we-goal P if and only if she has goal P, and believes that the others in the group (or most of them) have P and that this is mutually believed in the group. When the we-goal P is shared by all or most of the members the group, we can speak of a group goal: to the extent P is shared, the group has the goal P (in the I-mode). Supervenience with respect to I-mode goals of members then holds.16 The notion of “collective intention” is sometimes used to refer to an intention ascribed to a group or a collective, and sometimes to an intention ascribed to group members.17 There is a similar ambiguity concerning “collective goal.” It is sometimes used to refer to group’s goal and sometimes to collective goals of its members. The purpose of the above discussion was to disambiguate this distinction and to clarify the relations between these notions. Here we concentrate on group members’ collective goals, and we next discuss intended collective goals of the members. 2.3 The Collectivity Condition for Collective Goals Let us next discuss intended collective goals. Consider the shared goal of reducing the amount of greenhouse gases in the atmosphere. Most of the people who share the goal have never met and may not plan on getting together for joint action to achieve the goal. Yet they have a belief that many others share the goal. Indeed, there is a mutual belief about this at least in the sense that they believe that the others believe that many others have that goal. The goal holders must be prepared to somehow contribute to their goal, if they genuinely hold it, active endorsement of the goal may be enough in some cases. At least they are not supposed to contribute to the negation of the accepted goal, and if there is no active part for a member to perform, he is supposed to stand in reserve should it turn out that his active contribution is needed. One central aspect of intended collective goals is their satisfaction conditions. Here, the collectivity condition comes into play. What is this condition? While the idea seems to apply to all kinds of collectivity, our present version concerns

Collective Goals Analyzed

45

collective goals: if one or more agents satisfy the intention content, then it is on noncontingent, “quasi-conceptual” grounds satisfied for all participants. That is, it is necessarily satisfied because of the participants’ reflexive acceptance of the goal as their collective goal. The individual agents cannot normally intend the content in question in the standard “action-sense” of the notion of intention (requiring basically that they believe that they, by their own action, can satisfy the content). They can, however, have the content of the collective intention as their goal, and they can aim at its satisfaction and be in a collective sense committed to it, their action-commitment being to their own contributions to the goal. Accordingly, an intended collective goal in this sense is a goal in the we-mode. Consider the goal expression, “Our goal is to visit Naples.” Under the I-mode interpretation, my goal is to visit Naples, and similarly for you. My goal, of course, is satisfied (for me) when I have visited Naples, independently of what you do or have done. In contrast, in the we-mode, my we-mode goal to visit Naples is not satisfied merely by my visiting Naples. It is a necessary condition for my we-mode goal to become satisfied that your goal become satisfied, too. In accordance with our collective acceptance, my we-mode goal is satisfied only if we both visit Naples. Let us formulate the above collectivity condition as follows for further reference:18 (CC) It is necessarily true (based on the group members’ collective acceptance of P as group G’s goal) that P is satisfied for a member x of G (qua member of G) if and only if it is satisfied for every (other) member of G (qua member of G). Below we will discuss the collectivity condition as a criterion of we-mode goals. We argue that if P is a we-mode collective goal, then some version of (CC) must be true of it. (CC) concerns unstructured groups where the having and the satisfaction of the goal distributes to the members.19 As to the pro-group I-mode case, it does not satisfy (CC) because it is not based on collective acceptance for the group, but only on shared private acceptance and private commitments. Under this interpretation, in the above example, we share the goal that we both visit Naples. This goal can only be satisfied by both you and me visiting Naples. Yet, the pro-group I-mode is not based on the participants’ agreement to satisfy the goal as a group, and hence (CC) fails to be satisfied. If the participants indeed had made an agreement that commits them to the satisfaction of the goal as a group, they would have operated in the we-mode rather than in the pro-group I-mode. We noted that the notion of truth on quasi-conceptual grounds in the collectivity condition means this: because participants have collectively accepted the goal as their collective goal, the goal has the conceptually simultaneous satisfaction as its feature. The collective acceptance can vary in strength, so to speak,

46

C ollective Attitudes and A ctions

and range from joint, plan-based acceptance to shared, “acceptance-belief ”. 20 Figuratively speaking, the “stronger” the collective acceptance, the stronger the necessity. The agents need not even have beliefs about (CC)—the connection can obtain in a roundabout way due to their de re beliefs that they are engaged in the same “project.” If the collectivity condition were satisfied on contingent grounds only, P would not be a collective goal but only a shared private goal. Note that for a goal to be satisfied for a participant in a full subjective sense, we must require more than the objective occurrence of the goal state due to the participants’ action. Analogously with the case of intentional action, we must require that the participant believes that the goal has thus been satisfied and, given (CC), also believes that it has similarly been satisfied for the others. 2.4  Collective Acceptance, Mutual Belief, and the Collectivity Condition The we-mode approach basically relies on group acceptance (thus group members’ collective acceptance) of appropriate contents. Accordingly, the collectivity condition discussed in the preceding section also relies on collective acceptance. Therefore, we have to briefly discuss why this notion is introduced and what basic functional roles it has. We concentrate on collective acceptance of state goals like “G’s goal is that P,” where P is a state-expressing proposition.21 If group G accepts P as its goal, then G has goal P. If G does not accept goal P, then G cannot have goal P as a group. Suppose that group G would not accept P as its goal. Then the members would not have this (group) reason to act to achieve P. They might still intend to act, perhaps on the assumption that there will be sufficiently many members who act likewise, but, lacking collective commitment to achieve P, it is more likely that they are willing to give up their attempts. Therefore, in many cases, without collective acceptance, they would lack the required persistence, and P would remain unachieved. Even if they would persist with their intentions, acting in concert would be more difficult without collective acceptance. Accordingly, collective acceptance both strengthens their motivation and facilitates coordination. Elsewhere, Tuomela (2007, 2013a) has discussed group acceptance and collective acceptance at length, and we will here only present his final characterization and connect it to collectivity.22 In this account, the idea is to analyze a group G’s acceptance proposition p as true in terms of its commitment, of its having bound itself to see to it or to act on it that p is “premisible” (usable as a premise in inference and action) in group contexts with the right attitude-dependent direction of semantic fit. The proposition p could be, for example, “we will perform x jointly,” “P is our, viz. G’s, goal, “grass is green,” or “squirrel pelt is money.” Here, G’s acceptance of p is acceptance of p for the

Collective Goals Analyzed

47

group; thus, it is the “group-perspectival” truth of p that is involved (“G takes p to be true or correct”). The group is here for simplicity’s sake assumed to be an unstructured (“egalitarian”) we-mode group and thus treatable as an agent and capable of action. In this case, group acceptance ideally amounts to its members’ collective acceptance. Basically, group G qua group accepts p as true for G if and only if the members of G qua members collectively accept p as true for G. This statement can be analyzed as follows in terms of intentional “seeing to it that” (stit) action: (GA) Group G qua group accepts p as true for G (i.e., G accepts that its goal is P) if and only if the members come to share an attitude (explicitly or implicitly expressed by p) that has the world-to-mind direction of fit (one in the intention family of attitudes) or the mind-to-world direction of fit (one in the belief family), and where sharing the attitude entails the satisfaction of the relevant collectivity condition, the members being collectively committed to seeing to it that what p expresses is satisfied (in the world-to-mind direction of fit case) or true (in the mind-to-world direction of fit case), in conditions of shared we-knowledge of what the direction of fit of p is.23 The central aspect of the shared attitude (the “acceptance state”) here is its direction of fit rather than its being of a certain specific mental kind: the members of G are assumed to be collectively committed to seeing to it in the right way (namely, in terms of action obeying the direction of fit associated with p on that occasion) that p is true or premisible, given the mutual understanding and knowledge about p and its direction of fit here. When applied to goals, (GA) deals with the case in which p = G’s goal is that P, and then the analysandum becomes: group G qua group accepts that G’s goal is that P. Primarily, when accepting the goal that P, group members accept to see to it that the goal content, here P, will be true. As a consequence, they (cannot but) accept that (it is true that) G’s goal is that P. Thereby, when collectively accepting the goal that P, group members are collectively committed to seeing to it that P will be true. Accordingly, if G takes P to be true, G is committed to keeping it as it is, and if G takes P to be false, G is committed to changing the world to make P true. Collective acceptance is basically collective action, action resulting in collective commitment to what is accepted. Let CA(G,P) represent G’s acceptance of the goal P for itself. It entails that the members accept to see to it that P, and, accordingly, they are collectively committed to seeing to it that P. CA(G,P) represents external activity (“event” or “process” that can be described in overt terms). Below we will assume that CA(G,P) entails that the group members create a group goal P.24 In the present sense, the collective acceptance of a goal for

48

C ollective Attitudes and A ctions

the group is an umbrella term covering explicit and implicit agreements such as, for example, signing explicit agreements, promising and accepting promises, spontaneous suggestions and reactions to these nonverbal gestures and nods of approval, and so forth. From a linguistic and conceptual point of view, we can here speak of performatives, and now we can explicate the above theses in terms of them. Before the shared collective goal P emerges, interaction (typically linguistic) normally takes place between the members. The participants construct collective goals for the group by their conceptual activities and by their collective acceptances and allowances.25 Here, we call this activity the “collective acceptance of P.” The fact that the agents have the collective goal P is the necessarily inbuilt result (in the standard sense of action theory) of this collective approval: if the members did not come to have P as their collective goal, it would be false to say that they collectively accepted P as their collective goal. Thus the following is true: (1) CA(G,P) ->N Goal(G,P)

Also, the converse implication holds true for collective goals, although we will not discuss the matter here. What all cases of agreements have in common is that, if they are to count as a collective acceptance of a collective goal P, the members must have the collective goal P as a result. Because collective acceptance produces the members’ collective goals as its result, it also produces the conative element concerning the goal content. Having any goal implies having a “pro-attitude” toward its content:  the goal holder prefers the truth of the goal content to the truth of its negation. Thus, although collective acceptance was treated above as a performative and not an attitude, it gives rise to a shared pro-attitude and collective commitment toward the content of what has been accepted. Next, consider the following doxastic thesis where “MBg” stands for mutual belief (e.g., in the standard iterative sense) in G: (2) CA(G,P) ->N MBg(CA(G,P))

According to (2), although CA(G,P) stands for collective external activity, it implies mutual awareness of itself.26 MB need not be taken to stand for an infinitely iterative belief, but, depending on the case at hand, a suitable small number of iterations will suffice.27 In any case, the members not just happen to share the belief that they have reached the agreement, but there is also a mutual consensus in the group to this effect, viz. the parties of the agreement are aware that it is mutually understood that an agreement has been made. We can argue for (2) as follows. First, a group

Collective Goals Analyzed

49

capable of action can be regarded as an agent, and in analogy with individual agents, the group members functioning as a group must mutually believe what they collectively accept (cf. the individual case where, e.g., I accept something as true and must also believe what I have accepted on pain of not understanding what accepting means). Second, the group members must be able to act on what they have accepted—collective acceptances can ground their actions as group members. This obviously requires the kind of social awareness that mutual belief involves.28 It is not required that CA represents a collectively intentional action: the collective agreement need not be based on a joint intention, and it can be reached in various ways. It is, however, required for the truth of (2) that the individual parties do not enter the agreement unintentionally. Their having accepted a collective goal should not come as a surprise to them, and this gives a third, related argument for (2). Furthermore, collective acceptance can in some special cases be just collective acceptance-in-action, thus not premeditated. Note that individual performances by each party, however, are not required for collective acceptance. It suffices that the parties in the agreement are aware of what is going on and do not refuse to adopt the possible obligations ensuing from it. Because the group’s acting on its collective acceptance of a goal, in analogy with the single-agent case, clearly depends on the members’ collective awareness of the goal, (2) seems plausible, and we regard it as tenable. The mutual awareness, here MB, is allowed to be of varying degrees, but at least some degree of reflexive consciousness, over and above the mere shared belief, is required. Therefore, although collective acceptance is above treated as a performative and not just an attitude, it gives rise to a collective epistemic attitude. Collective acceptance thus gives rise to both a conative attitude concerning the content of what has been accepted (i.e., P) and an epistemic attitude concerning the fact that they have accepted P as their collective goal (i.e., CA(G,P)). Unlike the conative attitude, the collective epistemic attitude (here mutual belief) does not concern the content of what has been agreed upon. That the members also mutually believe that the goal content will become true is considered to be the “default,” but this (mutual) belief is not implied by the fact that the members have accepted the goal. Because the content of the epistemic and the conative attitude is not the same, “split-mind” cases—the agent believing that it is not likely that the goal content will become true and yet in another “mode” believes that it will—are avoided. Given that P has been collectively accepted and x shares the collective goal that eventually P will occur, his main doxastic premise in his practical reasoning is not his belief that eventually P will occur, but his belief that they have accepted P as their goal and that this is mutually believed in his group, that is, the following is valid for all members x of G:

50

C ollective Attitudes and A ctions

(3) CA(G,P) ->N Bx((CA(G,P)) & MBg(CA(G,P)))29

The consequent of (3) is x’s ordinary factual “we-belief ” having the fact that P has been collectively accepted for x’s group and the fact that this is mutually believed as its content. Speaking of collective acceptance involving agreement making, as x believes that they have collectively accepted P, he is also aware of the duty-right bonds between the members as specified and established by the concrete “deal” that has been agreed upon, and he believes that there is a mutual belief to this effect. Thus, for example, because of the existence of the agreement, he is aware of what, if anything, he himself ought to do, and he is aware that this is mutually believed. Let us now return to the collectivity condition, (CC), and add to it the mutual belief condition: (CC) It is necessarily true that P is satisfied for a member xi of G (qua member of G) if and only if it is satisfied for every (other) member of G (qua member of G); and this is mutually believed in G. Consider the following aspects of collectivity, formally rendered as follows with x ranging over the members of G: (CCa) CA(G,P) ->N (¬(∃x)Sat(x,P) ∨ (x)Sat(x,P)) (CCb) CA(G,P) ->N MBg(¬(∃x)Sat(x,P) ∨ (x)Sat(x,P))

What (CCa) amounts to is this:  given that the members have collectively accepted P as their goal, it is true that P either is satisfied for no member or it is satisfied for every member, and according to (CCb), the members mutually believe that P either is satisfied for no member or it is satisfied for every member. Together (CCa) and (CCb) cover the central content of what the collectivity condition intuitively can be taken to contain. Summing up some of the central features of the discussion of collective acceptance of goals when understood as a performative: Collective acceptance of goals

(i) necessarily results in shared collective goals of the members, (ii) implies mutual awareness of itself, over and above mere shared belief, at least to a significant degree, (iii) can be applied to account for the collectivity of goal contents, and (iv) involves normative or quasi-normative collective commitment to what has been accepted.

Collective Goals Analyzed

51

We argued above for these theses for collective acceptance understood as an action, a kind of performative agreement making. This basic account, with slight modifications, also applies to collective acceptance understood as a collective attitude, and it can be enlarged to cover agreement making in structured groups as well.30 2.5  Collective Goals and Related Notions In this section we will briefly consider some ideas discussed in the literature that are relatively closely connected to collective goals in the sense of this chapter. These ideas are related to Rousseau’s general will, to Sellars’ moral we-intentions, to Rawls’ views in his political philosophy, and, finally, to public and club goods in economics. Because we-mode goals are based on collective acceptance, our account is related to contractualist ideas in social and political philosophy. We recall that Jean-Jacques Rousseau distinguished between private will, the will of all, and general will. General will represents the true interests of everyone as members of a civil society. In The Social Contract, he wrote that every individual can personally have his particular or private will that contradicts or differs from “the general will which he has as a citizen” (Rousseau 1762, book I, ch. 7). We do not here literally consider what an individual’s will as a citizen amounts to but only what his “will” as a member of any group is. On the other hand, “There is often a great deal of difference between the will of all and the general will; the latter concerns only the common interest, while the former takes private interest into account, and is no more than a sum of particular wills” (ibid., book II, ch. 3). Accordingly, first, the general will is something the individual has as a member of a group, and second, it concerns common interest. Here, instead of the notion of common interest, we have mainly relied on the notion of collective goal. We implicitly assume that the collective goal reflects common interest of the members. In addition, the collective goal—especially if it is in the we-mode—also represents the notion of group on the level of the individual. When a group member has a goal in the we-mode, his intention is to satisfy the goal as a group member and for the group. Besides private will, the will of all, and general will, Rousseau occasionally takes up the notion of corporate will.31 This is the notion that is more closely connected to our notions of collective goal. It, so to speak, mediates between particular wills and the general will. It represents the agents’ common objectives as members of a group (“corporation,” “magistrate,” etc.). According to Rousseau, the general interest of the group is the object of corporate will. A person’s private will can be in conflict with his corporate will (e.g., when his desires

52

C ollective Attitudes and A ctions

are in conflict with his role), and a conflict may also occur between the corporate will and the general will. In this latter case, according to Rousseau, the requirements of the general will should overweigh those of the corporate will. Although Rousseau’s notion of general will, aiming at the satisfaction of general interest, shares many features with our notion of we-mode goal, not just any we-mode goal would qualify as an example of such a general will. For example, we-mode goals can result from any particular agreement, but the general will, if it ever exists, is not consistent with all (kinds of) agreements. What is or can be agreed upon can represent only a mere corporate will, which lacks the generality and impartiality of the general will. For Rousseau, the general will seems to be “general” in two respects. In terms of our classification, the goal carriers are all the members of the civil society. On the other hand, its content is general: it concerns the general principles of future interaction of the citizens.32 Accordingly, the general will not only “comes from all” but also “applies to all.”33 Our I-mode and we-mode goals can be considered in relation to “the moral point of view,” the view that morality applies to all humankind. A case in point is a distinction made by W. Sellars. In his 1968 book, he employs subindices “I” and “we” in his intention-forming shall-operator, thereby obtaining two different modes of intending, and he writes: “The functions of the indices is performed in ordinary language by the contrast between ‘from a personal point of view’ and ‘from the point of view of the group’ or, of more interest, ‘from a moral point of view’ ” (see Sellars 1968, 218n1). As for the contrast between the personal and the group’s point of view, our we-mode/I-mode distinction is broadly similar to Sellars’ distinction. However, his main interest is in the moral aspect of we-intentions, and thus in the case in which the group in question is “mankind generally” (220). Although collective goals typically involve obligations and rights for the parties, in our account, these can be related to all kinds of agreements that need not be moral in this general sense. Even though our account of collective goals need not involve morality in a general sense, it is often applicable in cases that “on the surface” have a competitive structure. Such cases can often be shown to presuppose an underlying collectively accepted goal. Not only cooperative “win-win” situations but also many competitive cases that at least appear to be zero-sum, or to have only “win-lose” types of solutions, presuppose a basic collectively accepted goal. For example, it can be argued that participation in competitive games such as chess or tennis, unlike participation, for example, in street fights, presupposes an underlying collective goal to play by the rules. Likewise, it can be argued that the continued existence of these kind of institutions presupposes such shared basic collective goals. Accordingly, although it cannot be argued that any society has such basic collective goals, it can be argued that if a society

Collective Goals Analyzed

53

is a “well-ordered” society “of justice as fairness,” the citizens share the basic collective goal of “supporting just institutions and giving one another justice accordingly” (Rawls 1993, 202). Conversely, it can be also argued that if there is no such collective goal, the society faces the risk of disintegration and decay, even if the members would privately acknowledge the viability of such a goal. Finally, if the citizens fail to share such collective goals, it can be argued that this is because they disagree about what is just and fair, viz. they disagree about the standards or paradigms of justice and fairness. Should this be the case, the people might be led to thinking that these criteria are given or revealed to them “from outside,” by a supreme authority, as it were, and what they should do is just to comply with them. In this case, if corresponding collective goals would result, the members would not regard these goals as their own construction. As a more detailed study of all of these cases would fall beyond the scope of the present chapter, we let the above suffice for the possible dynamics of these basic and general collective goals. Economists use the notion of public good in their theorizing. How does this notion relate to our notion of collective goal? A public good is one that satisfies the two conditions of indivisibility of benefits and the nonexcludability of benefits. A good is taken to be indivisible when a unit of the good can be consumed by one individual without detracting, in the slightest, from the consumption opportunities still available to others from that same unit. As to excludability, goods whose benefits can be withheld costlessly by the owner or provider generate excludable benefits. (See Cornes and Sandler 1996, 8–9.) A  good is nonexcludable just in case it is not excludable. Indivisibility can be construed as an ontological notion, for it concerns the existential nature of the good. Standard examples of indivisible goods such as national security or clean air or beautiful sunsets are concerned with properties of an external good. By analogy, knowledge (such as that E = mc2) is also indivisible. In contrast, a loaf of bread is divisible. As to nonexcludability, it can be construed as an enforceable normative notion. Thus a good is excludable for an agent in a collective g if and only if there is some agent in g who can be forbidden to consume this good, and the proscription here is enforceable by the authority controlling its use. A  good that is indivisible but excludable is called a club good by economists (a golf club permitting golf playing only to its members). How does all of this relate to our present account of collective goals and goal states as public goods? The first thing to note is that a public or club good is not a collective goal unless further criteria, discussed in the preceding sections, are satisfied. A collective goal need not be indivisible, thus it is not a public good in the economists’ sense. For instance, we can have making a bowl of pea soup as our

54

C ollective Attitudes and A ctions

collective goal. The corresponding goal state is yet a divisible one. However, if a collective goal (here, the achieved goal state) is indivisible (e.g., a piece of knowledge), it is either a public good or a club good in the economists’ sense. It is a public good if it is made available to everyone—also to people outside the group (full nonexcludability), but the group can alternatively exclude it from out-groupers or even make it available only to some group members, for this is also in the power of a we-mode group. When the availability of the good the goal state involves is made exclusive, we have a club good. Even scientific knowledge produced by a group is a club good if not published. Whether all indivisible goods can be made exclusive is a problem we cannot discuss here. The upshot of our present discussion is that collective goals, as discussed in this chapter, are closely connected—especially to the economists’ club goods, because as a default, collective goals and their involved goods are available to the group members but not, or not in general, to others.

2.6 Conclusion The aim of this chapter has been twofold. First, we have analyzed various features ascribable to goals and classified goals according to these predicates. In particular, we defined the notions of sharedness, divisibility, collective content and we-mode, which we then used to classify goals. Our purpose was to find and sharpen some basic tools for an analysis of collective goals. Second, our main focus has been to answer the question, “What makes a goal collective?” To this end, we analyzed the components of collective goals, concentrating on the notion of collective acceptance and on various conditions of the collectivity of goal contents in terms of the conditions of satisfaction, viz. satisfaction for the agent or agents in question. We also argued that full-blown collectivity of goal contents can be characterized basically by thesis (CC), the truth of which is due to the collective acceptance of a goal. In particular, we argued that collective acceptance entails at least one version of the collectivity condition. We also showed that the collectivity of goals is a matter of degree, depending on, for example, normative features and degrees of intentionality. Our basic formulations concern unstructured groups, even if the main ideas carry over to structured ones, as indicated in the text. Finally we sketched an outline of the relations between collective goals of the members’ and the group’s goal. A central aim of our investigation was to show that there is not just one kind of collective goal but many, differing from each other with respect to criteria on the basis of which they can justifiably be called collective. Collective goals have important explanatory functions in social theorizing as discussed in section 2.1. Detailed explanatory uses of collective goals fall

Collective Goals Analyzed

55

outside the restricted scope of this chapter. Another point to be made here is that economists’ public goods and collective goods are related but not equivalent to our collective goals. In fact, no relations of entailment obtain between the economists’ notions and our notion of a collective goal.34 Appendix: Classification of Goals Here, we briefly present a typology in which goals are classified by their carriers and contents, and then we will apply the typology to a concrete example. To keep things simple, the universe of discourse will consist of two agents, x and y, and one goal state, P. P is taken to represent the state that the window is closed. G is the goal operator. The index x in Gx stands for x as the goal carrier, and Px represents that x is the agent of the goal state P so that, for example, GxPx may denote “x’s goal is to close the window.” Gxy denotes that x and y have a goal together, and Pxy denotes that x and y act together to achieve P. Thereby we obtain four types of possible goal contents that are determined by the necessary satisfaction conditions of the goal. In addition to the four types of goal contents, we get three types of goal carriers and resulting twelve goals: Classification of goals by their contents and carriers:    P 

Px Py Pxy

Gx 1  2  3  4 Gy 5  6  7  8 Gxy 9 10 11 12

Cases 1, 5, and 9 do not require that any of the two agents acts, alone or together, to bring about P. In these cases, the goal is just that P will become true, no matter how or by whom, and these goals often represent mere wants, as intentions of either party need not be involved. In terms of our example, case 1 stands for x’s goal that somebody (or something) closes the window, and analogously for y in case 5. In case 9, both agents jointly have this goal. The satisfaction of any of the cases 2, 3, or 4 entails that 1 is satisfied, but 2, 3, and 4 cannot be jointly satisfied. In terms of our example, 2 stands for x’s goal that he himself closes the window, 3 for his goal that y closes the window, and 4 for his goal that they close the window together. Analogous relations hold between cases 5–8 and also between cases 9–12. Conjunctive cases like 1&5, 2&6, 3&7, and 4&8 are common, and they represent what we called shared private goals. For example, 2&6 stands for their independently shared goal that x closes the window. Disjunctive goal contents with any two or three of Px, Py, and Pxy as disjuncts are not infrequent, either.

56

C ollective Attitudes and A ctions

For example, 3 v 4 stands for x’s goal, which is satisfied if and only if either y closes the window or x and y close it together. Cases 9–12 are the central ones for the present paper. These goals of the Gxy type can, but need not, represent full-blown intended collective goals. Clearly, case 9 entails cases 1 and 5, but not vice versa. If x and y together have the goal that somebody closes the window, then x has this goal, and y has this goal, but the converse need not be true. Analogously, 10 entails 2 and 6, 11 entails 3 and 7, and 12 entails 4 and 8. It is important to note that the converse entailments do not hold; for example, 1 and 5 together do not entail 9. This suggests that the “togetherness” of GxyP cannot be reduced to an aggregate of GxP and GyP. This feature of togetherness is a defining characteristic of joint we-mode goals, and it is based on collective acceptance. Consider next cases 10 and 11. In case 10, agents x and y have the goal together (they collectively accept) that x closes the window. If they have made an agreement to this effect, this case yields an obligation or duty on x’s part and a right on y’s part concerning the satisfaction of the goal. Case 11 is analogously asymmetric. Finally, consider case 12, i.e., GxyPxy. It represents a collective goal par excellence. The goal is not only collectively endorsed but its content, and thus its satisfaction conditions, also necessarily involve the agents’ acting together to achieve it. In our example, the agents together have the goal that they close the window together. To satisfy the goal, both of the agents are collectively committed to seeing to it that P.  If they have made an explicit or implicit agreement to this effect, this case yields mutual rights and obligations for the parties.

Notes 1. In terms of the “stit (seeing to it that) theory,” this can be expressed as follows: agent x intends to see to it that the window is closed. Here, both the stit clause (that x sees to it that the window is closed) and the result of x’s action (that the window is closed) are propositional. For technical analyses of intentions and commitments within the AI framework, see Miller and Sandu (1997), where both individual and collective intentions are analyzed by relying on the notion of persistent goal that has propositional content. 2. However, the satisfaction of a collective goal neither entails nor is entailed by the satisfaction of a public good (for an argument, see Tuomela [2000, 401]). 3. In this chapter, both action goals and state goals are accepted. Both have generic content, but they become satisfied by singular actions or states (perhaps repeatedly). The satisfaction of the former presupposes the agent’s action, while the mere obtaining of the goal state will satisfy the latter. The satisfaction of the former typically presupposes that a goal of the latter type is also satisfied: the obtaining of the

Collective Goals Analyzed

57

goal state amounts to the necessary result without which the action could not have been accomplished. 4. Compare the compatible notions in Tuomela (2007, ch. 2). 5. For brevity, we here and below often say “goal P” rather than “goal that P.” 6. Indeed, if the collective acceptance is we-mode acceptance, it can be said that the goal is our joint (i.e., jointly held) goal, a shared we-mode goal in the sense of our category (d) below. 7. Sharing a goal or intention can be analyzed in many various ways. Here, it is taken “minimalistically” as the agents having the same goal content and believing so. Bratman (1993) analyzes “our” shared intention to J as requiring that we each intend “that we J” by way of our “meshing” and mutually reason-dependent subplans of which there is mutual knowledge. According to Velleman (1997), sharing an intention amounts to a kind of joint decision and a concomitant agreeing about a combination of intention expressions. 8. See below for a precise characterization of the collectivity condition. 9. See Tuomela (2013a) for a discussion of this. 10. These are the three most central criteria of the we-mode, as explicated in Tuomela (2007, 2013a). They all rely on collective acceptance for the group. 11. See ­chapter 1 in Tuomela (2007) for acting as a group member either in a strong sense (appropriate for we-mode groups) or in a weak sense (appropriate for I-mode groups). Here is a short definition of the weak sense: (AGMW) An action X performed by some member A of g is an action as a group member in a weak sense if and only if A performed X in the group context in the core sense (of Tuomela 2007, ch. 1) in part for the reason of promoting the ethos of g that she privately accepts and to which she is privately committed (and thus privately socially ought to promote). 12. In view of our above discussion, a more explicit linguistic criterion for a goal being a we-mode goal can be proposed. Let us assume the truth of the proposition, “our goal is P,” analyzed in logical terms as Forgroup(Goal(we,P)), where the “Forgroup” operator means simply that goal P is for the group. This proposition expresses a collective attitude for us in this situation if and only if it is true for us that Goal(we,P) entails and is entailed by CA(we,Goal(we,P)), where CA means our collective acceptance involving collective commitment to Goal(we,P). In other words, the proposition, “our goal is P,” expresses a we-attitude (for us) if and only if this proposition is collectively accepted and, in addition, entails and is entailed by the proposition, “We collectively accept (for us) to achieve P in the sense goals are satisfied.” See Tuomela (2000, ch. 2), and especially Tuomela (2007, ch. 8). 13. See Bacharach (1999) and Hakli, Miller, and Tuomela (2010). 14. In this chapter, group goals will not be properly analyzed; see Tuomela (2007, ch. 6) for an up-to-date discussion. 15. Actually, this condition strictly applies only to the operative members and allows that there be rank-and-file members who do not have the we-mode goal in the full sense of having a goal.

58

C ollective Attitudes and A ctions

16. Over and above collective goals in the senses (1)–(3), there are what might be called “mere” shared goals—states or actions that are private goals of several agents. 17. For recent detailed analyses of different types of “collective intention,” see Tuomela (2007, ch. 4), where the notion of joint intention mediates between a participant’s we-intention and the group’s intention. Parallel distinctions can also be made with respect to goals. Indeed, the contents of intentions can be regarded as goals. 18. See Tuomela (2007, 48); also see (2007, 50) for a version connecting satisfaction for the group and satisfaction for the members. 19. In the case of structured groups, (CC) applies directly to the operative members. However, normally at least “tacit” acceptance of the nonoperative members is also assumed. 20. Also see Tuomela (2000, ch. 6) for discussion. 21. The state-expressing proposition P stands for the content of the goal, and the agent who has the goal that P is committed, privately or collectively, to seeing to it that P will be true. The agent’s goals are primarily identified by their contents. Therefore— and for brevity and stylistic reasons, and when this does not cause confusion—we continue to write “goal P” instead of “goal that P,” but it should be kept in mind that P is propositional. 22. See Tuomela (2007, ch. 6). 23. The present account corrects a slip in the formulation given in Tuomela (2007, 127). 24. As noted earlier, collective acceptance can be taken in a broader sense that need not involve any kind of explicit agreement making. For example, standing collective goals need not involve agreement making, and yet they can be in the we-mode and thus involve collective commitment to satisfying the goal content. It can be argued that these we-mode goals also satisfy the theses of section 2.4. For simplicity, we will below discuss only cases of explicit agreement making. 25. This is a somewhat idealized account of the adoption of collective goals, and it depends on what “collective” acceptance means: it is, of course, trivially true that each goal holder somehow must have “accepted” the goal because he now has it. Yet this acceptance of the goal need not take place collectively even though the content of the goal is collective (collective in the sense discussed above, i.e., that an individual’s satisfying the goal is inseparably connected to the collective’s satisfying it). Our standing collective goals have typically been gradually adopted or learned. 26. Whether the converse implication also holds is another matter. In general, it seems that the answer is “no”—mainly because whether an agreement has been made is not entirely up to what the participants (mutually) believe. For example, whether a “formal” contract indeed has been made or not (is valid or not) depends on whether it satisfies criteria that are “fixed” in the community at large, criteria the parties may mistakenly believe to be satisfied. 27. See Tuomela (1995, ch. 1), where it is shown that more than four layers are hardly ever needed in real life and that often two layers will suffice. Weaker analyses of MB are possible, too. For example, only two “positive” iterations are assumed together with the idea that at any higher layer, the contrary belief content is missing. We will not discuss these alternative analyses here. 28. See Tuomela (2007, 2013a) for group agents and their analogy with individual agents.

Collective Goals Analyzed

59

29. The following proof of (3) applies a “fixed-point” property of mutual belief: Thesis (2) above, i.e., CA(G,P) ->N MBg(CA(G,P)), and the valid principle MBg(CA(G,P)) N Be((CA(G,P)) & MBg(CA(G,P)))

together imply CA(G,P) ->N Be((CA(G,P)) & MBg(CA(G,P))), where “Be” means “every member believes.” This statement implies for all x CA(G,P) ->N Bx((CA(G,P)) & MBg(CA(G,P))), i.e., thesis (3). 30. See Tuomela (2007, ch. 6) for the case of structured groups. 31. A we-mode group can be regarded as a corporate agent or a group agent, especially in a functional sense (see especially Tuomela 2013a). 32. For this latter aspect, cf. Ripstein (1999, 223–24). 33. For a discussion of further historical accounts of “collective volition”, see Tuomela 2013b, where it is shown that already W. McDougall and A. Vierkandt postulated an embryonic I-mode/we-mode distinction. 34. For an argument, see Tuomela (2000, ch. 10), and for the notions of public and club good, cf. Cornes and Sandler (1996).

References Bacharach, M. 1999. “Interactive Team Reasoning:  A  Contribution to the Theory of Co-operation,” Research in Economics 53: 117–47. Bratman, M., 1993, “Shared Intention and Mutual Obligation.” In Faces of Intention. Selected Essays on Intention and Agency. Cambridge, UK:  Cambridge University Press. Cornes, R., and T. Sandler. 1996. The Theory of Externalities:  Public Goods and Club Goods, 2nd ed. Cambridge, UK: Cambridge University Press. Hakli, R., K. Miller, and R. Tuomela, 2010. “Two Kinds of We-Reasoning.” Economics and Philosophy 26: 291–320. Miller, K., and G. Sandu. 1997. “Weak Commitments.” In Contemporary Action Theory, edited by G. Holmström-Hintikka and R. Tuomela, 2:  273–93. Dordrecht, The Netherlands: Kluwer. Rawls, J. 1993, Political Liberalism. New York: Columbia University. Ripstein, A. 1999. “The General Will.” In The Social Contract Theorists: Critical Essays on Hobbes, Locke, and Rousseau, edited by C. W. Morris, 219–37. Lanham, MD: Rowman & Littlefield. Rousseau, J.-J. “On Social Contract or Principles of Political Right.” In Rousseau’s Political Writings (1988), edited by A. Ritter and J. Conaway Bonadella. W.W. Norton & Company.

60

C ollective Attitudes and A ctions

Sellars, W. 1968. Science and Metaphysics. London: Routledge and Kegan Paul. Tuomela, R. 1995. The Importance of Us: A Philosophical Study of Basic Social Notions. Stanford Series in Philosophy. Stanford, CA: Stanford University Press. ———. 2000. Cooperation:  “A Philosophical Study” In Philosophical Studies Series. Dordrecht and Boston: Kluwer Academic Publishers. ———. 2007. The Philosophy of Sociality: The Shared Point of View. New York: Oxford University Press. ———. 2013a. Social Ontology:  Collective Intentionality and Group Agents. New York: Oxford University Press. ———2013b. “Who Is Afraid of Group Agents and Group Minds?” In The Background of Social Reality, edited by M. Schmitz, B. Kobow and H. B. Schmid, 13–35. Dordrecht: Springer. Velleman, D. 1997. “How to Share an Intention?” Philosophy and Phenomenological Research 57(1): 29–50.

3

Group Belief and Acceptance F r e d e r i c k F. S c h m i t t

3.1. Group Belief and Acceptance One way to get a handle on the nature of social groups is to ask how closely groups resemble individual human beings in action, cognition, and motivation. Here I refer to groups that it is natural think of as agents of a sort, such as partnerships, clubs, teams, corporations, government bureaus, and the like. In this chapter, I focus on cognitive similarities and ask whether such groups resemble individuals in holding beliefs. I assume that our everyday attributions of beliefs to groups make sense (whether or not they are true) and attribute cognitive attitudes to groups understood as entities, in much the same way that our everyday attributions of beliefs to individuals attribute cognitive attitudes to individuals on a naïve understanding of these attributions. I will not consider here what to say about how group belief (or talk of it) compares with individual belief (or talk thereof) if this assumption is denied. Although among those who accept this assumption there is general agreement that in some sense, groups assent to propositions, some writers have insisted that groups are capable of assent only in accepting propositions, not in believing them.1 This difference between groups and individuals is supposed to stem from the fact that groups have practical aims that prevent them from assenting for purely cognitive reasons, as required for beliefs, and from the fact that group assent is always voluntary, while beliefs are involuntary. These assumptions about the difference between belief and acceptance might be thought to be supported by one important account of the distinction between belief and acceptance, that of Michael Bratman (1992).2

62

C ollective Attitudes and A ctions

Here I defend the view that groups may both believe and accept propositions, as individuals do. More exactly, I argue that groups are not barred from believing because they differ from individuals in having practical aims that prevent them from assenting for purely cognitive reasons or in assenting only voluntarily. Nothing about these matters prevents groups from believing as well as accepting propositions. To argue for this claim, I rethink Bratman’s distinction between belief and acceptance. In the course of my argument, I fortify and generalize some points that Margaret Gilbert (2002) has already made in defense of the ability of her own joint commitment account of group belief to allow for the existence of group beliefs and not merely acceptances. However, I go on to cite one obstacle to identifying group belief with a joint commitment to believe, as Gilbert does. My argument in this chapter rebuts certain objections to the view that groups have beliefs. It does not yet give any reason to think that groups actually hold beliefs. I am indeed worried by objections to the existence of group beliefs other than those I treat here, such as the objection that group belief fails requirements of holism or systematicity.3 But these objections will tell against the existence of group acceptances as well as beliefs, so they do not by themselves favor a disanalogy between individual and group assent. Nor do they cast doubt on our operating assumption that everyday attributions of group beliefs attribute cognitive attitudes to groups understood as entities, in a manner analogous to our attributions of beliefs to individuals taken at face value. In fact, my argument contributes to a defense of a supraindividualist account of group belief on which there is little more to the attribution of group attitudes than an analogy with individual attitudes.4 More exactly, I defend that account from the objection that no attribution of belief to groups can amount to an analogy with individual belief because groups lack features necessary for having attitudes like individual beliefs. 3.2  Bratman’s Distinction between Belief and Acceptance I take it to be background to an account of belief and acceptance like Bratman’s that these states resemble one other in the following respects. Both the belief p and the acceptance that p take the form of cognitive assent to the proposition that p. Belief p and acceptance that p each entails or involves taking p for granted in reasoning—i.e., reasoning from the premise p without further ado, without considering or reconsidering whether p. Each, at least when reasonable, entails or involves readiness to defend p or, if defense is unavailable, to retract the belief p or acceptance that p, in light

Group Belief and Acceptance

63

of criticism. In these respects, belief and acceptance contrast with supposing p and pretending p. Beyond this, belief and acceptance differ importantly. I  paraphrase and embellish Bratman’s list of characteristics of belief, augmenting his explicit citation of (ii)–(v) with the commonly attributed characteristic (i): (i) It plays a role as a premise taken for granted in theoretical reasoning (i.e., reasoning concluding in further beliefs) and practical reasoning (i.e., reasoning concluding in actions).5 (ii) At least when it is reasonable, it is context independent: a subject’s belief does not vary with a context of theoretical or practical reasoning, though of course, beliefs may be formed or retracted as a result of reasoning. (iii) At least when it is reasonable, it aims at truth.6 This requires that belief is “shaped primarily by evidence” (Bratman 1992, 19). It also requires a disposition to defend p, or, if no defense is available, to retract the belief p, in light of evidence against p or evidence that the belief p is not well supported or reliably formed. (iv) The beliefs of a subject ideally ought all to be consistent with one another. (v) Belief is normally involuntary.7 I will have no reason here to contest the ascription of any of these characteristics to belief. Belief in any usual sense contrasts with supposing and pretending in the ways listed. Let me remark on each characteristic. Regarding (i), one may reasonably reason theoretically from the belief p to other beliefs as conclusions. It is reasonable to reason from one belief to another if the former provides enough reason to believe the latter. By contrast, one cannot reasonably reason theoretically from a supposition or a pretending that p to any belief as a conclusion. Moreover, one may reasonably reason practically from a belief to an action as a practical conclusion. By contrast, one cannot reasonably reason practically from a supposition or pretending to a practical conclusion. Regarding (ii), what one may reasonably suppose or pretend varies with a context of reasoning. Regarding (iii), reasonable supposition and pretending obviously do not aim at truth at all. They require no disposition to defend or retract them in light of criticism. Even ideally, suppositions and pretendings need not all be consistent, as (iv) says of belief. And suppositions and pretendings are normally voluntary, contrary to what (v) says of belief. I note regarding (iii) that contributors to discussions of belief generally interpret the claim that beliefs aim at truth very broadly. Aiming at believing the truth as to whether p may involve intentionally setting about to believe the truth as to whether p, or believing p as a result of having the goal of believing the truth, but it need not involve these. It may be enough that the subject

64

C ollective Attitudes and A ctions

regulates belief as to whether p with respect to the truth, in a sense in which regulation is automatic. Bratman argues that we need a contrasting attitude of acceptance to explain how we reason practically. I take Bratman’s notion of acceptance, qualified and elaborated in ways I will suggest, to be one we employ in everyday life. But before I turn to Bratman’s argument that we need to recognize an attitude of acceptance, let me set aside two notions of assent that might be called “acceptance” that do not usefully contrast with belief and so fall outside my focus in this article.8 One is an ordinary sense of “acceptance” defined in this way: I accept p just in case I linguistically assent to p either overtly or subvocally. In this sense, acceptance does not serve as a premise in reasoning at all, as both belief and acceptance in Bratman’s sense do, although it may sometimes express a premise in the overt or subvocal expression of reasoning.9 The other sense of “acceptance” that I set aside is employed by several writers to characterize convention, constitutive rules, or institutional facts. Thus, Gilbert understands our convention that we put the salad fork on the left as our joint acceptance of a principle of action, that we are to do so.10 I take the relevant notion of acceptance here to be one that entails a commitment to act in accordance with the principle that “we are to put the salad fork on the left.” Accepting in this sense is endorsing; to this extent it is related to linguistic assent. It is natural to think (though Gilbert does not say so) that joint acceptance in this sense entails jointly believing that one is to put the salad fork on the left. Gilbert can allow this. Granted, it is part of Gilbert’s case for her account of convention that a joint acceptance that p in the relevant sense can persist even though members of the group (perhaps all members) cease to believe that p. But a joint belief p on Gilbert’s view of joint belief can also persist even though no member of the group believes that p. So Gilbert’s case for her account of convention is compatible with the view that joint acceptance entails joint belief: there can be a joint belief that one is to put the salad fork on the left even though members of the group have ceased to believe that this is so. Joint acceptance in the sense that defines convention does not, however, entail joint acceptance in the sense of acceptance at issue in this chapter, since the latter does not entail the relevant sort of endorsement of a principle of action. Similar remarks apply to John Searle’s characterization of the constitutive rule or institutional fact “This piece of paper counts as a five dollar bill” as our joint acceptance that one has the power to exchange this piece of paper for goods of a certain value.11 Searle’s notion of acceptance in his characterization is not the notion at issue here. I turn now to Bratman’s argument that we need a notion of acceptance that contrasts with belief. I freely paraphrase his presentation and adopt some of his examples. Bratman proposes that we need such a notion for these purposes:

Group Belief and Acceptance

65

(1) Simplifying reasoning: I may take it for granted that p in my reasoning, despite not believing p, in order to simplify the reasoning. In reasoning to the practical conclusion to leave my umbrella at home when taking my walk, I  take it for granted that it not will rain, even though I  lack confidence in the proposition that it will not rain and so do not believe it. And my taking this for granted enables me to avoid complex reasoning from assignments of probabilities (to the propositions that it will rain and that it will not rain) to an action—e.g., reasoning by weighting utilities by the probabilities. This is accepting p without believing p.12 (2) Avoiding costly error:  I  believe my two-story ladder to be sturdy, but I do not take this for granted. Rather I check to make sure it is sturdy, since a fall from two stories would be catastrophic. Before I  check, I have belief; I  accept the proposition only after I  check. Acceptance here requires me to have a very high degree of belief, given my valuation of the benefits and costs of my action depending on whether or not p. For the converse point that I can make an acceptance without having a belief: I  may have little confidence that my chair is sturdy before I  sit in it, but the floor has a thick rug, and the chair’s collapse would be harmless, so I  judge it not worthwhile to check whether the chair is sturdy, given the expected benefits and potential costs of the action. I take it for granted in my reasoning that the chair is sturdy, accepting that it is, even though I do not believe that it is. (3) Facilitating coordinated behavior: In a case where I coordinate my behavior with others, I  accept their expressed view in order to make coordination possible, even though I don’t believe the view.13 (4) Fulfilling the demands of special relations to others: My friend is charged with a crime. I take his innocence for granted in inviting him to a dinner party. But if placed on a jury to decide his guilt, I would not take his innocence for granted. I both accept his innocence and refrain from accepting it in different contexts, though there is only one context-independent fact as to whether I believe in his innocence. (5) Preconditions for practical reasoning: I take for granted that I will be alive tomorrow when I plan for tomorrow, even if I have good reason to believe that I will die in battle today. I may believe that I will die, but accept that I will not die in the context of my future planning. Acceptance, then, serves as a premise in reasoning even where belief is lacking. Before I examine more closely Bratman’s contrast between acceptance and belief, I want to propose that the uses of acceptance Bratman lists here belong to two fundamentally different kinds of uses—so different that it is misleading

66

C ollective Attitudes and A ctions

to describe them as uses of the same kind of attitude, acceptance. The difference between these kinds of uses matters for whether there is a contrast between beliefs and acceptances in the case of groups. In two of the uses Bratman mentions, (1) and (2), I accept p in a sense that entails a belief of sorts. It entails at least having a suitable degree of belief that p, a degree sufficient for my reasoning to my action given my valuation of the benefit and cost of the action and perhaps also of the benefit and cost to my reasoning of reasoning from the premise p. When, in use (1), I simplify my reasoning to the action of leaving my umbrella at home by accepting that it will not rain, I have a degree of belief that it will not rain sufficient to support my action of leaving my umbrella at home given my valuation of the benefit and cost of my action and the benefit and cost to my reasoning of reasoning from the premise that it will not rain. If I did not have such a degree of belief that it will not rain, then I would not reason simply from that premise but in a more complex way from degrees of belief that it will not rain and that it will rain.14 I have a strong enough degree of belief that, although I do not believe it, I ignore the contrary alternatives in planning what to do. I ignore these alternatives because, although I believe them to some degree, ignoring them simplifies my reasoning to an action. When, in use (2), I accept without believing that the chair is sturdy, I have enough belief that the chair is sturdy that I reason from the premise to the action of standing on the chair, even though I do not believe full stop that the chair is sturdy. My degree of belief is sufficient for me to act given my valuation of the benefits and costs of the action, depending on whether or not the chair is sturdy. Accepting p in uses (1) and (2) thus entails having a suitable degree of belief that p. It is both directed toward the truth and crafted to accommodate a practical concern. In these uses, acceptance simultaneously aims at truth and at practical concerns. In both (1) and (2), the practical concerns permit acceptance in light of a lower degree of belief than would be required if the aim were only truth.15 The other uses, (3)–(5), differ from uses (1) and (2) in that my acceptance in the former uses does not entail that I have a suitable degree of belief that p. It is not directed toward the truth at all, but is entirely crafted to accommodate a practical concern. When in a use of acceptance that falls under (3), I accept that the cost of a house we are building together will be at the top of the estimated range offered by each of the sub-contractors in order to facilitate our construction of the house, I need not have any degree of belief in the proposition. When I  fulfill the demands of friendship by taking my friend’s innocence for granted in use (4), I need not have any degree of belief in his innocence. And when I fulfill the demands of jury duty by not taking my friend’s innocence for granted, I need not doubt his innocence. Finally,

Group Belief and Acceptance

67

when I  take for granted that I  will be alive tomorrow in order to plan for tomorrow in use (5), I need not have any degree of belief in that proposition. What Bratman calls accepting p in these uses does not entail having a suitable degree of belief that p. In this respect, it resembles supposing p and pretending p. It would be aptly described as merely assuming p, a description inapt for accepting p in (1) and (2). Thus, it seems that Bratman’s uses cover two rather different attitudes. Our ordinary term “acceptance” carries with it the suggestion of a suitable degree of belief that p. So it covers uses (1)  and (2)  but not the rest. The term “assumption” in one ordinary sense does cover both these and the remaining uses. For convenience, I will generally follow the literature in using acceptance as a technical term to cover all of these uses. When I want to make clear that I am speaking of uses (1) and (2), I will speak of narrow acceptance. And when I want to speak of uses (3)–(5), I will speak of mere assumption. Let me now list features that Bratman attributes to acceptance (including both narrow acceptance and mere assumption) in contrast to belief. Again, I freely paraphrase and qualify in light of the preceding discussion: (i) Acceptance plays no role in theoretical reasoning (i.e., concluding in a belief) but only in practical reasoning (concluding in an action).16 (ii) Acceptance is context dependent. I  can reasonably accept p relative to one context of practical reasoning (where context is individuated by the decision problem that practical reasoning tries to solve) and reasonably not accept p relative to another context, without changing my mind. Indeed, the example of a precondition for practical reasoning under use (5) above suggests that I can even reasonably accept p relative to one context and not-p relative to another. I may take for granted being alive tomorrow relative to planning for tomorrow and take for granted being dead tomorrow when today I write my last letter to my loved ones back home before going into battle. (iii) Acceptance, when reasonable, aims at practical concerns whether or not it aims at truth: simplicity of reasoning, avoiding costly error, facilitating coordinated behavior, fulfilling the demands of special relations to others, and preconditions for practical reasoning. More exactly, in light of my distinction between narrow acceptance and mere assumption, narrow acceptance aims simultaneously at truth and practical concerns, while mere assumption aims only at practical concerns. (iv) Reasonable acceptance need not be consistent across contexts of reasoning, only within contexts of reasoning.17 (v) Acceptance is normally subject to direct voluntary control.18

68

C ollective Attitudes and A ctions

3.3 Unnoticed Similarities between Belief and Acceptance I endorse Bratman’s contrasts (i), (ii), and (iv) between belief and acceptance. I  have qualified—so qualified I  endorse—contrast (iii). It remains for me to resist contrast (v). The question of whether acceptances are voluntary is complex. To discuss the issue, I will note several important and overlooked similarities between belief and acceptance. These will prove to have consequences not only for whether acceptances are voluntary but also for the aims of acceptances and for whether a group can only accept and not believe propositions. (a) I begin by emphasizing a point that Bratman leaves unmentioned: narrow acceptances are not occurrent takings-for-granted any more than beliefs are. It is commonly allowed that believing p normally involves a disposition to take p for granted occurrently in any theoretical or practical reasoning when the question arises whether p. It is no less plausible that narrowly accepting p normally involves a disposition to take p for granted occurrently in any episode of practical reasoning relative to which one accepts p when the question arises whether p. This point does not immediately show that a narrow acceptance is involuntary, since it could perhaps be up to me whether I am disposed to take something for granted, and equally it could perhaps turn out (as I later argue it does) that some occurrent takings-for-granted are not up to me. But the point is an important preliminary for getting clear on whether mere assumptions are dispositions or takings-for-granted, and identifying mere assumptions with the latter rather than the former favors voluntarism about mere assumptions. We can see that an acceptance is not a mere occurrent taking-for granted. An occurrent taking-for-granted that p may be understood as an act of premising p in reasoning without further ado. We may contrast this occurrent understanding with a dispositional one on which a taking-for-granted is a disposition to premise p in reasoning without further ado. On either the occurrent or the dispositional understanding, it is possible for a taking-for-granted to be intentional, or it may be inadvertent, unnoticed, or even unwanted. It sometimes happens that I come to realize that I am taking something for granted in reasoning only after others point out to me that I am doing so. I may take something for granted when I wish not to. I may respond to my interlocutor’s observation that I am taking p for granted by conceding, “I see that I am taking p for granted, but I did not intend to do so. In fact I do not wish to do so. I do not even accept that proposition, though I am inadvertently reasoning as if I do.” It seems that my response need not be interpreted as offering to retract an acceptance that I have already made. Rather, I am taking p for granted occurrently, but my doing so does not yet amount to my narrowly accepting p. This is enough to show that my taking something for granted occurrently in practical reasoning does not

Group Belief and Acceptance

69

entail that I narrowly accept the proposition taken for granted relative to the pertinent practical issue. Narrowly accepting p, then, entails a disposition to take p for granted occurrently in relevant practical reasoning, just as believing p entails a disposition to take p for granted occurrently in relevant theoretical and practical reasoning. And just as believing p when reasonable involves a disposition to defend p from criticism or, if defense is unavailable, to revise the belief in light of criticism, so narrow acceptance involves a disposition to defend p from criticism relative to my valuation of the benefit and cost of action. In the case of belief, the criticism from which I  must be disposed to defend casts doubt on whether p is true, while in the case of acceptance, the relevant criticism casts doubt on whether p is probable enough to permit action given my valuation of the benefit and cost of action. (This criticism can involve casting doubt on p in a way that raises a doubt as to whether my degree of belief p is reasonable, given my valuation of the benefit and cost of action or my judgment of how much probability is sufficient.) Thus, narrow acceptance resembles belief in requiring, when reasonable, a disposition to take p for granted in reasoning as well as to defend p from relevant criticism or otherwise revise my degree of belief in light of criticism. It is plausible that when I narrowly accept p, I am so disposed because I have a degree of belief p that is sufficient to take p for granted given my valuation of the benefit and cost of action. Our reliance in reasoning on narrow acceptances as dispositions rather than occurrent takings-for-granted suggests that we take dispositions to have some advantage over occurrent takings-for-granted that do not manifest dispositions. What might that advantage be? It might be one analogous to the advantage we reap by forming dispositions in the case of belief. For many propositions p, we must recurrently take p for granted in reasoning about topics and actions related to p. Given that we must do so, cognitive economy is served by forming just one disposition to take p for granted in reasoning, a disposition that is then manifested on many occasions of reasoning, rather than by forming an occurrent taking-p-for-granted on each occasion of reasoning without the benefit of a disposition to do so. The latter practice would evidently require either recalling or judging p anew on each occasion. The former practice would require only a single consideration once and for all (as long as the disposition lasts). The former practice requires less effort and is thus the more economical one. In short, believing carries with it more economy than repeated occurrent takings-for-granted. I take it that the same goes for narrow acceptance, the only difference being that accepting p involves a disposition to take p for granted that is manifested in episodes of reasoning relative to a single context of practical reasoning about a type of action related to p, while belief involves a disposition to take p for granted in any episode of reasoning related to p.

70

C ollective Attitudes and A ctions

The foregoing claims hold for narrow acceptances. I do not say that they hold for mere assumptions. It seems that mere assumptions do not always, or perhaps ever, involve a disposition to take for granted; they typically involve only occurrent takings-for-granted.19 In typical cases of mere assumption, it would not happen that I take p for granted—for example, I assume the expressed view of others in coordinating my behavior with theirs—but I come to see that I did not intend or even want to do so. If someone were to point out to me that I am reasoning from that assumption, I would not naturally respond by saying, “That’s odd; I don’t mean to assume that view.” Typically, when I assume the expressed view of others, I intentionally take their view for granted in this particular episode of practical reasoning, but I am not disposed to take it for granted in all related practical reasoning, even within a single context.20 There remains the question of why mere assumptions differ in this respect from narrow acceptances. If what I have said is right, the explanation must lie in the utility of a disposition for recurrent reliance on the attitude. Mere assumptions must recur less frequently than narrow acceptances, so that economy in forming mere assumptions is less valuable than economy in forming narrow acceptances. But I offer no argument that mere assumptions do recur less frequently than narrow acceptances. (b) I wish to emphasize another point that Bratman does not mention:  a reasonable acceptance, whether a narrow acceptance or a mere assumption, is parasitic on a belief or degree of belief. An acceptance is reasonable only when inferred from a belief or degree of belief.21 We then reason from the acceptance to an action. Note that the inference from a belief to an acceptance that p is neither theoretical reasoning to a belief p (since it does not conclude in a belief but rather an acceptance) nor practical reasoning to the action (since it does not conclude in the action). A reasonable mere assumption—Bratman’s uses (3), (4), and (5)—is parasitic on a belief without that belief serving as a premise in the reasoning to the action. Thus, in the case of facilitating coordinated behavior (use (3)), I rely on my belief that I must take p for granted if I am to conclude my reasoning in coordinated behavior with others. Here I rely on a belief about how to reason to the concluding action, and I rely on this belief in my reasoning about how to reason. It is my reasoning about how to reason that leads me to the mere assumption, my taking p for granted. This resembles practical reasoning to an action, my taking p for granted being the present analog of an action. From my taking p for granted, I then reason to the concluding action. Similar points apply to cases of fulfilling the demands of special relations to others (use (4)) and cases of preconditions for practical reasoning (use (5)). In each of these cases, I take for granted something I do not have sufficient evidence to believe, and I do so on the basis of a belief that I must, or do best to, take this for granted in my

Group Belief and Acceptance

71

reasoning to an action. But the belief on which I rely is about how to reason to the action, and it forms a premise of my reasoning to my taking p for granted, from which I then reason to the action. A reasonable narrow acceptance—Bratman’s uses (1)  and (2)—is inferred from a degree of belief that p. This degree of belief is in effect a premise in my reasoning to the acceptance, from which I then reason to the concluding action.22 In the example of the two-story ladder, I have a sufficient degree of belief to believe the proposition full stop. But I do not have so high a degree of belief, given the high cost of error, that I take the proposition for granted. In the example of the chair, I do not have a sufficient degree of belief to believe the proposition full stop. But I have a high enough degree of belief, given that the cost of error is not too high, that I take the proposition for granted.23 Thus, in cases both of mere assumption and of narrow acceptance, where these are reasonable, I infer the acceptance from a belief or degree of belief and then employ the acceptance in practical reasoning by taking p for granted in the same way I would employ a belief in practical reasoning. Although I have just argued that we in fact infer reasonable acceptances— both narrow acceptances and mere assumptions—from beliefs, I do not rule out here the possibility that we, or beings rather like us, could make reasonable acceptances and rely on them in reasoning without inferring them from beliefs. It is perhaps conceivable to derive reasonable acceptances, at least mere assumptions, from mere takings-for-granted rather than from beliefs. If this is so, reasonable mere assumptions need not be inferred from beliefs. However, as I will observe, any takings-for-granted from which acceptances are inferred will share important features with beliefs—notably, that they aim at truth and are involuntary. So the fact that it is possible for reasonable acceptances not to depend on beliefs, if it is a fact, does not license the conclusion that it is possible for acceptances not to depend on involuntary truth-directed cognitive attitudes. And this means that we cannot deny that groups have beliefs on the ground that they do not have involuntary truth-directed cognitive attitudes, if we admit that groups make acceptances. (c) Finally, Bratman’s distinction needs a modification that is important for our discussion of group belief. He is mistaken in proposing difference (v) between beliefs and acceptances. Specifically, he is mistaken in proposing that narrow acceptances are always voluntary. I can do no more in this chapter than urge that the best argument that beliefs are normally involuntary, if good, carries over to some narrow acceptances.24 However, I am inclined to endorse this argument and the conclusion that beliefs and some narrow acceptances are normally involuntary. Bratman’s thesis in the case of belief is normal doxastic involuntarism. According to this thesis, a belief is not normally under the subject’s direct

72

C ollective Attitudes and A ctions

voluntary control. I cannot normally will to believe p in the way I can will a basic action like wiggling my finger, a nonbasic action like pumping a well, or a cognition like imagining a unicorn. Nor can I normally will actions that foreseeably produce belief in the way that I  can will actions that foreseeably produce an accumulation of savings in a bank account. Nor can I  even normally will a practice of actions that foreseeably produces a belief as I can will a regimen of exercise and diet that foreseeably reduces my blood pressure. Limitations of space force me to focus on what I take to be the best and only forceful argument for normal doxastic involuntarism, so understood. It is an argument based largely on observation of our inability to will beliefs. The argument begins by conceding that I can normally will whether or not to reason theoretically or practically from my belief p, since reasoning is normally an action under my control. And I can often will whether or not to form or hold a belief as to whether p (where forming or holding a belief as to whether p is understood to involve either forming or holding the belief p or suspending belief p). What I cannot normally do is will to believe p or control whether I believe p in some less direct way, these ways to be specified in the course of the argument. Consider, for example, whether I can normally will to believe that an asteroid has just hit Pluto. The first point to make in compiling evidence for a negative answer is that believing it is not an action, so cannot be willed in the manner of a basic or nonbasic action. We next observe that I cannot normally will to believe it as I can will imagining a unicorn. It seems that I cannot normally will to believe it without having considered the reasons for believing it; nor can I do so once I have considered the reasons. I observe in my own case that I cannot will to believe that an asteroid has just hit Pluto before I have judged whether the reasons pertinent to whether to believe this favor believing it (where my options for judgment are judging that the reasons favor believing it, judging that they do not favor it, or suspending judgment as to whether they favor it). So the question is whether I can will to believe this after I have made a judgment on this matter. It seems clear that I cannot will to believe it if, as in this case, I judge that these reasons do not favor believing it. And if I judge that these reasons do favor believing it, I cannot help but believe it. It is not an option for me to refrain from believing it. Thus, the only circumstance that remains in which I might possibly will to believe it is one in which I suspend judgment as to whether the reasons favor believing it. And here, too, it seems that I cannot normally will to believe it; I  simply refrain from believing it. Thus, I  cannot normally will to believe the proposition, either before or after I  have judged the reasons. What holds for belief in this case with regard to willing would seem to hold for any normal case.25 So far, the argument, if successful, shows that I cannot normally will to believe that an asteroid has just hit Pluto.26 But

Group Belief and Acceptance

73

could I control whether I believe a proposition in a different way—in the way I control whether I save a sum of money? Could I control it by willing actions that foreseeably produce the result? The question is whether I could manage the reasons available to me in such a way that they foreseeably produce the belief p. Could I will to acquire or note evidence as to whether my in-laws like me (p) in such a way that, as I foresee, I will judge that the reasons favor believing p and so I will believe p? I can steer clear of the looks and remarks of relatives who are notoriously hard to please. But my doing so enables me to foresee that I will believe p only if I recognize the effect of this selection on my judgment of the reasons. Yet if I recognize this effect of selection on my judgment, that will tend to prevent me from believing p. (I take it that this is so even if I do not continue to recognize the effect over its entire duration.)27 So either I do not foresee the result or my action tends not to produce the belief p. My will does not normally bring about the belief p in a manner analogous to that in which it brings about my savings. To take the final step in this observational argument for normal doxastic involuntarism, it is doubtful that a belief normally arises by willing a practice of actions that produces a disposition by habituation. A belief is a disposition to do something analogous to an action, namely, to take p for granted in theoretical and practical reasoning. In many cases of action, I can normally bring about a disposition to perform the action and do so by willing a practice of such actions. I can bring about a disposition in myself to take an umbrella when it rains by willing a practice of doing so. One might hope that I can normally will a disposition to take p for granted in theoretical and practical reasoning by willing a practice of taking p for granted. But is this so? Certainly, I can will to take p for granted in some specified episodes of reasoning; and I can will such a thing repeatedly for various episodes of reasoning. I can will a practice of such takings-for-granted. I allow that this might give rise to a disposition of taking p for granted that amounts to supposing or pretending p. But it seems doubtful that I can will to take p for granted in the relevant kind of reasoning for a belief—theoretical reasoning to beliefs and practical reasoning to actions. My recognition that I will to take p for granted will prevent me from concluding a belief or an action from this premise. To conclude a belief or action I must place enough stock in the proposition I take for granted to form the belief or to act on the basis of my taking that proposition for granted. But if I take the proposition for granted merely because I will to do so, I will not have the confidence that I need to rest a belief or an action on my taking p for granted, any more than I do when I merely suppose or pretend that p. To be sure, if I take the proposition for granted because I judge the evidence to favor it, I will have the necessary confidence; but in this case, my judgment will cause me to take it for granted in a way incompatible with my willing to take it for granted. Thus, it

74

C ollective Attitudes and A ctions

seems that I cannot normally will the practice of actions necessary to produce belief by habituation. In sum, observation shows that I cannot normally will to believe p, nor can I normally will an action or a practice that foreseeably gives rise to a belief p.28 This is the best and most forceful argument for normal doxastic involuntarism.29 I believe that this argument does carry over from beliefs to some narrow acceptances—in particular, to narrow acceptances like case (2), in which assumptions enable us to avoid costly error. To begin, a narrow acceptance is no more an action than a belief is. Can I, in normal conditions, will my acceptance that the chair is sturdy when I reason practically to the action of sitting on the chair, as in case (2)? Suppose I assign a cost of error to my relying on an acceptance that the chair is sturdy, based on the value I place on avoiding falling off the chair. Suppose I assign a degree of belief to p that is not high enough reasonably to accept p, given the cost of error I assign. Then if conditions are normal, there is nothing I can do to will the acceptance of p, given the cost of error I assign. I simply do not sufficiently incline to the truth of p to accept p. Suppose instead that I assign a high degree of belief relative to my assignment of the cost of error. Let us assume it is not yet a high enough degree of belief for me reasonably to believe p, but it is high enough that I reasonably accept p. I do sufficiently incline to the truth of p to count as accepting it. Is there room in this case for me to have willed to accept p? To be sure, I can normally will whether or not to reason from my acceptance as to whether p if I do make such an acceptance. For reasoning is normally an activity under my control. And I can normally will whether or not to make any acceptance at all as to whether p. But in these regards, acceptance is no different from belief. What I cannot have done, if the case is normal, is to will to accept p by willing a sufficient degree of belief that p given the cost of error. I could not do so by willing my degree of belief that p, since this is not normally up to me. (I have not set out an argument for normal doxastic involuntarism regarding degree of belief, but I take that argument to parallel that for full-stop belief.) Nor could I have willed to accept p by willing my assignment of the cost of error. It is not normally up to me what value I ascribe to my own health and safety. Normally, the assignment is a product of my outlook on life as applied to a case, and I do not will either my outlook on life or what cost that outlook ascribes to error in the case. Nor could I will what counts as a sufficient degree of belief given the cost of error. This, too, is a product of my outlook on life. It is not up to me to will that a low degree of belief that the two-story ladder is sturdy is sufficient, given my judgment that error would be catastrophic. For it is not up to me to place a low value on my own life. Since I cannot will acceptance by willing a degree of belief sufficient given the assigned cost of

Group Belief and Acceptance

75

error, the only remaining option for having willed my acceptance that p is to have willed it once I have formed my degree of belief, assigned the cost of error, and judged what is a sufficient degree of belief given my valuation. But this is just as implausible as that I can will my degree of belief once I have weighed the evidence. I do not think the following would be an accurate description of a normal train of thought: “I have considered the evidence and have a low degree of belief that the chair is sturdy, and I recognize that falling off is a big deal; I have considered for a moment accepting that the chair is sturdy and could have done so even though I recognize it to be ruled out by the value I place on my own life, but in the end, I suspended acceptance.” This argument for normal involuntarism about some narrow acceptances seems as forceful as the argument for normal doxastic involuntarism.30 Could I will actions that foreseeably produce a narrow acceptance like that in case (2), as I will actions that result in my accumulating a sum of money? Here, too, such narrow acceptances seem to be in the same boat as beliefs. If I am to will an action that foreseeably produces my acceptance that the chair is sturdy, it would normally not be the action of selectively acquiring or noting evidence about the sturdiness of the chair. I can engage in this action. But doing so enables me to foresee that my degree of belief will be sufficient for acceptance only if I recognize the effect of the selection. Yet my recognition of this effect tends to reduce my degree of belief and hence tends to prevent me from inclining to the truth of p, and so from accepting p. Thus, normally the action does not foreseeably produce the acceptance. Could I normally control whether I accept p by willing actions that foreseeably affect my assignment of the cost of error, given a degree of belief I  do not will? Evidently this could occur only by my willing actions that foreseeably affect my judgment as to what the consequences of error are, on the basis of which I value these consequences and assign the cost of error. For once I judge that these are the consequences of error, there is no room for me to will an assignment of value. My assignment is fixed by my outlook on life. Yet I cannot normally will actions that foreseeably affect my judgment as to what the consequences of error are. That judgment is a belief and so normally is not subject to control in this way. Thus, such a narrow acceptance is not normally under the control of actions that foreseeably produce it.31 Nor can such a narrow acceptance normally arise by willing a practice of actions that produce a disposition by habituation. My willing to take p for granted repeatedly in a context of reasoning might give rise to a disposition of taking p for granted, but taking p for granted does not manifest acceptance unless what is manifested is a degree of belief that p. But a disposition that arises in this way does not manifest a degree of belief. For my recognition that I will be taking p for granted prevents me from concluding an action from this

76

C ollective Attitudes and A ctions

premise. I do not have the confidence that I need to rest an action on it—I do not have a degree of belief that p. Thus, the observational argument for normal doxastic involuntarism carries over to such narrow acceptances.32 I conclude that there is as much reason for normal involuntarism in the case of narrow acceptances to avoid costly error as there is for normal doxastic involuntarism. The question arises whether the observational argument for doxastic involuntarism also applies to narrow acceptances to simplify reasoning, as in case (1). It is not entirely clear how to answer this question. Note first that in deciding whether to take my umbrella, I may choose to reason in a simplified way or in the more complex way that involves solving the decision problem by relying on my degrees of belief as to whether it will rain. The decision problem does not by itself fix whether I am to reason in the simplified or the complex way. It is true that I may place so high a value on simplified reasoning that I prefer it to complex reasoning. And once I set that value, it is determined that I am to reason from the proposition that it will not rain. Still, it is up to me whether to take this proposition for granted in my practical reasoning; for it is up to me whether to place so high a value on simplified reasoning. This contrasts with case (2), where my need to rely on an assumption is determined by an outlook over which I have no control rather than by a choice about how to reason. One might think that this favors the position that my narrow acceptance in case (1) is voluntary. But this is too quick. For the question is not whether it is up to me to take this proposition for granted in this episode of reasoning. The question is whether my narrow acceptance is voluntary. And my narrow acceptance is not merely my taking the proposition for granted but rather a disposition to do so in reasoning of this kind. It is not clear that it is up to me whether to be disposed to take this proposition for granted in such reasoning. For this reason, I leave open the possibility that the observational argument for doxastic involuntarism does carry over to narrow acceptances to simplify reasoning. We may ask the same question about mere assumptions:  does the observational argument for doxastic involuntarism carry over to them? Here, the argument seems less compelling, if only because mere assumptions are not dispositions but merely takings-for-granted. In case (5), for example, I assume something that I believe to be false—that I will be alive tomorrow—in order to make plans for the future. It is up to me whether to take this for granted in reasoning to my future plans; I could just as well reason from what I believe and conclude that my future plans are to be vacuous. In general, mere assumptions would seem to be voluntary. The key points I have made so far in this chapter are that both narrow acceptance and mere assumption are inferred from belief. Reasonable narrow acceptance aims at truth in light of the cost of error if reasonable belief aims at truth,

Group Belief and Acceptance

77

although reasonable mere assumption does not. And there is as much reason to take normal narrow acceptances to avoid costly error to be involuntary as there is to take normal belief to be so. I will now deploy these points against arguments for group exceptionalism, that groups accept but do not believe because group aims are practical and group cognition is voluntary. 3.4 The No-Aim-of-Truth Objection to Group Belief Let us turn to our primary question in this chapter:  are groups restricted to accepting rather than believing propositions?33 Writers on group belief have raised two objections to the existence of group belief on a nonsummative account of group belief (i.e., an account on which a group’s believing p is not merely a matter of each member of the group believing p).34 I will not set out the specific formulations of the objections made by various writers since Gilbert (2002) has carefully examined and persuasively rebutted a number of these. Rather, I will sketch some general objections and supplement Gilbert’s replies to them. I will then provide my own argument that, to speak roughly, there can be no more objection to group belief than to group acceptance. The first objection to the existence of group belief concerns the aim of truth as opposed to practical value in believing, and the second concerns the involuntariness of belief. Each objection purports to show that the assents we call “group beliefs” in everyday life are more aptly called group acceptances. In discussing these objections, it will be convenient to follow Gilbert in referring to the assents we call “group beliefs” as group beliefs*, a term intended to be neutral between group beliefs and group acceptances. Parties to the dispute under discussion will agree that group beliefs* exist and that if there are genuine group beliefs, they must be group beliefs*. So our question of whether there are genuine group beliefs comes to whether some group beliefs* are genuine beliefs. The first objection to the claim that some group beliefs* are genuine beliefs is this: believing p, when reasonable, aims only at believing the truth as to whether p; it does not aim at a practical value.35 (Here, “aiming” is taken in the same broad sense generally employed in discussing individual belief, as satisfied by either intentionally setting about to form a belief p only if p is true, or merely regulating belief with respect to the truth.) More exactly, this is so in the following sense: the subject believes p as a result of the purely cognitive aim of believing p only if p is true, and the role of this aim in regulating the subject’s belief guarantees that the subject would believe p were truth the only aim. In this sense, the aim of truth determines that the subject believes p. The objection is that group belief*, even when reasonable, does not aim only at the truth in this broad sense. For aiming only at the truth in belief is incompatible with aiming

78

C ollective Attitudes and A ctions

at something other than the truth, such as a practical value, in such a way as to figure in the regulation of the subject’s belief so that it does not guarantee that the subject would believe p were truth the only aim. Yet, the objection proceeds, the members of a group aim at practical values in forming a group belief*, and such an aim does figure in the regulation of the subject’s belief in the way just described. Consequently, the group belief* does not aim only at the truth. So a group belief* is not a genuine belief. One might think this objection to be particularly compelling on Gilbert’s joint commitment account of group belief*, on which a group believes* p just in case members are jointly committed to believe p as a body in so far as this is possible (Gilbert 1987; 1989, 288–314; 2002; for further discussion, see the final section of this chapter). In making such a commitment, the objection proceeds, the members of the group aim at achieving certain practical values—the practical values they would hope to achieve by fulfilling this commitment. Consequently, one might think, practical aims affect whether the group believes* p, to such an extent that the role of the aim of truth does not guarantee that the subject believes what he or she would believe were truth the only aim. So the group does not believe p. Such a joint commitment cannot constitute a genuine belief. It can amount only to an acceptance. So on Gilbert’s joint commitment account, a group belief* is not a genuine belief. Gilbert has shown that this objection to her account rests on an invalid inference. I freely adapt her response. We may grant to the objection that since group belief* is a joint commitment to believe as a body insofar as this is possible, and members have an aim in making this joint commitment, the members’ aim figures in regulating the group belief*. But there are two ways to understand what it is for the members to have an aim here. We can understand this to mean that each member has a singular aim. Or we can understand it to mean that the members have a joint aim. Now, group belief* aims only at the truth only if the members jointly aim only at the truth. We may grant that in many instances of group belief*, each member will aim singularly at practical values in joining the joint commitment; each is trying to satisfy personal practical goals in making the commitment. But it does not follow that the members jointly aim at practical values or that the group aim does not figure in regulating group belief* so that the group would believe* p if the members were jointly to aim only at the truth. So it does not follow that the group belief* does not aim only at the truth. Indeed, it might be that members’ diverse singular aims cause them to aim jointly only at the truth. The objection rests on an invalid inference from what the members singularly aim at to what the group belief* aims at. Gilbert’s response to the stated objection to her account is telling. The objection does not establish that group belief* on her account must fail to aim at the truth and thus fail to amount to group belief. This is a significant point

Group Belief and Acceptance

79

since it seems unlikely that any analogous argument that group belief*, on some account of it other than Gilbert’s, is not group belief will be any more convincing. In short, Gilbert’s response is perhaps enough to deflect the objection that on no account of group belief* can it be group belief, since group beliefs* aim at practical values. But deflecting the objection in this way does not yet make it plausible that group belief*, when reasonable, aims only at the truth. We may ask whether this is so without assuming any particular account of group belief*—Gilbert’s or any other. To answer this question, I begin by observing that a reasonable group belief* can aim only at the truth. I will assume that it is sufficient for a group belief* to aim only at the truth that it is regulated with respect to the truth, and for the latter it is enough that the group belief* p is so regulated that group members tend not to believe* p jointly unless p fits the relevant evidence possessed by the group and this evidence meets certain standards (such as being ample and being representative of available relevant evidence). For example, individuals might form a poetry club because they wish to discover correct interpretations of poetry and believe that discussion with others will further this aim. They constitute the club in such a way that the club too aims to discover correct interpretations. If the individuals are sufficiently skilled at discussion, the group will not form a belief* p unless p fits the evidence meeting standards. This is an example of how a group can form reasonable beliefs* that aim only at the truth by inheriting this aim from the aims of the members in forming the group. A group can also form such beliefs* because it is regulated by incentives to do so regardless of the members’ aims in forming or joining the group. These incentives may arise from the group’s own goals or from incentives imposed from without. For example, the board of directors of a corporation may be so constituted that it has an incentive not to form a corporate belief* that an expenditure, investment, sale, marketing policy, hiring policy, and so forth benefits the corporation unless the evidence favors thinking so. There could be penalties for violating this requirement imposed on the board by its charter. There could even be legislation that prohibits the board of directors of a corporation from forming a corporate belief* unless the evidence favors thinking so, and enforcement and penalties might be strict enough that the board conforms to the requirement despite incentives to deviate from it. The question arises whether such conditions are sufficient for aiming only at the truth. It is true that these conditions are not purely cognitive. But I take it that if it turned out that an individual’s belief-formation conformed to the analogous requirement of assenting only if the evidence favors the proposition as a result of similar pressures, this would suffice for aiming only at the truth. We should say the same about the group case.

80

C ollective Attitudes and A ctions

This is enough to show that a reasonable group belief* can aim only at the truth. But we would also like to know whether group beliefs*, when reasonable, do often enough aim only at the truth. The answer is affirmative if, in cases of reasonable group beliefs*, there are phenomena like those just cited as sufficing for aiming only at the truth. Although I can give no decisive argument for an affirmative answer, I can make headway toward it by arguing that such phenomena are common enough that group belief* does not suffer greatly in comparison with individual belief. First, groups formed with the aim of discovering or establishing the truth are common. These include universities, think tanks, research divisions of corporations, medical clinics and hospitals, government agencies that collect information like the census bureau, government investigative agencies, courts, and informal discussion groups like the poetry club. Plausibly, these groups often enough aim at group beliefs*, and often enough when they do, they aim only at true group beliefs*. They may, of course, aim at realizing incentives offered to the group, but often enough these incentives function to motivate the group to aim only at true group beliefs*. Second and more important, virtually all groups have group goals and sometimes pursue them reasonably. Reasoning practically to actions expected reasonably to satisfy these goals requires reasoning practically to these actions in a reasonable way from certain sorts of premises. Such reasoning requires that these premises are themselves reasonably expected to be true. In general, only practical reasoning from true premises can reasonably be expected to satisfy goals. But reasonably expecting the premises to be true requires in turn that this expectation aims only at the truth. To the extent that groups recognize the force of this line of thinking, they will recognize that if they are reasonably to expect their goals to be satisfied, they must rely in their practical reasoning on expectations that aim only at the truth. Such groups will attempt to satisfy their goals by relying in their practical reasoning on expectations that aim only at the truth. A  corporation, a street gang, or even a research team of scientists can perhaps assent to propositions with an aim other than assenting to the truth. They can perhaps tailor their assents to their interests without regard for what the evidence tells them. In such cases, their assents more closely resemble presumptions or pretendings than reasonable beliefs, even if they count as instances of group belief*. But groups cannot tailor all of their assents in this way if they are to engage in reasonable pursuit of their goals. The expectations on which they rely for goal satisfaction must generally aim only at the truth. Since groups rely on group beliefs* in the practical reasoning that leads to goal satisfaction, these group beliefs* must generally aim only at the truth. Third, and also important, there are social pressures on groups to form beliefs* that aim only at the truth, and it seems likely that these pressures

Group Belief and Acceptance

81

often enough cause groups to hew to the straight and narrow path of aiming only at the truth. We can see the effect of these pressures when we observe that groups will strive, as individuals do, to make others think that the propositions they present themselves as reasonably believing* aim only at the truth. I have never heard a board of directors of a corporation admit anything like “this is our belief, it’s reasonable, but it doesn’t arise from a concern to fit our belief to the evidence.” Nothing like this is admitted because groups take their beliefs* to be beliefs, and they, like everyone, presuppose that everyone assumes reasonable belief to require aiming only at the truth. This unanimous presupposition about what everyone assumes pressures individuals and groups alike to avoid appearing, in cases in which reasonable belief is called for, to form a belief* not aimed only at the truth. The pressure to avoid so appearing in turn drives individuals and groups alike to aim only at the truth in cases in which reasonable belief is called for since the easiest and most assured way to avoid so appearing is to aim only at the truth in these cases. These three points support the view that group belief* does not compare poorly with individual belief with respect to aiming only at the truth. This seems to me enough to cast doubt on the “no aim of truth” objection to group belief. There is the further point against this objection that, as I argued earlier, if groups genuinely make acceptances, whether narrow acceptances or mere assumptions, then, since acceptances are inferred from prior beliefs or degrees of belief of the subject, groups must also have beliefs. Reasonable group acceptances will surely require reasonable prior beliefs, and the latter aim only at the truth. So one who maintains that groups make acceptances and that some of these acceptances are reasonable must admit that groups have some reasonable beliefs that aim only at the truth. 3.5 The Voluntariness Objection to Group Belief Let us turn now to the second, voluntariness objection to the existence of group belief. The objection is that belief is normally involuntary and not subject to the will. But group belief* is voluntary and subject to the will. The normal voluntariness of group belief* is alleged both to have intuitive plausibility and to follow from Gilbert’s joint commitment account of group belief*. Since on this account a group belief* is a joint commitment—one that obtains when each member of the group is willing to be jointly committed to the attitude—and according to the objection such willingnesses are subject to the will, it follows that group belief* is subject to the will. It is then suggested that since, on Gilbert’s view a group belief* is voluntary, it is a voluntary assent and so is an acceptance rather than a belief.

82

C ollective Attitudes and A ctions

There are three components in my reply to this objection. The first component is that the point that rejecting the view that group belief* is belief in favor of the view that it is acceptance does not by itself support the claim that group belief* is voluntary. Suppose that group belief* is group acceptance. This supposition by itself leaves open whether the group belief* is narrow acceptance to avoid costly error or instead narrow acceptance to simplify reasoning or mere assumption. I conceded earlier that the latter two sorts of acceptance may be voluntary. But I argued that the former sort of acceptance is involuntary. So we should conclude that those group beliefs* that are narrow acceptances of this sort are involuntary. Thus, rejecting the view that group belief* is belief in favor of the view that it is acceptance leaves open the possibility of involuntarism about group belief*. The second component in my reply to the voluntariness objection is to the claim that on Gilbert’s view, group belief is voluntary since it is a voluntary joint commitment of the members. Gilbert responds convincingly to this claim. A joint commitment to believe as a body is under the voluntary singular will of any given member, in the sense that no joint commitment obtains unless that member singularly wills something. And the persistence of the joint commitment is under the voluntary singular will of any given member in the sense that it persists unless that member grants its rescission (in tandem with all other members) since the joint commitment remains in effect until each member grants rescission. But it does not follow from these points about the singular will of each member that the joint commitment is under the voluntary will of the subject of the group belief*—that is, of the group itself or of the members willing jointly. Yet what an involuntary group belief requires is that the belief is not under the will of the subject of the group belief*. To elaborate this reply, the following are consistent: Gilbert’s joint commitment account of group belief, the voluntariness of each member’s contribution to a joint commitment, and the existence of involuntary group beliefs. There are possible cases of involuntary group belief* that meet the requirements of Gilbert’s account even though each member’s contribution is voluntary. Suppose there is an overriding incentive—say, payment of a large sum of money—for each director of a corporation not to go in for a joint commitment to believe any proposition unless the proposition fits the evidence. This incentive structure induces a tendency to form a joint commitment to believe only what aims at the truth. When this tendency is operative, the joint commitments formed aim at the truth. Suppose this tendency is operative in the formation of a joint commitment to believe p. On Gilbert’s account, this commitment amounts to a group belief, and this belief satisfies the requirement of aiming only at the truth. Now each director voluntarily contributes to the joint commitment, by expressing a willingness to enter into the commitment. Each has a sort of control over

Group Belief and Acceptance

83

the formation of the joint commitment since each must contribute if the joint commitment is to obtain. Despite this, the group itself might lack control over whether the joint commitment obtains. The group might lack control over whether the joint commitment obtains by lacking control over the formation of the commitment or by lacking control over the rescission of the commitment, as required for retracting a group belief on Gilbert’s view. The group might develop a habit of joint commitment-formation that is so strong that the group does not have the power to refrain from forming a joint commitment to believe p. The habit involves an inertia in the members; although each has voluntary control over whether to join the commitment, the members jointly lack control because they cannot agree to depart from habitual commitments. This is compatible with the habit being one that regulates belief in such a way that the group tends to believe only what fits the evidence. It might happen that the group lacks control over the rescission of the joint commitment to believe p. What is required for the rescission of the joint commitment, on Gilbert’s view, is that each member joins the rescission. A member would normally have the power to join the rescission and do so by expressing a willingness to rescind. Although each member can voluntarily express a willingness to rescind, it might nevertheless happen that the members do not jointly control whether rescission occurs. Perhaps the monetary incentives for continuing with the joint commitment are so strong that the group lacks the ability to bring it about that all members rescind the joint commitment; they cannot, for example, convince each member to express a willingness to rescind.36 In this case, the group lacks control over whether the joint commitment to believe p obtains. If involuntary group belief of these sorts is frequent enough, then on Gilbert’s account, group belief is normally involuntary. I therefore reject the objection to Gilbert’s view that because the joint commitment is voluntary, and belief is normally involuntary, Gilbert cannot identify the former with the latter. (However, I will later raise the concern that although a joint commitment need not itself be voluntary, it is sensible to join a commitment only if one thinks of the item to which one is committed as voluntary, so that a belief cannot be a joint commitment sensibly joined.) The third component in my reply to the voluntariness objection is independent of Gilbert’s account of group belief. Plausibly, some groups control whether they believe* propositions by controlling whether they have a tendency to form a group belief* only if it aims only at the truth. But in any instance in which a group controls whether it believes* a proposition q in this way, there must be some prior belief* p such that the group does not control whether it believes* p by controlling whether there is such a tendency. For if the group controls whether it believes* q by controlling whether there is a tendency to form a group belief* only if it aims only at the truth, it must do so by relying

84

C ollective Attitudes and A ctions

on a prior belief* p. A group controls states of affairs only by performing group actions, and these depend on group practical reasoning, which in turn begins with group beliefs as premises. But these further beliefs* cannot always be controlled, on pain of a regress of controlling the prior beliefs required for control. Thus, a group cannot control whether it believes* q, for every belief* q, by controlling whether there is a tendency to form a group belief* only if it aims only at the truth. If a group is restricted to this way of controlling its beliefs*, then not all of its beliefs* are voluntary. These replies show that group beliefs* can be involuntary, and this is so even on Gilbert’s view. They suggest that actual group beliefs* are frequently, perhaps normally, involuntary. If so, the normal involuntariness of belief is consistent with the claim that some group beliefs* are beliefs rather than acceptances. In sum, neither the no-aim-of-truth nor the voluntariness objection shows that group beliefs* are acceptances rather than beliefs, whether group beliefs are understood in Gilbert’s way or in some other way. 3.6 Group Acceptances, Hence Group Beliefs To these replies to the objections to group belief, I add a final response. As I have argued, individual acceptances, whether narrow acceptances or mere assumptions, are inferred from prior individual beliefs. So to the extent that group cognition is analogous to individual cognition, if groups make acceptances of either kind, and group acceptances are analogous to individual acceptances in being inferred from prior individual beliefs, then groups must hold beliefs. Now, how analogous group assent is to individual assent is precisely what is at issue in this chapter. A proponent of group exceptionalism about assent might therefore disallow my latest response on the ground that it begs the question by assuming that group beliefs*, if acceptances, are like individual acceptances in being inferred from beliefs. I have assumed that group acceptances are like individual acceptances overall, hence like them in being inferred from beliefs, because the typical proponent of group exceptionalism assimilates group beliefs* to assents like individual acceptances. But suppose the proponent of group exceptionalism sidesteps my assumption by suggesting that group beliefs* differ from individual acceptances in being inferred only from voluntary assents, not beliefs. Of course, it is clear enough that the assents from which group beliefs* are inferred must resemble beliefs in aiming only at the truth if these assents are to play the right role in the formation of group acceptances, a role analogous to the one played by individual beliefs and degrees of belief in the formation of individual acceptances. But one might maintain that these assents nonetheless differ from beliefs in being voluntary. Yet on what ground might one say so? One might say that these are mere takings-for-granted. But

Group Belief and Acceptance

85

it does not follow from this that they are voluntary. The case that aiming at the truth prevents voluntariness, whatever its force may be, would apply to these takings-for-granted. And as I  have argued, even the takings-for-granted that manifest some individual narrow acceptances are involuntary. The plausible view is that not all takings-for-granted from which group beliefs* are derived are voluntary. This is enough to counter the worry that I  beg the question by assuming that group belief* is analogous to individual acceptance. I conclude that groups believe if they accept, and one cannot maintain that in all instances group belief* is group acceptance or voluntary assent rather than genuine belief. 3.7 To What Extent Do Group Beliefs Resemble Individual Beliefs? I have addressed the question of whether groups can believe as well as accept propositions because the answer to this question bears on whether groups differ fundamentally from individuals in their capacities for psychological attitudes. I  have urged that the main arguments that groups cannot believe but only accept propositions lack force, and I have argued that groups must believe if they accept propositions. There remains the amorphous question to what extent group beliefs resemble individual beliefs. In this closing section of the chapter, I  address this question by considering to what extent Gilbert’s joint commitment account makes group beliefs analogous to individual beliefs. Gilbert has given various formulations of her joint commitment account (1987; 1989, 288–314; 2002), and she is always meticulous in the formulations of her views. For present purposes, I will have to rely on my own rough summary statement of her mature account:  a group belief p arises when, under conditions of common knowledge, each member of a population expresses an unconditional personal willingness or readiness to be jointly committed with the others to believe p as a body insofar as this is possible, with the understanding that a joint commitment is formed when each other member expresses a like unconditional personal willingness with a like understanding. Under these conditions, a collectivity is formed (if it does not exist already), the members of the population constitute a group, and the group believes p. The formulation of the account raises the question of whether a joint commitment to believe p as a body insofar as this is possible can really be sufficient for a group belief. I assume that the qualifying phrase “as a body” is inserted to render “to believe” neutral between the attitude of an individual and the attitude of a group. I will not address the question of whether this succeeds in preventing the account of group belief from implicitly referring to group belief in its characterization of the content of the joint commitment.37 But I want to

86

C ollective Attitudes and A ctions

point out that the phrase “insofar as this is possible” poses a dilemma for the account. If such a joint commitment—to believe as a body insofar as this is possible—could exist, even though it is not possible for the members to believe p as a body, then it is hard to see how this commitment suffices for group belief.38 There might be a joint commitment to believe p as a body insofar as this is possible even though it is not possible for the members to believe p, and hence there is no group belief. But if such a joint commitment could not exist if it is not possible for members to believe p as a body, the phrase “insofar as this is possible” would seem redundant, and the account would then come simply to the claim that a joint commitment to believe p as a body suffices for group belief. On the account simply formulated in this way, it is less obvious that the relevant joint commitment falls short of group belief; perhaps a commitment to believe is enough for group belief. But if so, it would seem that group belief requires significantly less than individual belief does, since an individual’s mere commitment to believe p does not plausibly suffice for the individual’s believing p, as I will argue momentarily. It would make no sense to use the phrase “insofar as” comparatively, so that “believe p as a body insofar as this is possible” means “do something sufficiently like believing p,” to allow for the possibility that a group cannot exhibit the processes or conditions required for belief. If we were to read “insofar as this is possible” in this comparative way, the joint commitment account would maintain that a joint commitment to believe p as a body suffices for belief when members jointly commit to doing something that may fall short of believing p. But it makes no sense to maintain on the one hand that a group might not be able to exhibit the processes or conditions required for belief, but on the other hand that a joint commitment to do something that might fall short of believing suffices for belief. For these reasons, I am inclined to think that the joint commitment view is most plausible if we drop the phrase “insofar as this is possible.” The view is then simply that a group believes p just in case members are jointly committed to believe p as a body. I will ask how similar group belief is to individual belief on this simplified formulation of the joint commitment account. It seems that on this formulation, group belief is not closely parallel to individual belief. Granted, the account does not require for group belief that individual group members are personally committed to a belief. But it does entail that the group members’ joint commitment to believe is necessary and sufficient for belief. Presumably, the account entails that the group’s commitment to believe is necessary and sufficient for belief. This is quite disanalogous to the individual case. For an individual’s commitment to believe p is neither necessary nor sufficient for the individual’s belief.39 Such a commitment is not necessary for the individual’s belief. It is plausible enough that an individual belief p entails various commitments—to reason

Group Belief and Acceptance

87

from p, to defend p from criticism, and the like. But do these commitments add up to a commitment to believe p? If these are sensible commitments to do things, and I can sensibly commit to doing something only if I do not take myself to lack control or influence over whether I do it, then these commitments can add up to a commitment to believe only if I do not take myself to lack control or influence over whether I  believe. But in many normal cases of belief, I take myself to lack control or influence over whether I believe. So I cannot be sensibly committed to believe. Surely I do not normally commit myself to believe in these cases. So there are many cases of belief in which I am not committed to believe. Nor is an individual’s commitment to believe p sufficient for believing p. It would be possible for an individual to commit to belief, whether sensibly or not, without having a chance at bringing about the belief. Since commitment to believing is neither necessary nor sufficient for individual belief, the joint commitment account makes group belief significantly dissimilar to individual belief. I take it to be an important desideratum for an account of group belief that it makes group beliefs and individual beliefs similar. An account that meets the desideratum better explains why we label as “beliefs” both the group and the individual beliefs than does an account that makes them dissimilar. So I take the consequence that group beliefs are dissimilar to individual beliefs to be a substantial drawback of the joint commitment account. But there is a further problem for the account. Our earlier argument that many group beliefs are no more controlled by groups than normal individual beliefs are controlled by individuals now casts doubt on whether, on the joint commitment account, the joint commitment to believe p can be sensible in many cases, and relatedly on whether a joint commitment to believe is necessary for group belief. It is plausible that a similar argument will show that many group beliefs are no more influenced by groups than many individual beliefs are by individuals. To the extent that the conclusion of this argument is recognized by groups, there must be cases of group belief in which the group takes itself not to control or influence the group belief. In such cases, the members jointly take themselves not jointly to control or influence the group belief. But then it is not sensible for the members jointly to commit to believe p. It follows that in cases of group belief in which all joint commitments associated with the group belief are sensible but members take it that they do not jointly control or influence the group belief, there is no joint commitment to believe p. So a group can believe p without there being a joint commitment to believe p. These reflections suggest that the joint commitment account suffers from its reference to a joint commitment and would do better simply to identify the group belief p with the members’ believing p as a body. Such an account says little by itself, but it may be supplemented with the substantive view that all

88

C ollective Attitudes and A ctions

there is to an attribution of members’ believing as a body is an attribution to the group of a state analogous to individual belief. Omitting reference to joint commitment of course eliminates the source of members’ obligations and entitlements that go along with group belief, according Gilbert’s account of group belief. If we wish to omit the reference to joint commitment but also explain why members have such obligations, we must say either that these obligations arise from a source that always or frequently accompanies group belief (such as the members’ putative expressions of willingness to form a joint commitment, apart from whether they really are willing) or that group beliefs frequently occur when such obligations obtain because fulfilling the obligations often brings about the conditions required for group belief. The core of this alternative to the joint commitment account is that an attribution of group belief is an attribution that an organized plurality of individuals is in a state analogous to individual belief. Call this view supraindividualism. In this chapter, I have in effect defended supraindividualism from the objection that an attribution of a group assent cannot sensibly be taken merely to be an analogy to an attribution of an individual assent because group assents differ significantly from individual assents since individuals believe and groups do not. In this chapter, I have maintained that the case of belief poses no threat to a close analogy between individual assent and group assent. I have done so by fending off arguments that groups accept rather than believe propositions. And I have argued to the contrary that groups believe propositions if they accept them. These points contribute to the position that group beliefs and acceptances, insofar as groups have them, are analogous to individual beliefs and acceptances. That position in turn fits a supraindividualist view on which attributions of attitudes and actions to groups are analogies with attributions of attitudes and actions to individuals.40

Notes 1. For example, Cohen (1989, 1992), Meijers (1999), Tuomela (2000), and Wray (2001). 2. For other distinctions between belief and acceptance, see Perry (1980), van Fraassen (1980), Stalnaker (1984, ch. 5), Cohen (1989, 1992), Lehrer (1997, 3ff), Engel (1998), Tuomela (2000), and Buckareff (2004). 3. For such objections, see my (2004a). 4. I sketch supraindividualism in (2004a). 5. It is clear that I can believe a proposition without its being reasonable for me to take it for granted in theoretical reasoning to other beliefs. It might be that I could reason reasonably from my belief p, in combination with other beliefs, to beliefs in many other propositions of theoretical significance (i.e., having interesting content,

Group Belief and Acceptance

89

explanatory power, and the like). But if I lack high confidence in my belief p, I will be loath to draw these consequences. For example, I believe that God exists. From that I could reason reasonably that this is the best of all possible worlds, and from this I could infer all sorts of interesting consequences. But I do not have enough confidence in my belief in God to draw these conclusions. My modest confidence in my belief in God limits my reliance on this belief in theoretical reasoning. In this case, refraining from reasoning from my belief is like refraining from an acceptance of sturdiness in the case of the two-story ladder, mentioned below. For this reason, beliefs resemble acceptances in not agglomerating across episodes of reasoning. The converse situation of taking a proposition for granted in theoretical reasoning without believing it is less clearly possible. Can I reasonably rely on acceptances or assumptions short of belief in theoretical (or belief-concluding) reasoning? The most promising case of this is one in which I simplify my theoretical reasoning by taking for granted something that I do not believe. I do so in order to arrive at any conclusion at all. For my reasoning would otherwise be too complicated to be feasible. Another promising case is one in which I reasonably take for granted something I do not believe in order to arrive at a belief because the stakes for that belief are low. Imagine that the belief is not expected to figure in any further theoretical or practical reasoning. Yet it is questionable whether in either of these cases, given that I reason from a proposition I do not believe, I can reasonably reach a belief as my conclusion. In each case, I have only enough support for my conclusion to form a degree of belief or to accept it as a premise restricted to practical reasoning or to theoretical reasoning to further conclusions not expected to figure in further reasoning. 6. The suggestion that belief in some sense aims at truth is commonplace in the contemporary literature on belief. See, for example, Williams (1973). For an interesting attack on this commonplace, see Owens (2003). I remark below on the meaning of the claim that belief aims at truth. 7. On van Fraassen’s (1980) and plausibly on Lehrer’s (1997, 3ff) distinctions between belief and acceptance, if a subject accepts propositions, then the subject also believes propositions. Van Fraassen applies his distinction to scientific theory: to believe a scientific theory is believe that it is true, whereas to accept is merely to believe that it is empirically adequate. According to van Fraassen, one should only accept and not believe scientific theories. On this view, we may accept a proposition that we do not believe when doing so satisfies a nonveritistic goal (namely, empirical adequacy). But accepting a theory entails believing something else (that it is empirically adequate). If this view is applied to groups, a group cannot accept any theory without holding a belief. By contrast, Lehrer’s distinction between belief and acceptance is between a belief that arises without evaluation of the proposition believed as being worthy of belief and an acceptance that is an evaluation of a belief (or perhaps of a proposition) as being worthy or unworthy to hold (or believe). Lehrer’s view nominally allows acceptance without belief. But it would seem that a plausible account of accepting p in Lehrer’s sense is that it is a belief that the belief p (or proposition p) is worthy or unworthy to hold (believe). On this account, it is undeniable that there are acceptances. Obviously, on this account too, Lehrer’s view does not permit a

90

C ollective Attitudes and A ctions

group to accept anything without believing something. So no issue of whether groups accept rather than believe arises on either van Fraassen’s or Lehrer’s distinctions between belief and acceptance. 8. I mention a possible third, performative notion of acceptance of this sort in note 18. 9. Acceptance in this sense is not relative to a context of reasoning, as Bratman’s acceptance is, but simply occurs in an episode of expression. It is normally done, when overt, to satisfy a practical verbal aim, though when subvocal, it may carry with it an inclination to the truth of p. And it is voluntary. 10. A joint acceptance contrasts with a joint agreement in that the former entails a commitment to act in accordance with a principle of action, while the latter is a joint decision on a course of action. See Gilbert (1989, 373–85). 11. Searle (1995, ch. 4). 12. A rather different case of simplifying reasoning is suggested by Stalnaker (1984, ch. 5). I may simplify my reasoning by simplifying the proposition taken for granted rather than by ignoring the alternatives to it. For example, I  reason practically from the proposition that the table is 4 feet long, though I know it to be 3 feet, 11 7/16 inches by measurement. Here, the proposition I take for granted contradicts the proposition I believe, though it approximates it. See notes 15, 22, and 31 for further discussion of this example. 13. It might be denied that these are cases in which I personally or singularly accept the proposition believed by the others. An alternative view is that I merely accept the proposition jointly with the others. I then do my part or participate in a joint action in light of this joint acceptance of the proposition. What it is for me to do my part in a joint action in light of this joint acceptance might be understood in different ways, as sketched in Searle’s (1990) account of collective intention or in Gilbert’s (1989, 417–27) account of participant reasons for action. 14. If I had no inclination one way or the other, then presumably I could reason simply by assigning a probability of one half to the proposition that it will not rain. 15. The example of note 12 requires an alteration of my treatment of narrow acceptance here. It is not unnatural to assimilate this example to other narrow acceptances on the ground that my acceptance may be described as directed toward the truth. Yet I am certain that p is false. I have a high degree of belief that p is approximately true but a negligible degree of belief that p is true. I should say that what I incline to and indeed believe is the proposition that p is approximately true, rather than the proposition p. It is worth noting, too, that the example of note 12 differs from other narrow acceptances in that which proposition I choose to reason from is under my voluntary control. One might take the example to call for a more complex classification of acceptances than my division into narrow acceptances and mere assumptions. But I will not complicate the discussion by introducing this classification. It is clear that even in this example, acceptance requires belief (albeit belief that the proposition accepted is approximately true); so no subject who lacks beliefs altogether can accept in the relevant sense. 16. It is sometimes said that belief can be either a matter of degree or instead full stop, while acceptance can only be full stop (see, e.g., Tuomela 2000), although we can hold an acceptance more or less firmly. On the view of acceptance I am developing here, this proposed contrast is a bit misleading for narrow acceptances, although

Group Belief and Acceptance

91

not for mere assumptions. A narrow acceptance, in my view, arises from a degree of inclination and a practical consideration. The degree of inclination is a degree of belief. So although a narrow acceptance is full stop, there is an underlying phenomenon that admits of degrees, and a narrow acceptance obtains when a function of that phenomenon meets a threshold. 17. I note that Bratman says only that acceptances need not be mutually consistent, not that they must be mutually consistent within a context of reasoning. 18. I allow that the verb “accept” has a performative use on which, in the right circumstances, I can accept p merely by uttering “I accept p.” Acceptance of this sort may be viewed as an individual action or state, but one that does not require all that Bratman’s acceptance does. Uttering “I accept p” in this performative use does not always or even usually bring it about that one has a disposition to take p for granted in reasoning. The acceptance achieved by the performance is therefore usually a pale shadow of Bratman’s acceptance. One often makes such an utterance in a context of discussion or negotiation in which one wishes to concede that p, despite not believing it or accepting it in Bratman’s sense. Conceding that p entails allowing members of the audience one addresses to take p as a premise in dialogical reasoning with one. It may be that the performative use of “I accept” is often an expression of readiness to take p for granted in group reasoning, and such an expression is intended to contribute to the formation of a joint acceptance with one’s audience. This fits Gilbert’s account of group belief and conversation (1989, ch. 5). 19. I caution that a mere assumption that I hold to coordinate my behavior with that of the group might correspond to my participation in a group acceptance of the same proposition, and that acceptance may be a narrow acceptance rather than a mere assumption. What other group members accept or believe may give rise to a group narrow acceptance in which I  participate, despite not believing. And if this is a group narrow acceptance, it involves the group’s disposition to take the proposition for granted. My mere assumption might even correspond to a group belief in the proposition. 20. Might it be that I must be disposed to defend the view as what it is reasonable to take for granted for purposes of practical reasoning in this context? I think not. It would be enough for me to stand ready, upon taking the view for granted, to defend it in this way. 21. I do not assume here that the beliefs from which we reason to acceptances must themselves derive from theoretical reasoning. So I  do not make the common assumption that practical reasoning depends on prior theoretical reasoning. 22. In the example in note 12, I  reason from a proposition that approximates what I believe. 23. Clarke (1994) argues that accepting p entails believing p. This is mistaken for mere assumptions. But it holds (with qualification) for acceptances—uses (1) and (2). For these, I must have a degree of belief that p whenever I accept p. For the special case of simplifying reasoning described by Stalnaker, where I accept a proposition that is a mere approximation to what I  believe, I  must believe a proposition to which the proposition I accept is an approximation. There are responses to the details of Clarke’s argument in Cohen (2000) and Buckareff (2010). However, these responses do not circumvent my argument that accepting that p entails a degree of belief that p.

92

C ollective Attitudes and A ctions

24. There are, of course, standard arguments for doxastic voluntarism, and to the extent that they establish that view, they undermine the objection that groups do not believe because they lack voluntary cognition. But although these arguments would support my overall conclusion in this chapter, I find them unpersuasive. A typical argument is that we speak of deliberating as to whether to believe p and of deliberating as to whether p, and we take it that believing and refraining from believing p result from such deliberation in many normal cases. Some have inferred from this that belief is sometimes analogous to action with regard to willing. It is natural to think that deliberating as to whether to act and acting or refraining from action as a result of this deliberation entails willing to act or refrain from acting. It is tempting to conclude that deliberating as to whether to believe p and believing or refraining from believing p as a result entails willing to believe or refrain from believing p. But the case of deliberation as to whether to believe seems crucially different from the case of deliberation as to whether to act. After deliberation as to whether to act and recognition that there are adequate reasons for acting, it is still up to us whether to act or not. Acting requires a willing beyond the conclusion of deliberation. But after deliberation as to whether to believe and recognition that there is adequate evidence for p, it is no longer up to us whether to believe or not. I have never heard anyone say, “I see that there is adequate reason, but it’s still up to me whether to believe, and I have decided not to believe.” It is not even clear to me that we normally engage in deliberation as to whether to believe analogous to deliberation as to whether to act. For further arguments for doxastic voluntarism, see Montmarquet (1986), Steup (2000), and Ginet (2001). For a succinct review of arguments for and against doxastic voluntarism, see Vitz (2009). 25. For variants of this argument, see Curley (1975) and Alston (1989). For resistance to Curley’s argument, see Ryan (2003). 26. Ginet (2001, 64–65) offers an example in which, after departing on a road trip, he has only a vague memory of having locked his front door, and so (as I take it) he suspends judgment as to whether his reasons favor believing that he locked the door. Ginet claims that, given the inconvenience of returning to check, he can decide to believe that he locked it. But this is clearly narrow acceptance, not belief. And it would be hasty to describe Ginet as being able to decide to accept the proposition. What he decides is to continue his journey, and what enables him to decide in favor of this action is that he accepts the proposition. But his acceptance of the proposition results from recognizing that his evidence, though inadequate for belief, is substantial enough for acceptance, given the cost of error and of checking. 27. This is not to say that we cannot conceive of beings whose recognition of selection at some time during its effect does not prevent the production of a belief. But their psychology would differ from normal human psychology in this respect. 28. It is a further question whether I  can will a belief in the long run by willfully selecting evidence. The great duration of the deception allows me to foresee in the beginning the result of believing p, but eventually to forget that I had selected the evidence, so that I can believe p. Pascal’s Wager depends on an analogous forgetting that we manipulate ourselves into belief in God. I don’t want to deny that this could and even does happen. If so, there are cases of belief that are like acceptance according to normal acceptance voluntarism.

Group Belief and Acceptance

93

29. A common argument for normal doxastic involuntarism is that belief normally aims at truth or is regulated with respect to truth, and any such attitude must be involuntary. (See Williams (1973) for an elaborate version of such an argument.) Clearly, this argument carries over to narrow acceptances, since these too aim at or are regulated with respect to truth. For yet another argument for doxastic involuntarism, see Scott-Kakures (1994). See note 32 for further discussion of this argument. 30. Evidently, some of my argument in this paragraph applies to a mere taking p for granted in the case of avoiding costly error. Not only can I not will my acceptance that p when my degree of belief is insufficient given the cost of error I assign, but I  also cannot even will to take p for granted in practical reasoning under these conditions. I  cannot bring myself to reason to the action of sitting on the chair from the premise that the chair is sturdy when I am not sufficiently convinced of its truth or do not fear the consequences of error. However, the remaining argument of the paragraph does not apply to a mere taking p for granted. If I am sufficiently convinced of the truth of p or do fear the consequences of the action, then it is up to me whether to take p for granted in reasoning to the action of sitting on the chair, given that I am engaging in reasoning as to whether to perform the action. So I have a partial and asymmetrical control over whether I take p for granted, even though I lack control over accepting p—I cannot take p for granted unless I have a high enough degree of belief, but I can refrain from taking p for granted if I have a high enough degree of belief. 31. Acceptance involuntarism also holds for the example of note 12 as for other cases of simplifying reasoning. I no more control whether to accept that the table is 4 feet long given my assignment of the utility of this approximation than I do whether to believe that the table is 3 feet, 11 7/16th inches long. 32. It is worth asking whether the argument from the aim of truth mentioned in note 29 also carries over to narrow acceptance in the case of avoiding costly error. I think it does. If beliefs normally aim at truth, plausibly narrow acceptances normally aim at what we might call truth in light of the cost of error. Such an aim brings about an acceptance only if the degree of belief is high enough, given the cost of error. Thus, if the argument from the aim of truth succeeds for belief, it equally shows that the subject does not control acceptance in this case. 33. As far as I know, Cohen (1989, 1992) was the first to answer this question affirmatively. His argument for this answer rests on idiosyncratic views of group acceptances and beliefs, which I will not review here. 34. It is obvious that groups can believe on a summative account of group belief, on which a group believes p just in case under conditions of common knowledge, each of its members believes p. A summative account clearly allows group beliefs that aim only at truth and also group beliefs that do not aim only at truth, assuming that aiming only at truth requires that the group is so constituted or regulated that its members tend not to believe p jointly unless p fits the evidence possessed by the group, and this evidence meets certain standards. And the summative account also allows voluntary and involuntary group beliefs, depending on whether the group controls what its members believe. I will not further discuss the summative account. See Gilbert (1989, ch. 5) for criticisms of it. For a review of nonreductive accounts of groups, see my (2004b).

94

C ollective Attitudes and A ctions

35. I assume that some plausible view of what it is for degrees of belief to aim at the truth can be worked out. For example, we may say that degrees of belief assigned values between zero and one aim at the truth when regulated in such a way as to minimize the difference between the value assigned and unity when the relevant proposition is true (see Goldman 1999, 87–94). 36. Is the following a worry for this suggestion? One might think that it is impossible for the group to convince a member to rescind when that member is unmotivated to rescind. A group action to influence members to rescind must be taken in light of joint goals. But a joint goal requires a commitment from all members. Yet an unmotivated member will not join such a commitment. I would reply that the fact that a member is unmotivated to rescind a joint commitment to believe does not entail that the member is unmotivated to join a commitment constituting a joint goal to influence members to rescind. 37. Bratman introduces an analogous qualification in his account of shared intention and shared activity by restricting the action types for which an account is given to those that are “cooperatively neutral”—can be instantiated by individuals acting singularly as well as individuals acting jointly (1999, 96–98). Bratman’s qualification on the face of it excludes an account of shared intention and shared activity for action types, like playing tennis, that are not instantiated by individuals acting singularly. Gilbert’s qualification does not clearly exclude an account of the group activity for these action types since “as a body,” it might be said, applies to playing tennis. Moreover, Bratman’s account suffers from the difficulty that it is not clear how individual intentions that we instantiate a cooperatively neutral action type can mount up to a shared intention that we instantiate a joint activity meeting that action type. Gilbert’s qualification does not obviously suffer from any analogous difficulty. See my (2004a, 151–152) for discussion of Bratman on these matters. 38. Possibly, the phrase “insofar as this is possible” is not intended to allow for the possibility that the group does not literally believe, but rather to recognize that group belief requires a joint commitment and in this respect differs from an individual belief. If this is all that the phrase is intended to do, I see no difficulty with including it. 39. I take it that the qualification “as a body” poses no problem for the comparison: an individual’s commitment to believe p as a body, assuming it is possible at all, is simply the same as the individual’s commitment to believe p. 40. I would like to thank the members of my social metaphysics seminar of 2008 and the anonymous referees for helpful criticisms of this chapter.

References Alston, William P. 1989. “The Deontological Conception of Epistemic Justification.” In William P. Alston, Epistemic Justification: Essays in the Theory of Knowledge, 115-152. Ithaca, NY: Cornell University Press. Bratman, Michael. (1992) 1999. “Practical Reasoning and Acceptance in a Context.” Mind 101(401): 1–15. Reprinted in Faces of Intention: Selected Essays on Intention and Agency, 15–34. Cambridge: Cambridge University Press.

Group Belief and Acceptance

95

———. 1999. “Shared Intention and Mutual Obligation.” In Faces of Intention: Selected Essays on Intention and Agency, 130–41. Cambridge: Cambridge University Press. Buckareff, Andrei A. 2004. “Acceptance and Deciding to Believe,” Journal of Philosophical Research 29: 173–90. ———. 2010. “Acceptance Does Not Entail Belief.” International Journal of Philosophical Studies 18: 255–61. Clarke, D. S. 1994. “Does Acceptance Entail Belief?” American Philosophical Quarterly 31(2): 145–55. Cohen, L. Jonathan. 1989. “Belief and Acceptance.” Mind 98(391): 367–89. ———. 1992. An Essay on Belief and Acceptance. New York: Oxford University Press. ———. 2000. “Why Acceptance that P Does Not Entail that P.” In Believing and Accepting, edited by Pascal Engel, 55–63. Dordrecht: Kluwer. Curley, Edwin. 1975. “Descartes, Spinoza, and the Ethics of Belief.” In Spinoza: Essays in Interpretation, edited by Maurice Mandelbaum and Eugene Freeman, 159–89. LaSalle, IL: Open Court Publishing, 1975. Engel, Pascal. 1998. “Believing, Holding True, and Accepting.” Philosophical Explorations 1(2): 140–51. Gilbert, Margaret. 1987. “Modelling Collective Belief.” Synthese 73(1): 185–204. ———. 1989. On Social Facts. London: Routledge. ———. 2002. “Belief and Acceptance as Features of Groups.” Protosociology 16: 35–69. www.protosociology.de. Ginet, Carl. 2001. “Deciding to Believe.” In Knowledge, Truth, and Duty, edited by Matthias Steup, 63–76. Oxford: Oxford University Press. Goldman, Alvin I. 1999. Knowledge in a Social World. Oxford: Oxford University Press. Lehrer, Keith. 1997. Self-Trust:  A  Study of Reason, Knowledge, and Autonomy. Oxford: Oxford University Press. Meijers, Anthonie. 1999. “Believing and Accepting as a Group.” In Belief, Cognition and the Will, edited by A. Meijers. Tilburg: Tilburg University Press. Montmarquet, James. 1986. “The Voluntariness of Belief.” Analysis 46: 49–53. Owens, David. 2003. “Does Belief Aim at Truth?” Philosophical Studies 115(3): 283–305. Perry, John. 1980. “Belief and Acceptance.” In Midwest Studies in Philosophy, edited by Peter French, Theodore Uehling, Jr., and Howard K. Wettstein, vol. 5, 533–42. Minneapolis: University of Minnesota Press. Ryan, Sharon. 2003. “Doxastic Compatibilism and the Ethics of Belief.” Philosophical Studies 114: 47–79. Schmitt, Frederick F. 2004a. “Joint Action: From Individualism to Supraindividualism.” In Socializing Metaphysics, edited by Frederick F. Schmitt. Lanham, MD: Rowman and Littlefield. ———. 2004b. “Social Metaphysics: An Introduction.” In Socializing Metaphysics, edited by Frederick F. Schmitt. Lanham, MD: Rowman and Littlefield. Scott-Kakures Dion. 1994. “On Belief and Captivity of the Will.” Philosophy and Phenomenological Research 54: 77–103. Searle, John. 1990. “Collective Intentions and Actions.” In Intentions in Communication, edited by Philip R. Cohen, Jerry Morgan, and Martha E. Pollack, 401–15. Cambridge, MA: MIT Press. ———. 1995. The Construction of Social Reality. New York: The Free Press.

96

C ollective Attitudes and A ctions

Stalnaker, Robert. 1984. Inquiry. Cambridge, MA: MIT Press. Steup, Matthias. 2000. “Doxastic Voluntarism and Epistemic Deontology.” Acta Analytica 15: 25–56. Tuomela, Raimo. 2000. “Belief Versus Acceptance.” Philosophical Explorations 2: 122–37. van Fraassen, Bas. 1980. The Scientific Image. Oxford: Clarendon. Vitz, Rico. 2009. “Doxastic Voluntarism.” The Internet Encyclopedia of Philosophy. Accessed April 24, 2009. www.iep.utm.edu/d/doxa-vol.htm. Williams, Bernard. 1973. “Deciding to Believe.” In Problems of the Self, 136–51. Cambridge: Cambridge University Press. Wray, K. Brad Wray. 2001. “Collective Belief and Acceptance.” Synthese 129: 319–33.

4

Against Group Cognitive States R o b e rt D.   R u p e rt

English users are not fazed by such sentences as “Microsoft intends to develop a new operating system” and “England wants to retain the pound as its unit of currency.” We produce and consume such claims frequently and with ease. One might nevertheless wonder about their literal truth. Does Microsoft—the corporation itself—literally intend to develop a new operating system? Does England—as a single body—genuinely want to retain the pound as its unit of currency. More generally, it is a substantive philosophical and empirical question whether groups of individuals (who themselves instantiate mental states) instantiate mental states properly so called. In what follows, I employ a squarely naturalistic methodology in my attempt to answer this question of group, or collective, mental states. Accordingly, the fundamental principle guiding my investigation is this:  if a group has mental states, those states must do causal-explanatory work. On this way of approaching the issue, our practice of attributing mental states to groups— noted above—provides no more than a provocative point of departure. Unless empirical-cum-theoretical considerations can underwrite these attributions, we should adopt an error theory (although, as I argue below, the error involved may be subtle, and acknowledging it may not require that we treat the sentences in question as false). So, we should ask: “Do genuine mental states inhere in groups, considered as individuals?”1 Even the question of the instantiation of group mental states will not be addressed directly. Rather, I  ask whether groups instantiate cognitive states, where cognitive states are, at a first pass, those manifesting properties that distinctively contribute to the causal production of intelligent behavior (the

98

C ollective Attitudes and A ctions

explanandum of cognitive science). My circumscription rests on three considerations. First, in this chapter, I reach a skeptical conclusion regarding group psychological states, and I do not wish to make my case too easily. According to one common conception of the mental—going back at least as far as Descartes—mentality cannot be cleaved from phenomenological experience, that is, immediate conscious awareness of our thoughts or “what it’s like” to have a certain experience. Given how little has been done to show that a group can itself be the subject of a phenomenal state (Rupert 2004, 404n27), my case might be open-and-shut, and thus unenlightening, were I to focus directly on mental states. Second, many of those who advocate extension of mind or cognition beyond the boundary of the skull—and who are so inclined at least partly on naturalistic grounds—take a deflationary view of phenomenal states and their supposed qualia (Clark 2001; Dennett 1996, 2005). So it is only fair when engaging, even indirectly, with such authors to bracket questions about the (supposedly) irreducible qualitative character of mental states, which groups seem so clearly to lack. Third, many philosophers inclined toward naturalism take—rightly I  would say—cognitive science to be the ontologically relevant heir to our mentalistic thinking (cf. Churchland 1981).2 The remainder of the chapter unfolds as follows: first, I give a quick argument against the ontological reality of group cognitive states, focusing on everyday cases; then I respond to a series of objections; and I conclude by arguing that matters do not change when we turn to the causal-explanatory project in cognitive science. 4.1 On Causal-Explanatory Work, and the Lack of It It would, of course, be foolish to pronounce once and for all that group cognitive states have no causal-explanatory work to do. This matter is, after all, an empirical one, and who knows what new data and theoretical advances the future will bring? Nevertheless, from our current perspective, the situation does not look promising for the proponents of group cognitive states, or so I shall argue. The fundamental problem appears to be this:  in the cases in which, in everyday speech, we’re inclined to attribute cognitive states to groups, there seems to be available a complete causal explanation couched in terms of the cognitive states of individuals (together with the noncognitive, physical structures that individuals manipulate and transmit) (cf. Wilson 2004, ch. 12). Take one of the examples mentioned at the outset. At a number of times in recent history, it has seemed that “Microsoft intends to develop a new version of its operating system” was true. Did the alleged intention causally explain any data? If so, which ones? One possibility is the hiring of new employees. Take a particular person—call her “Sally”—who recently received a letter of

Against Group Cognitive States

99

offer of employment from Microsoft and who now has a key to an office on the Microsoft campus and now has funds transferred to her bank account from an account with the name “Microsoft Corporation” attached to it. What explains these data? It seems clear that the occurrence of these events is due entirely to communication among individuals (e.g., members of human resources at Microsoft), cognitive activities of individuals (e.g., each individual on the hiring committee who voted to extend an offer of employment to Sally), and actions of individuals (e.g., the person who transmitted the letter of offer to Sally). It is gratuitous to include an additional cognitive state, a state of Microsoft as a single entity, that causally accounts for the data. 4.2 Does Ockham’s Razor Remove too Much? Goldstone and Gureckis (2009, 426; cf. Huebner 2008) object to this style of reasoning on the grounds that its eliminativist implications overreach in obvious ways. In particular, a parallel application of Ockham’s razor to the case of individuals’ cognitive states would eliminate them from the set of things to which we are ontologically committed.3 After all, there is available, at least in principle, a complete neural causal explanation of intelligent behavior in the individual’s case. By parity of reasoning, then, it is gratuitous to posit individual cognitive states, and they should be eliminated from our ontology. That result is, however, clearly unacceptable. Therefore, my argument against group cognitive states must be flawed (see Rupert 2005, 180, for a statement of this objection and an initial response to it). There are, however, at least two significant points of disanalogy, which undermine the supposed reductio of the casual-explanatory argument against group cognitive states. The first concerns the motive for thinking that individuals have cognitive states. Each of us has what would seem to be decisive first-person evidence of the existence of mental states. For this reason and others, the quick causal-explanatory argument against group cognitive states should be qualified in an important way: P.1. If it cannot be established that group cognitive states do distinctive causal-explanatory work4—and there is no independent reason to think they exist—then we should tentatively conclude that they do not exist. P.2. The antecedent of P.1’s conditional is satisfied. Conclusion. Thus, we should tentatively conclude that group cognitive states do not exist. So qualified, this argument cannot be adapted to the case of individual states in a way that preserves the eliminativist result; for we have independent

100

C ollective Attitudes and A ctions

evidence, the first-person evidence alluded to above, of the existence of individual cognitive states. I do not want to place too much emphasis on the preceding point, given the unreliability of introspective access (Nisbett and Wilson 1977). Keep in mind, though, the weakness of the demand made by the independent-evidence clause. It may be reasonable to doubt specific claims of first-person access, for example, of direct access to the particular motive that caused some instance of one’s behavior. The issue at hand, however, pertains to the very existence of cognitive states. Although we frequently are wrong about the details of our cognitive processes, the empirical evidence does not seem at all to undermine our confidence that we have some cognitive states or other. Consider one way of arguing for such minimal introspective access. Many have claimed, largely on empirical grounds, that humans simulate other persons’ thought processes as a way of predicting and understanding others’ behavior: we put ourselves in their shoes, so to speak, and reason away to see what we would do. On the simulation-based view, broadly speaking, we have first-person access to our own case, and then project it onto others. At least on most ways of fleshing out this view (e.g., Nichols and Stich 2003), the model presupposes some access to one’s own states (and see ibid., 172 for some experimental evidence of such access). A verbal report of a prediction of another’s behavior is, for instance, a report of the output of one’s own cognitive processes. If this view of mental-state attribution is correct, a certain amount of introspective access—sufficient for present purposes—seems to follow. Moreover, even if one endorses the theory-theory of mind, the everyday (i.e., first-order, not philosophical) success of our use of that theory provides independent evidence of the existence of individual cognitive states. I am not making a point about first-person access. The theory-theory doesn’t presuppose first-person access to one’s cognitive states; one’s own cognitive states are treated as theoretical entities just as are everyone else’s (Gopnik 1993). We clearly succeed, however, at predicting and explaining others’ behavior (Fodor 1987). If an application of the folk theory of mind grounds such predictive and explanatory success, and that theory is partly constituted by talk of cognitive states, the success in question itself supports a claim to the mere existence of individual cognitive states. Setting aside questions about introspective access and the theory of mind, I want especially to emphasize a second point of disanalogy between arguments for the elimination of group cognitive states and those for the elimination of individual cognitive states. Goldstone and Gureckis’s worry fails to distinguish between intra- and inter-domain elimination. I have argued for a kind of intra-domain elimination: we should eliminate group cognitive states because of the availability of other cognitivist explanations of the relevant

Against Group Cognitive States

101

data—those invoking the cognitive states of individuals. In general, once we have reason to introduce a certain kind of property, then we should eliminate redundant or otherwise gratuitous instances of that kind of property. It is simply a matter of constructing the most parsimonious theory of that particular domain. In the case at hand, the theory is psychology or perhaps cognitive science. The objection under consideration presupposes that this intra-domain strategy can be freely extended to the inter-domain case. In the inter-domain case, though, we face a substantially different question: why should we introduce, or preserve commitment to, a distinctive kind of property or domain of phenomena at all? The most common way to defend psychology against such wholesale eliminativism appeals to distinctive patterns—patterns in intelligent behavior that have no theoretically natural expression outside of psychology. Psychology is a relatively successful endeavor and thus should be taken to identify distinctively psychological properties. One might hope for reduction in the face of this, but such reductions do not seem forthcoming given the (perhaps open-ended) variety of realizers of psychological properties or kinds (Fodor 1974). Therefore, we have independent evidence of the existence of cognitive states: the evidence against the reducibility of psychology to neuroscience (or some other nonintentional science). One might wonder, then, whether any distinctively intra-domain principles might block my appeal to Ockham’s razor. (Such a principle could be thought of as providing, when applied, independent evidence for the existence of group cognitive states, thus neutralizing the first premise in my causal-explanatory argument.) Further possibilities are considered in the next section, but for the present, consider just this one: our best theory of individuals’ cognitive states (or properties) itself provides independent reason to accept the existence of group-level cognitive states. Within a domain, once one has posited certain distinctive properties or state-types—as a way of doing necessary causal-explanatory work—certain instances of those properties might get an ontological free pass, so to speak. Here is an analogy. The mass of, say, a rigid body is a straightforward agglomeration of the masses of the body’s component parts. Now, it might be that, according to the most defensible metaphysics of a body’s mass, it (the mass of the body) has no causal power of its own. From the standpoint of physics, though, this would seem to be irrelevant. The theory of mass itself entails that the rigid body (the object occupying the space-time region in question) has a mass. Perhaps groups, similarly, possess cognitive states because our best theory of cognition—of, say, the structure or information-processing steps distinctive of the states that do causal-explanatory work in individuals—entails that groups instantiate such states (given the empirical facts about groups).

102

C ollective Attitudes and A ctions

Regardless of whether those group states do distinctive causal-explanatory work, it might be that the groups in question simply possess the right internal structure or information-processing properties or that they enter into the right apparent causal relations (even if, according to some deeper philosophical analysis, the relations should not be counted as genuinely causal). We would then seem to be ontologically committed to the existence of group cognitive states because we are justified in accepting a theory that entails the existence of cognitive states wherever certain structural properties or apparent causal patterns appear, and these happen to appear in groups. This tack does not seem at all promising as a way to resist the causal-explanatory argument against the existence of group cognitive states. It is a demanding task to find group states that have the properties of cognitive states, as these are described when posited to do causal-explanatory work in the science of individual cognition. For example, representation plays a central role in individual cognition, yet typical groups do not have mental representations of the sort the positing of which has proven useful in cognitive science (see Rupert 2005 for a detailed discussion of mental representation and groups). This immediately precludes the defense currently under consideration with regard to any content bearing cognitive states—which would seem to be nearly all of them. Moreover, when one attends to the structural, information-processing, and (apparent) causal profiles of just about any particular kind of individual cognitive state—say, having items in a short-term memory buffer—the prospects for group states seem dim, at least as their existence might be established by the current strategy.5 Consider the way in which a human performs actions. Typically, a person executes a selected action by executing other actions, which all together constitute a chain of actions grounded in the execution of a relatively independent motor routine (Grush 1997, 2004): I warn the police by flipping the light switch; I flip the light switch by reaching for it—and here the chain ends with the activation of a motor routine. In the case of groups, although there may be a more or less centralized action-selection mechanism, the actions bottom out in the actions of individuals. The basic actions of humans—those that are not executed by any other actions—are performed via the activation of relatively autonomous motor routines. In contrast, for the typical group, insofar as it has what might be thought of as basic actions, these are executed by a system that itself is made up of both a more or less centralized decision-making system and relatively autonomous motor routines (both in the individual).6 There could be exceptions to this generalization: an automated factory might execute the actions of a corporate entity in a way that is architecturally similar to that of the typical individual human. My point, however, is that architectural

Against Group Cognitive States

103

dissimilarities are some of the many that stand in the way of the “side-effect” defense of group cognitive states currently under consideration. I should reiterate the point of the preceding comments. I have not offered an argument of the following form: group cognitive states exist only if such states have the properties of individuals’ cognitive states; they do not have such properties; therefore, they do not exist. Absolutely not. Rather, I have been criticizing the argument that even if group states do not have any causal-explanatory work to do, they exist so long as they have the same profile as individual states; they do have the same profiles; therefore, they exist. Given the differences between group-level states and individual ones, the second premise is false. So the differences to which I have directed the reader’s attention do not ground an independent argument against group states; instead, they ground a criticism of a certain strategy for getting around my original causal-explanatory argument against group cognitive states. To be fair, Huebner (2008, 101–05) develops a more subtle version of the rejoinder from overreaching. He argues that there must be something wrong with my original eliminativist argument because, even in the case of individual cognition, component parts of the individual cognitive system—subpersonal representations, in particular—carry content that is also the content of the personal-level cognitive states. To my mind, this version of the objection appeals to an indefensible notion of personal-level states. Individual cognition simply is the activity of the component mechanisms of the cognitive system housed in the individual organism (see Rupert 2009, ch. 3 for more detailed discussion). On this view, there is no distinct personal-level at which we find states of the sort that, according to Huebner, my casual-explanatory argument would eliminate (unacceptably). It is no objection to my causal-explanatory argument that, if its logic were applied across the board, it would eliminate personal-level states; for we already have independent reason to eliminate them, at least if they are supposed to be distinct from collections of subpersonal ones. 4.3  Further Objections and Replies The remainder of this chapter addresses further objections, some found in the literature and some raised by audiences members.

4.3.1 Distinctive Patterns at the Group Level Above, I rehearsed a standard argument against the elimination of individual cognitive states, an argument that appeals to distinctive patterns that can be captured only at the psychological level. One might wonder then why such

104

C ollective Attitudes and A ctions

considerations would not themselves speak in favor of a commitment to group cognitive states. I take it that the ontological importance of distinctive patterns concerns the recognition of a distinctive kind of phenomena. But when distinctive patterns involve straightforward combinations of patterns of entities of the same kind, the mystery is solved; there is no need to think that a new domain of inquiry must be opened up. Recall the famous multiple realizability arguments against the reduction of psychological properties to physiological ones. According to such arguments, distinctive kinds of facts appear at the psychological level because there is a pattern that is only explicable by a hodgepodge of facts in a distinct domain. In the case of alleged group cognitive states, we are not faced with a domain of facts that can be explained only awkwardly by reference to odd groupings of properties from a distinct domain. Digression:  The objection currently on the table, and related objections raised below, is cast in commonly accepted terms; this is perhaps the best way to proceed from a dialectical standpoint. I am more inclined, however, to think that realization relations tie distinct domains together too tightly—engendering problems of overdetermination, causal exclusion, and, in some forms, the virtus dormativa problem (Rupert 2006, 2008), in which case, realization relations, within a level or otherwise, count against the ontologically autonomous status of the states in question (in this case, group cognitive states). On this view, ontological commitment depends on whether distinctive patterns are not realized by (or do not supervene on) same- or different-domain states. The realization relation is too tight to establish the existence of the realized states. At least in the standard examples, group cognitive states almost certainly do not exhibit the requisite form of distinctiveness (i.e., they are not distinctive qua not being realized by anything else).

4.3.2  Within-Domain Realization The immediately preceding remarks disparage realization relations as a ground for autonomy, but let us set these concerns aside, treating them as motivated by idiosyncratic metaphysics. It is much more common to think of the realization (or supervenience) relation as the appropriate relation to hold between properties in distinct domains, each of which is ontologically autonomous. Moving back beneath the umbrella of orthodoxy, then, consider the following objection. Why shouldn’t we think that individual cognitive states realize group-level cognitive states? On a standard view of realization, this realization relation would allow both individual and group cognitive states autonomous status

Against Group Cognitive States

105

while accounting for the way in which individual states seem to determine group states. My reaction to this thought parallels my reaction to the first objection. To me, it seems to be based on a misunderstanding of the very purpose of talking, in philosophy of science, about realization relations: the theory of realization provides a way to understand the relations between different domains, that is, to make sense of the relation between different sets of relatively autonomous but internally nomically related properties. The idea of realization is meant to illuminate both the way in which such domains are autonomous while at the same time being part of a single, ontologically coherent (physical) universe. To hold that individual cognitive states realize group cognitive states—all states within the same domain of cognition—muddies such a picture to no good end. At the very least, much independent empirical motivation would be required for such a view, and it is part of my brief that there is no such motivation.

4.3.3 Aren’t Many Sentences That Attribute Cognitive States to Groups True? Consider again the sentence: “Microsoft intends to develop a new version of its operating system.” Perhaps when we utter this sentence, we defer in some way to facts about individual employees at Microsoft (and their use of merely physical media). Why, however, should we not take such facts to be truth-makers of the sentence at issue? If they are, then Microsoft has a cognitive state; it has an intention. How could it be true that Microsoft intends to P if it does not have an intention to P? This objection does not speak to the question in which I am ultimately interested. I take a state to be the instantiation of a property by an individual. Thus, if the statements in question are true, but their truth-makers involve nothing more than the instantiation of cognitive properties by individuals (together with their use of merely physical materials), then there are no group cognitive states in the relevant sense. No group, as an individual, instantiates a cognitive property. For comparison, consider the following sentence:  “The average American family consists of 3.14 persons.” As of the 2000 census, this sentence (or something near enough to make my point) was true (see http://factfinder2. census.gov/faces/tableservices/jsf/pages/productview.xhtml?pid=DEC_00_ SF1_P033&prodType=table). But there is no individual family that instantiates the property of being comprised of 3.14 persons. The truth of the census-based claim no more entails the existence of partial persons than does the truth of “Microsoft intends to develop a new version of its operating system” entail that

106

C ollective Attitudes and A ctions

Microsoft itself instantiates the causal-explanatory (or natural) cognitive property intending that.

4.3.4  What about Shared, or Collective, Intentions? It might seem simply obvious that there are shared, or collective, intentions. When we work together to move a heavy table, we intend to move the table. This is all well and good, but the most promising analyses of our concept of group intentions do nothing to establish group cognitive states. Take, for example, Natalie Gold and Robert Sudgen’s (2007) recent account, which emphasizes the individual’s identification with a group or commitment to the goals of the group. Such an analysis might apply to cases in which people are willing to work together to bring about certain goals—that is, to contribute causally to the production of what we might think of as data to be explained. It does so, however, in a way that coheres with my response to the preceding objection. It explains how sentences that ascribe group intentions might be true without the group’s instantiating the properties that appear in the causal-explanatory generalizations (or laws) of our best cognitive science. I might care about the fortunes of a group (e.g., the number of games the group wins by some particular metric), and I might believe that others on my team have the same concern; and the others on the team might, in fact, instantiate these same cognitive states. That collection of facts exhausts the cognitive materials required to explain the relevant outcome:  the team’s number of wins— insofar as team spirit contributes to that outcome. Similar comments apply to the entire range of analyses according to which individuals group members’ beliefs (or knowledge) about the beliefs (or knowledge) of other members of a group determine collective intentions (see, e.g., Chant and Ernst 2008, and many of the proposals surveyed therein).

4.3.5 The Effects of Groups on Intelligent Behavior It seems indisputable that being in a group can affect the individual’s behavior. Thus, the group itself plays a causal role in the production of intelligent behavior—that of the individual at least. Absent a principled way of identifying only the genuinely cognitive contributors to intelligent behavior, such causally contributing group states count as cognitive. Consider, though, what might be thought of as the independent cognition-related status of the relevant group state—its status apart from its causal contribution to the intelligent behavior of the individual. At least as I presented the objection above, there is nothing particularly cognitive about the group states; the mere perceived physical presence of a group (by the

Against Group Cognitive States

107

individual) affects the individual’s behavior. Seen in this light, then, one might think there is nothing specially cognitive about the group. If it is cognitive, then a tree perceived by an individual counts as being in a cognitive state; for when perceived, it can causally contribute to the production of intelligent behavior on the part of the individual—perhaps her discourse on the tree’s many important biological properties. This alone might ground a straightforward response to the objection at hand. The group’s contribution to the production of intelligent behavior is on par with the contribution of such things as perceived trees. Given that the latter is clearly not, simply on account of its causal contribution, in a cognitive state, neither is the group in a cognitive state. Larger matters loom, however, including the question of what is to count as cognition. Elsewhere (Rupert 2009; cf. Wilson 2002), I have argued at length for the following view: a state is cognitive if and only if it is the state of a mechanism that is part of a relatively persisting, integrated set of mechanisms that produces intelligent behavior. In the parlance of contemporary cognitive science, the integrated set of mechanisms is a cognitive architecture; this appears to be the fundamental explanatory construct of cognitive science, whether one’s proclivities run toward computationalist, connectionist, or dynamicist forms of modeling. Limiting cognition to the activities of the cognitive architecture would seem obviously to exclude the tree from the collection of things in cognitive states. Moreover, in most (perhaps all) commonly cited cases, it precludes the group from doing so as well. This approach raises a new question, though: do groups instantiate a cognitive architecture? This, I think, is an open empirical question. Nevertheless, two concerns should temper our optimism about finding architectures in groups. First, the architecture is an integrated collection of mechanisms that plays a privileged role in causal-explanations of intelligent behavior, that is, in accounting for the distinctive explananda of cognitive science. To find a cognitive architecture in a group thus requires the identification of relevant explananda. And although I make some effort below to do this, it is no easy task to locate such data. Second, the positing of cognitive architectures seems to me to be constrained by the same intra-domain principles of elimination discussed above. In particular, in cases in which there are existing cognitive architectures, the operation of which account for the data in question, it is gratuitous to add another. To be more concrete, if we already have reasons to posit cognitive architectures in individual employees of Microsoft, for example, and the activities of those cognitive architectures account for the data, then it is gratuitous to invoke the contribution of some further cognitive architecture. If we assume that the side-effect considerations discussed above do not secure the existence of group-level cognitive architecture, this concern about redundancy eliminates

108

C ollective Attitudes and A ctions

them. Of course, this leaves open the possibility that group-level architectures do distinctive causal-explanatory work, a matter I return to in closing.

4.3.6 Groups as Subject to Norms Much philosophical work has stressed the importance of norms. If groups are subject to norms of evaluation—for instance, of the rationality of their behavior—this might itself establish the ontological status of groups as cognitive agents. Norms, however, do not appear from nowhere. The institution and enforcement of norms are the work of individuals. Individuals must invent norms, they must learn norms others have invented, they must choose to (and agree to) be bound by them, they must choose to write down statements (policy statements, for example) that they themselves or other individuals can access later (and they must decide, as individuals, whether to act in keeping with explicitly formulated norms), and so forth. Cognitive science explains these facts by adverting to the standard mechanisms of individual cognition, and it seems no surprise that it would do so. It is difficult to see any other naturalistically viable option, at least at present. 4.4  Causal-Explanatory Power Revisited Up to this point, the discussion has proceeded mostly at an abstract level; where examples have been used, they involve everyday attributions of mental states to groups. Is there, however, any cognitive scientific evidence for the causal-explanatory role of group states—evidence that would sink the simple argument around which discussion has been centered? Our initial task is to identify something in the vicinity of intelligent behavior on the part of group. Here is a possibility that I think will be instructive. An existing body of literature on foraging identifies what are perhaps surprising patterns in the distribution of individuals across food sites. The basic result of interest is that under many conditions, groups of animals distribute themselves in a way that is very nearly proportional to the distribution of food across sites. To keep things simple, if 80 percent of a quantity of food appears in one site and 20 percent at another, then roughly 80 percent of the animal population in the vicinity go to the first site, and 20 percent go to the second. Thus, we might think of the population as a whole as engaging in a kind of problem solving: efficient consumption of a resource (Goldstone and Gureckis 2009, 423–24). If efficient consumption of a resource is itself intelligent behavior, then, one might think, whatever states explains it count as cognitive, regardless of whether those states’ causal or structural profiles resemble those of the states that explain the intelligent behavior of individuals. What might the relevant

Against Group Cognitive States

109

state of the group be? The obvious candidate is the distribution of the group in space—or perhaps a series of successive distributions of the group in space. Here again, however, the group state plays no causal-explanatory role. The distribution of the group does help to explain why the food gets consumed efficiently. Nevertheless, there seems to be a complete causal, and clearly cognitive, account of the distribution. Individual organisms themselves pursue a strategy—implemented in the standard ways by perception, computation, representation, and the like—that determines the relevant distribution of the group. Consider the underpinnings of optimal foraging in individuals. The frequency with which, for instance, an individual great tit visits a given food site, relative to the frequency which it visits other food sites, is itself proportional to the amount of food the bird has found at that site, relative to the other sites. Release one hundred great tits, then, and the probabilistic workings of their individual cognitive mechanisms lead unmysteriously, by simple aggregation, to the group’s distribution that eventuates in efficient consumption. These results hold for various species of fish, fowl, and rodents (for review and citations, see Gallistel et al. 1991, 18–21). 4.5 Conclusion Nothing I have said precludes the existence of group cognitive states. Nor have I tried to show that no noncognitive group states do causal work. Rather, I have argued that the naturalistically minded proponent of group cognitive states must meet a significant burden—one that, at present, does not seem to be met, regardless of whether we focus on cases of interest to the folk or those of interest to cognitive scientists.7

Notes 1. I frame the question in this way for the following reason. I do not think the truth of a sentence of the form “things of kind P exist” has immediate ontological import; that is, I do not think it entails the presence of things of kind P among the furniture of the universe (cf. Armstrong 1989, 89). Of central importance is what makes such a sentence true. There may, for example, be a legitimate meaning of “group mental states exist” in keeping with which the sentence expresses a true proposition, even though it is never the case that groups, taken as individuals, instantiate mental properties. Further discussion will follow. 2. Note, too, that a negative answer in the case of group cognitive states may well entail a negative answer to the question of group mental states. It is somewhat plausible that where there are mental states, there is cognition—from which it follows that where there are no group cognitive states, there are no group mental states. 3. Throughout, when I  use “individual” without qualification, I  have in mind individual organisms (in most cases, humans).

110

C ollective Attitudes and A ctions

4. In this context, to do distinctive causal-explanatory work is to provide a causal account of data for which there is no other, more entrenched causal explanation. 5. Bear in mind that it is of little consolation to the proponent of group cognitive states to find a relatively small amount of match in casual profiles. Normally, a cognitive state that has been posited to do causal-explanatory work in the human case exhibits a wide range of theoretically important interactions with other cognitive states. 6. This suggests a more a prioristic argument against group cognitive states, at least for typical groups: any intentional action of system S must be executed via one of S’s basic actions (where a basic action is an action not executed by the performance of any other intentional action); collective systems execute no basic actions (all of their actions are executed via the basic actions of other intentional systems); therefore, collective systems perform no intentional actions. 7. Earlier versions of this chapter were presented at Cornell University, Bielefeld University, and at the Universities of Utah, Pittsburgh, and Colorado. Thanks to all of the audiences for stimulating questions and discussion. Whether for their reactions as audience members or as interlocutors in other contexts, I  would particularly like to thank Amber Arnold, Margaret Battin, David Boonin, Sara Rachel Chant, Chris Heathwood, Harold Hodes, Bryce Huebner, Mitzi Lee, Elijah Millgram, Sandra Mitchell, Alastair Norcross, Sydney Shoemaker, Mark Sprevak, and Mariam Thalos, as well as all of those whom I have forgotten.

References Armstrong, D. M. 1989. Universals: An Opinionated Introduction. Boulder, CO: Westview. Chant, S. R., and Z. Ernst. 2008. “Epistemic Conditions for Collective Action.” Mind 117: 549–73. Churchland, P. M. 1981. “Eliminative Materialism and the Propositional Attitudes.” Journal of Philosophy 78: 67–90. Clark, A. 2001. Mindware:  An Introduction to the Philosophy of Cognitive Science. Oxford: Oxford University Press. Dennett, D. C. 1996. Kinds of Minds:  Toward an Understanding of Consciousness. New York: Basic Books. ———. 2005. Sweet Dreams:  Philosophical Obstacles to a Science of Consciousness. Cambridge, MA: MIT Press. Fodor, J. 1974. “Special Sciences, or the Disunity of Science as a Working Hypothesis.” Synthese 28: 97–115. ———. 1987. Psychosemantics:  The Problem of Meaning in the Philosophy of Mind. Cambridge, MA: MIT Press. Gallistel, C. R., A. L. Brown, S. Carey, R. Gelman, and F. C. Keil. 1991. “Lessons from Animal Learning for the Study of Cognitive Development.” In The Epigenesis of Mind:  Essays on Biology and Cognition, edited by S. Carey and R. Gelman, 3–36. Hillsdale, NJ: Laurence Earlbaum. Gold, N., and R. Sudgen,. 2007. “Collective Intentions and Team Agency.” Journal of Philosophy 104: 109–37. Goldstone, R. L., and T. M. Gureckis. 2009. “Collective Behavior.” Topics in Cognitive Science 1: 412–38.

Against Group Cognitive States

111

Gopnik, A. 1993. “How We Know Our Minds: The Illusion of First-Person Knowledge of Intentionality.” Behavioral and Brain Sciences 16: 1–14. Grush, R. 1997. “The Architecture of Representation.” Philosophical Psychology 10: 5–23. ———. 2004. “The Emulation Theory of Representation: Motor Control, Imagery, and Perception.” Behavioral and Brain Sciences 27: 377–96. Huebner, B. 2008. “Do You See What We See? An Investigation of an Argument against Collective Representation.” Philosophical Psychology 21: 91–112. Nichols, S., and S. P. Stich. 2003. Mindreading:  An Integrated Account of Pretence, Self-Awareness, and Understanding Other Minds. Oxford: Oxford University Press. Nisbett, R., and T. Wilson. 1977. “Telling More Than We Can Know: Verbal Reports on Mental Processes.” Psychological Review 84: 231–59. Rupert, R. 2004. “Challenges to the Hypothesis of Extended Cognition.” Journal of Philosophy 101: 389–428. ———. 2005. “Minding One’s Cognitive Systems:  When Does a Group of Minds Constitute a Single Cognitive Unit?” Episteme:  A  Journal of Social Epistemology 1: 177–88. ———. 2006. “Functionalism, Mental Causation, and the Problem of Metaphysically Necessary Effects.” Noûs 40: 256–83. ———. 2008. “Ceteris Paribus Laws, Component Forces, and the Nature of Special-Science Properties.” Noûs 42: 349–80. ———. 2009. Cognitive Systems and the Extended Mind. New  York:  Oxford University Press. Wilson, M. 2002. “Six Views of Embodied Cognition.” Psychonomic Bulletin and Review 9: 625–36. Wilson, R. A. 2004. Boundaries of the Mind:  The Individual in the Fragile Sciences. Cambridge: Cambridge University Press.

5

The Ontology of Collective Action K i r k   L u dw i g

5.1 Introduction What is the ontology of collective action? I  have in mind three connected questions. 1. Do the truth conditions of action sentences about groups require there to be group agents over and above individual agents?1 2. Is there a difference, in this connection, between action sentences about informal groups that use plural noun phrases, such as “we pushed the car” and “the women left the party early,” and action sentences about formal or institutional groups that use singular noun phrases, such as “the United States declared war on Japan on December 8, 1941” and “the Supreme Court ruled that segregation is unconstitutional in 1954 in Brown vs. Board of Education”? 3. Under what conditions does it make sense to speak of a group doing something together, and what, if anything, is a collective action? These questions are of interest in themselves, and they are of interest also for the light their answers promise to shed on the nature of social reality and on the difficult question of the nature and distribution of moral responsibility for the consequences of group action.2 In the following, I argue that a) understanding action sentences about groups does not commit us to the existence of group agents per se, but only to the existence of individual agents;

Ontology of Collective Action

113

b) there is no difference in this regard between sentences that attribute actions to informal groups on the one hand and institutional groups on the other; c) collective action can be both intentional and unintentional; d) any random group of agents, each of whom does something, is also a group that does something together; e) although there is a sense in which groups per se perform no primitive collective actions, and therefore no actions at all, f) there is a sensible extension of talk of actions to groups, though it should be treated, strictly speaking, like talk of group agents, as a façon de parler, for g) the only agents per se are individuals, and the only actions are theirs.3 In section 5.2, I argue that an analysis of the logical form of plural action sentences shows that plural action sentences are not committed to the existence of group agents.4 In section 5.3, I argue that the account of the logical form of plural action sentences can be extended to institutional action sentences, which employ grammatically singular subject terms. In section 5.4, I argue that groups do things intentionally and unintentionally, that a group can do something that is not intentional under any description, and that for any random group of agents, all of whom do something, there is something the group they constitute does. In section 5.5, I argue that since groups are not per se agents, they do not per se perform actions, but that there is a sensible way to extend talk of primitive actions to groups, and so of actions, though it should be treated as a façon de parler. 5.2 The Logical Form of Plural Action Sentences Prima facie the only difference between [1]‌and [2] below is that a plural referring term occupies the subject position in [2], where a singular referring term does in [1]. [1]‌ I built a boat [2]‌We built a boat This suggests that, just as the agent in [1]‌is the referent of the subject term, so in [2] the referent of the subject term is the agent. We can show that this is a mistake, however, by attending to an ambiguity in [2] and how to project the standard event analysis of singular action sentences to the plural case in its light. [2] has a distributive and a collective reading. [2] is true on the

114

C ollective Attitudes and A ctions

distributive reading if for each of us there is a boat that he built. On the collective reading, [2] involves us building a single boat together. The event analysis of singular action sentences treats the verb as introducing an implicit quantifier over events. The matrix of [2] may then be represented as follows, where “t” ranges over time intervals, “t*” is an indexical that denotes the time of utterance, and “ t*][The x: x is(t) the forty-fourth President of the United States] [Et′: t′ > t* & t′ is on January 20, 2009](x is(t′) inaugurated) [4]‌has two readings. If we are talking about someone in view at the moment, the nominal may be intended to be true now of a unique person as in [4a]; if we are taking about what happened at a party last night, it may index to the event time of the main verb as in [4b]. [5] will typically be interpreted as about someone who uniquely satisfies at the time of utterance the nominal as in [5a], though if I assert it after we have been looking at a photograph taken last year, it will be understood as involving a time in the past of the present as in [5b]. In

Ontology of Collective Action

119

[6], the sense of “fugitive” requires, if we take the speaker to be competent and rational, that it index to a time in the past of the event time of the main verb, as in [6a]. In [7], if uttered before January 20, 2009, the nominal must be interpreted relative to a time in the future of the utterance in order to pick out the right person, given that the forty-fourth president began his first term in office on January 20, 2009, as in [7a]. Now consider [8]. The entity we have in mind is not any particular group of nine justices who have served on the Supreme Court at some point in its history, but the entity that has had various members (justices serving on it) over its history. We can treat this as the group whose members (in the generic sense) are everyone who has been a member of (in the sense of having been officially appointed to and served on) the Supreme Court, since the Supreme Court at any given time consists of exactly those individuals who are the justices at that time. The nominal formed from “Supreme Court (of the United States)” has an argument place for time, as shown by: there was no Supreme Court (of the United States) in 1700, but there was one in 1800. Since one can utter [8] without knowing when the Supreme Court was created and how long it will last, I suggest that we simply read it as involving an existential quantifier over a time interval. Thus, we can interpret [8] as in [8a], where I use a capital letter as a variable taking groups of individuals as values.10 [8] Four women have served on the Supreme Court. (a) (∃t)[The X: X is(t) a Supreme Court (of the United States of America)] [∃t´: t´ ≤ t*](served-on(X, four women, t´))11 Let us turn now to an action sentence involving the “the Supreme Court.” [9] The Supreme Court went to lunch. [9] has both a distributive and a collective reading. On the distributive reading, each of the nine justices of the Supreme Court went to lunch, but not necessarily together. On the collective reading, they went to lunch together. In uttering [9], on either the distributive or collective reading, however, we do not mean that everyone who was ever a member of the Supreme Court went to lunch. Rather, we have in mind the members who are justices at the time of utterance. On the distributive reading, then, the quantifier has to be restricted to members of the group who are justices at the time of utterance. This is shown in [9d], where we use “εt” to express the socially constructed membership relation, that of serving on the court at t—that is, having been appointed and officially approved and not having resigned or been removed at t. This is to be distinguished from the generic sense of membership in a group

120

C ollective Attitudes and A ctions

(for which we have used “∈”) in which for any random selection of individuals there is a group G of which each of them is a member (each x in the selection is such that x ∈ G). When I mean to speak of a socially constructed membership relation or of members understood in terms of such a relation, I  will speak of ε-membership or ε-members. I  omit the temporal argument place in the agency relation in the uniqueness condition, and this should be read as only members of X are agents in the relevant way of e at any time. We get the collective reading in [9c]. [9d] [∃t][The X:  X is(t) a Supreme Court][Each x εt* X](∃e)[∃t′ < t*] (agentG(x, e, t′) and [only y εt* X]agentG(y, e) and going(e) and to(e, lunch)) [9c] [∃t][The X:  X is(t) a Supreme Court](∃e)[Each x εt* X][ ∃t′ < t*] (agentG(x, e, t′) and [only y εt* X]agentG(y, e) and going(e) and to(e, lunch)) This gives us the tools we need to interpret sentences about the Supreme Court involving its institutional actions. We give an analysis first of [10] in [10c]. Since “ruled” expresses an action type that can only be performed intentionally by an appropriately constituted institutional group, only the collective reading can be rationally intended. [10] The Supreme Court ruled in Plessey vs. Ferguson in 1896 that segregation is constitutional. [10c] [∃t][The X: X is(t) a Supreme Court](∃e)[∃t′ < t* & t′ lies in 1896] [Each x εt′ X](agentR(x, e, t′) and [only y εt′ X]agentR(y, e) and ruling(e) and content(e, segregation is constitutional) and in(e, Plessey vs. Ferguson)) The key here is that since the ruling is located in the past of the present and in 1896, we must let the temporal argument place in the membership relation be bound by the quantifier introduced by the past tense marker on the verb. Now we can show how to handle sentences that attribute to the Supreme Court actions at different times at which there is no overlap in membership. Consider [11]. [11] The Supreme Court ruled in 1896 that segregation is constitutional, but in 1954, it reversed itself and ruled that segregation is not constitutional. [11c] [∃t][The X: X is(t) a Supreme Court](∃e)[∃t′ < t* & t′ lies in 1896] [Each x εt′ X](agentR(x, e, t′) and [only y εt′ X]agentR(y, e) and ruling(e) and content(e, segregation is constitutional) and (∃e′)[∃t′′ < t* & t′′ lies in  1954][Each x εt′′ X](agentR(x, e′, t′′) and [only y εt′′ X]agentR(y, e′) and reversal(e′, e, X) and ruling(e′) and content(e′, segregation is not constitutional)))

Ontology of Collective Action

121

This shows that the fact that we speak of institutional groups doing things at different times does not commit us to group agents per se and that the account of the logical form of plural action sentences extends straightforwardly to singular group action sentences. We can also now explain the appearance that institutional groups like the Supreme Court can change their membership over time. There is something of an illusion here, generated by the fact that there are different senses of “membership” in play. When we speak of the Supreme Court as such, as in [8], it is the group of individuals who are its ε-members at any time that we have in mind. That group does not change its membership over time, though not all of its members exist at the same time (think of a family extending over generations). But which individuals bear the socially constructed ε-membership relation to the Supreme Court, the serving-on relation, is different at different times, and in that sense of “membership” we can speak of the membership of the Supreme Court changing over time (a member of it in the first sense may resign and so cease to be a member of it in the second sense, but once a member in the second sense, one is always a member in the first sense—that is, a member can resign and so cease to be an ε-member, but once an ε-member, one is always a member). This is clearly compatible with all action sentences about what the Supreme Court does being understood in terms of the agency of its then appointed ε-members. Finally, to deal with modal statements involving singular group action sentences, we have only to note that the fact that the Supreme Court could have had different members in toto and different ε-members at any time is simply a reflection of the fact that “the Supreme Court” is a definite description and not a proper name, that is, that it is not a rigid designator. The fact that the Supreme Court might have had different members than it actually has in 2012 no more supports the claim that when we talk about what the Supreme Court does, we are talking about anything other than what its members at the time do, than the fact that the forty-fourth President of the United States might have been someone else shows that when we talk about what the president does, we are talking about someone other than the occupier of that role at this time. We would analyze “the forty-fourth President of the United States might have been someone other than he is” as in [12]. [12] [The x: x is(t*) a forty-fourth President of the United States](it is possible that: [the x′: x′ is(t*) a forty-fourth President of the United States](x′ ≠ x)) In parallel fashion, we analyze “the Supreme Court could have had different members in 2012 than it does” as in [13].

122

C ollective Attitudes and A ctions

[13] [∃t][The X:  X is(t) a Supreme Court][the Y:  Y are ε-members of X in  2012](it is possible that:  [∃t′][The X′:  X′ is(t′) a Supreme Court][the Y′: Y′ are ε-members of X′ in 2012](Y ≠ Y′)) This simply expresses the fact that the description, “the Supreme Court in 2012,” might have picked out a different group of people than it in fact does. 5.4  When Does a Group Count as Doing Something? It sometimes seems to be supposed that collective action is always intentional. But this is incorrect. For example, it is undeniable that we human beings are poisoning the environment through what we do and the way we live. This is not something that any one of us alone could do. But we are not doing it intentionally. It is an unintentional by-product of other things that we are doing. Clearly, then, unintentional collective action is possible, and it is easily explained on my account: it is simply a matter of there being multiple agents of an event. This example is interesting for another reason. It shows that we can do something together though, in contrast to individual action, there is no description of it under which it is intentional. There is nothing we are all doing together intentionally that has the poisoning of the environment as its unintentional consequence. This point is also made nicely by an example of Sara Chant’s in “The Special Composition Question in Action” (2006). Imagine that if someone flips two switches at the same time, an alarm goes off. Suppose he does so, not knowing the consequence. Then he sets off the alarm unintentionally, though he did do something intentionally, namely, flip both switches. Now suppose there to be two people, each of whom flips one of the switches at the same time as the other, with neither knowing the consequence or even what the other is doing. They set off the alarm, albeit unintentionally. However, though each did something intentionally, there is nothing that they need have done intentionally. This is likewise a consequence of my account. It is also a consequence that whenever two people do anything, there is something that they do together, provided that we are generous enough about what counts as an event. Thus, if we count as an event any mereological sum of events, then there is an event of which any group of agents who each do something are the joint agents: each is an agent (in the sense introduced above) of the mereological sum of the events of which they are sole individual agents. I want to address now an objection to this, drawing on an example that Chant gives in the paper I just mentioned. We imagine two boy scouts, in different

Ontology of Collective Action

123

parts of the same town, at the same time each helping a little old lady across the street, neither of them knowing about the other or aware of what the other is doing. In this case, Chant says, we do not accept the collective reading of “they helped two little old ladies across the street.” This shows, Chant argues, that when they each do what they do, they are not participating in a collective action. Consider first the case of two boy scouts who have made a compact with each other to help two little old ladies across the street every day for a month by each helping one little old lady across the street every day. It is day one. Each goes looking for a little old lady to help across the street. At the end of the day, each reports that he found someone to help across the street, and they agree that they have helped two little old ladies across the street on this first day of the project. This seems clearly to be a case in which they jointly help two little old ladies across the street and do so intentionally. Suppose now, instead, that they have just each independently decided to help a little old lady across the street each day for a month. Events proceed as before. Each boy scout is responsible for helping a little old lady across the street, but it is also the case that, in virtue of that, two little old ladies were helped across the street, and that is something that was not brought about by the efforts of only one of the boy scouts. So the helping of two little old ladies across the street, what was done intentionally in the former case, is in this case done by the boy scouts together but unintentionally. Why is there an initial reluctance to accept that the two good boy scouts who act independently jointly aided two little old ladies across the street? I  suggest that it is because we ordinarily think of helping someone across the street as something that is done intentionally. It is difficult, though not impossible, to imagine circumstances in which one could do that without doing it intentionally. So if we hear that someone helped another person across the street, we imagine that it was done intentionally. This suggestion that the action was undertaken intentionally then naturally carries over to plural action sentences whose main verb is “help.” When we consider an utterance of “they helped two little old ladies across the street,” we can interpret it distributively, in which case we imagine each of them intentionally helping a little old lady across the street, or collectively, in which case we imagine them intentionally together helping two little old ladies across the street. Since in the case as initially described it is made clear that there was no joint intention, we feel a reluctance to endorse the truth of the statement on the collective reading because of the pull of this default reading. Yet we can see after reflecting on similar cases that there is a collective reading on which they together help two little old ladies across the street but do not do it together intentionally.

124

C ollective Attitudes and A ctions

5.5  What Are Collective Actions? What do we mean to speak about when we speak of actions? We have in mind what people do. But the question “what did he do?” is answered canonically by an action sentence: he F-ed. It is not like: what did he build? To this, we can answer with a name or a noun phrase, “H.M.S. Endeavour,” “a tree house,” “an empire,” and so on, which specifies a particular object or an object of a certain type. Here “what” stands in for the direct object of the verb “build.” In the question “what did he do?,” “what” stands in for an action verb, and action verbs are not referring terms or noun phrases. Do we need to countenance actions at all then? It is best to consider this question from the standpoint of the event analysis of action sentences. Take [2]‌as an example, repeated here, with its full-dress analysis in [2a]. [2]‌I built a boat. [2a] (∃e)[∃t < t*](∃f)(primitive-agent(I, f, t) and RB(f, e))and [only y = I] agentB(y, e) and building(e) and of(e, a boat)) [2a] shows that the action verb introduces three quantifiers, one over time and two over events. The action verb then relates the agent to a time and to two events, in addition to (in this case) a direct object. Thus, even though we answer the question “what did he do?” with an action sentence, supplying an action verb, say “F,” for the placeholder “what”—“He F-ed”—we may take this to indicate a thing of some type to which he is related as the agent. There are, however, two candidates. The time, of course, is not something one is an agent of, though one is an agent of things at a time. But there are two events that are candidates for being the action. One is the event that is expressed by the action verb “to build” (represented by the variable “e” in [2a]), and the other is the event (which may be complex) of which I am a primitive agent, which results in the event of boat assembly (represented by the variable “f ” in [2a]). We have two choices. We can say that any event of which I am an agent is an action of mine, or we can say that only events of which I am a primitive agent are my actions. The first thing to be said is that this does not appear to be a substantive issue, but rather a terminological one, so long as we are agreed upon the distinction between being an agent of an event primitively, and then being an agent of an event by way of something of which we are primitively an agent bearing the right sort of relation to it. Deciding how to use (or how we in fact use) “action” in relation to this picture does not add any details to it. On the point of usage, I favor the view that “action” counts events of which we are primitive agents on the following grounds. First, we do not treat consequent

Ontology of Collective Action

125

events described as such as actions. Suppose that I kill someone. The consequent event is his dying. But his dying is not an action of mine. Rather, what I did to cause his death—that is, what I did that caused his death—is my action. Second, there is no end to the consequences of events of which we are primitive agents. These continue long after we are dead. But we do not continue to perform actions when we no longer exist. We can say that things Napoleon did are still affecting the present long after his death, but not that Napoleon is still doing things. Third, intuitively the time to which an action verb indexes is the time of action. When we ask when Brutus killed Caesar, the answer is when he stabbed him, not when Caesar dies. If Brutus stabbed Caesar on the Ides of March and Caesar dies on March 22, we do not say that Brutus killed Caesar on the 22nd. He may have been imprisoned or have died himself in the meantime. Intuitively, the time of Brutus’s killing of Caesar is the time of his action, which is the time of his primitive action, the movement of his arm as he stabbed Caesar. This is reflected in [2a] in the fact that the temporal variable has a place in relating the agent to the event of which he is primitive agent, but not to the consequent event. If this is right, then we would expect that the consequent event could occur after the time of utterance, and this is indeed the case. For example, in Hamlet, Laertes says, after having wounded Hamlet with a sword dipped with poison and while dying himself from a poisoned wound, “Hamlet, thou art slain. . . In thee there is not half an hour of life.” Hamlet’s slaying here lies in the past of the speech act, but his death lies in its future. It is easy to think of examples of a similar sort. For example, one may say of the central administration at the university, after they have closed the PhD program to cut costs: “They have destroyed the department because the senior faculty will leave.” The destruction of the department coincides with the leaving of the senior faculty, which lies in the future of the utterance act, though the administration’s action lies in the past of the utterance act. Similarly, a chess player who has just executed a knight fork of his opponent’s queen and king may say correctly: “I have just won the game,” though he is not able to checkmate his opponent without further play. And so on. When we turn to collective agency, the question whether there are collective actions now becomes the question whether, or in what sense, groups can be primitive agents of events. The answer is in one sense “no” and, in another sense, “yes.” If the analysis of collective action given above is correct, it is immediately clear that there is a sense in which groups per se are not primitive agents of any events, because they are not per se agents of any events—not at least so far as the truth conditions of collective action sentences go. So if we think of group action as what the group per se is a primitive agent of (or even an agent in any sense), then there are no group actions, because groups are not agents.

C ollective Attitudes and A ctions

126

Groups as such never figure as the agent argument of the relation expressed by “primitive-agent(x, f, t)” and so do not figure as the agent argument in “agent(x, e, t).” In this sense, the primary sense, we will say, then, that there are no collective actions at all, only individual actions—whether we choose the events of which we are primitive agents as our actions or any events of which we are agents. There remains the question whether there is a close enough analog to a primitive action for a group for it to deserve the honorary title of the group’s “collective action.” Let us consider a team of carpenters building a house. We can represent what the structure of what a group does in cases like these in diagram 1. Here, while each member of the group bears the primitive agent relation to an event, there is no event to which the group bears the primitive agent relation, as we noted above. There is, however, a best candidate for a primitive action of the group in these kinds of cases, namely, the mereological sum of the primitive actions of its members. To establish this, we bring to bear two points. The first is that it is an undeniably useful façon de parler to speak of group agency. Groups are not, of course, per se agents of any events. When we speak of a group as an agent of an event in some particular way, we have in mind that there is an action sentence involving the group that is true on its collective reading. Thus, we may speak about a group being an agent of an event in a derivative sense when all and only members of the group bear the relevant agency relation to the event. In this sense, we may say that the group of carpenters is an agent of the building of a house because all and only those carpenters are direct agents of its building. The second point is that a primitive action is defined as something done, but not by doing anything else. We then get a derivative notion of group action as to what the group does, but not by doing anything else. The team of carpenters builds the house by way of its members doing the various things they do to contribute. The individuals’ primitive actions are not actions of the group. But if we allow their mereological sum as an event, that is an event that is brought about only

Agents Primitive agency relation Primitive actions Mereological sum Consequent event

Diagram 1  The Case of the Carpenters

Ontology of Collective Action

127

by the group as a whole and toward which all and only members of the group bear the agency relation. When the group builds the boat, it is by way of the complex event consisting of the primitive contributions of its members. But there is no event the group is an agent of in the derivative sense (I will suppress the qualification from now on) by which it brings about the mereological sum of the relevant primitive actions of its members. Thus, the mereological sum of the primitive contributions of the group’s members may count in an extended, or secondary, sense as the group’s primitive action. In the case we are considering, it is important to note that what we have identified as the primitive action of the group, though it is an event of which each member is an agent, is not an event of which each is a primitive agent. Rather, each member of the group contributes to it by way of doing something that is a part of it, in the way in which an individual’s clapping is a constitutive part the complex event of an audience’s applauding a performance. We may say, then, that a group performs a primitive action F when its members contribute to its coming about by way of performing primitive actions themselves that are constitutive parts of F. Then we can represent what is required for F to be, in the extended sense, a primitive action of the group G in [14]. [14] [Each x ∈ G](∃t)(∃f)(primitive-agent(x, f, t) and f is a constitutive part of F) and [only y ∈ G](∃t´)(∃f´)(primitive-agent(y, f´, t´) and f´ is a constitutive part of F)) In certain cases, it may be possible to narrow the gap between collective primitive actions and individual primitive actions even more. Consider a hypothetical case, suggested by Paul McNamara, of Siamese twins—identical twins whose bodies are joined in utero, who share control over a shared arm. Let’s imagine that each has independent control over the arm. In this case, there is no reason to say that each does not perform a primitive action when each moves the arm, or, say, clenches the fist, independently of the other. In a case in which they struggle to control the arm, one wishing to do one thing and the other another, it would seem that neither does what he intends, and neither succeeds in performing an action. In a case in which they intend to do the same thing at the same time with the arm, say, clench the fist, and that comes about through their overdetermining it, they are joint agents of the clenching in the sense that both are agents of it in the way they are when they clench the fist independently. Since there is a single event in question, it would appear that they would each be a primitive agent of the same event. We would then have the situation represented in diagram 2. Let us designate the event in question as “C.” We can then say that they are the primitive agents of C both as individuals and as a pair. But we still cannot

C ollective Attitudes and A ctions

128

Agents Primitive agency relation Primitive action

Diagram 2  The Case of the Siamese Twins

say that the pair as such is a primitive agent of C in the sense in which the two individuals are. For to be a primitive agent of an event, one must have a de re intention-in-action directed at it, and while each of the twins has a de re intention-in-action directed at what they are doing, the pair of them, as such, does not, and so the pair does not figure in the primitive agency relation as such. In a variant, we might suppose that they cannot move the arm at all without their willing it together. But this does not affect the conclusion. The only difference is whether they jointly overdetermine the movement of the arm or whether they each make necessary contributions that are only together sufficient. They would each be a primitive agent of it in that case too, but the pair still would be a primitive agent of it only in the secondary sense. The case of the Siamese twins is a special case of [14] in which we trade the twins for G, and C for F, and the event that makes the matrix [15] true is likewise C, and we allow identity as a limiting case of being a constitutive part of a thing. [15] [Each x ∈ the twins](∃t)(primitive-agent(x, f, t) and f is a constitutive part of C)) and [only y ∈ the twins](∃t´)(∃f´)(primitive-agent(y, f´, t´) and f´ is a constitutive part of C)) Here, neither twin is the sole primitive agent of any event, and so neither gets to claim any action as his. But this is captured by the above without there having to appear on the scene any agents other than each of them.12 5.6 Conclusion I have argued that (1) when we assert group action sentences, whether plural or singular, we are asserting that certain events had multiple agents; (2) collective action can be intentional and unintentional; (3) a group may perform an action though it is not intentional under any description; (4) any random group of agents is a group that does something together;

Ontology of Collective Action

129

(5) in a certain sense, there are no collective actions, if we mean by that events of which groups are primitive agents in the way individuals are primitive agents of events; (6) but the mereological sums of events of which the members of a group are primitive agents in standard cases of collective action are reasonable surrogates; and (7) although in some conceptually possible cases in which we press the analogy between primitive actions of individuals and groups, like the Siamese Twin cases, the distance between individual and collective primitive action can be reduced, it cannot be completely eliminated. The picture that emerges of collective agency, understood as the subject matter of collective action sentences, is that talk of groups acting, of group agents, and of group actions is a façon de parler.13 Groups are not per se agents, and they do not per se perform actions. Rather, collective action sentences are true when there are multiple agents of a relevant event in relevant ways. The only actions, strictly speaking, are the actions of individuals—that is, it is only individuals who stand in the agency relation to events. There is nothing objectionable in the vocabulary of group agents and group actions, so long as it is understood as shorthand for there being multiple agents of certain events that stand as primitive relative to further events of which they are all thereby similarly agents. All agency, strictly speaking, is individual agency; all collective action is a matter of there being multiple agents of events, in the first instance of aggregates of individual primitive actions and then of their consequences.

Notes 1. This way of putting the question is important. I am not concerned with whether, as a matter of fact, some of the groups picked out using plural or singular group-referring terms are agents, but with whether the truth of the action sentences that such terms figure in as the subject requires that they be agents. 2. I won’t be developing it in this chapter, but it is clear that the question whether there are genuine group agents is of central importance in considering whom to hold to blame for what groups do, intentionally or unintentionally. If there are only individual agents on the scene, then the blame and responsibility must be distributed over them—which is not necessarily to say divided among them. Two recent collections of papers attest to the contemporary interest in the intersection of issues in the ontology of collective action and collective responsibility (French and Wettstein 2006; May and Tuomela 2007). 3. This puts me at odds with a number of theorists, including Baier (1997), Copp (2006), French (1984), Pettit (2003), Schmitt (2003), Stoutland (1997, 2008), and Tollefsen (2002, 2006). Many more talk of collective agents without making it clear whether they regard such talk as a façon de parler or as literally true. On the

130

C ollective Attitudes and A ctions

individualist side are, inter alia, Bratman (1992, 1993), Miller (2001, 2002), and Searle (1990). None of these authors on either side of the issue approach the topic from the standpoint of the logical form of action sentences. 4. My account of the logical form of plural action sentences is developed in Ludwig (2007). I  provide a sketch of the account in section 2.2 and show in section 2.3 how it can be extended to grammatically singular group action sentences. See also Ludwig (2010) for discussion of the details of the event analysis and its motivation. 5. One should not be misled here by thinking the conditions specified do not secure that we were intending to build a boat. It is compatible with our building a boat that we not have been aiming at building a boat. We might have been intending to build a tree house but got the wrong plans out of the drawer. But even so, when we were done, we would have built a boat, not a tree house, as a result of our mistake. 6. This leaves the question of what distinguishes the intentions of individuals who are participating in group intentional action from the intentions of individuals who are not—that is, it leaves the question of how to analyze we-intentions. I provide a reductive account of we-intentions that builds on this account of the logical form of plural action sentences in Ludwig (2007). I use this to provide, in turn, an analysis of collective intentional behavior. 7. This is not invariably so. The Paris Mob was not an institutionally organized group, though it is picked out using a grammatically singular term. A mob can change its membership over time, does not have its members essentially, can engage in acts (storming the Bastille, rioting) that individuals cannot, and arguably, in some cases act through spokespersons. The account I sketch for action of institutional group agents through changes in membership extends to mobs, crowds, gangs, packs, and other similar groups. The main difference has to do with how we conceive of the relation of an individual to other individuals at a time for them to be members of such a group. Membership in an institutional group involves having a certain status function, the conferring of which on an individual is a matter of a settled practice (such as being appointed by an appropriate official or being selected in an election or paying dues and being inscribed in the membership roll). Being a member of a mob is a matter of intending to participate in a joint action with others characterized by (a) some relation they bear to oneself (being in the vicinity, for example) and by (b) their sharing a goal and the same conception of the group with whom they are acting, i.e., as participating in a collective action (perhaps rather vaguely defined) with a group satisfying some unique description whose various members think the same. 8. Some philosophers seem to have assumed that if there is a type of event that only groups can bring about by their members intentionally doing something together, then there must be an agent over and above the individuals who make up the group (Stoutland 2008, 537–38). But our analysis shows that this is a mistake. Suppose the faculty convened on Friday afternoon. Convenings are essentially intentional collective action types. A  group convenes only if the agents involved each intend to be meeting the others as a part of their coming together. So for an event to be a convening, it has not only to exhibit a certain pattern of behavior involving more than one individual but also to have come about as the result of the individuals who instantiate the pattern each having the intention

Ontology of Collective Action

131

to do so with the others jointly intentionally. But the only agents of the pattern required to instantiate the description are individual agents. To put it in terms of the account here, “they convened on Friday afternoon” has the logical form: [Each x ∈ them](∃e)[∃t < t*](agentC(x, e, t) and [only y = x]agentC(y, e) and convening(e) and on(e, Friday afternoon)). Everything we have said about what is required for convening to take place has to do with there being an event that satisfies “convening(x),” and we explained that in terms of conditions on the agents and their intentions. As the analysis of the logical form of “they convened on Friday afternoon” shows, this is compatible with only individuals standing in the agency relation to the event. There remains the question whether the participatory intentions of individuals who participate in such actions can be analyzed in terms of concepts already at play in individual action. As noted above (endnote 6), I argue that they can in Ludwig (2007). However, the present point— that group agents per se are not required for essentially collective intentional action—is unaffected by the outcome of this further issue. 9. There are various other objections that can be raised to the account that I will not address in this chapter. For example, it may appear that institutional groups may act through the actions of appropriately authorized subgroups, such as spokespersons, though not all of the members of the group make a contribution. I argue elsewhere that a proper understanding of the type of acts involved in these cases and of the authorization relation show that this is not, in fact, a counterexample. See Ludwig (2007, 378–87; 2013) for responses to this and other connected objections. My main goal in this section is to show how the account of the logical form of plural action sentences can be extended smoothly to singular group action sentences compatibly with the points in 1–5 above. 10. This gives rise to a puzzle in the case of “the Supreme Court did not exist before 1779” because the group picked out by “the Supreme Court” on the analysis indicated includes individuals who existed prior to the institutional arrangements necessary for anyone to be appointed to it. Thus, if the group that is the Supreme Court exists when some of its members do, the Supreme Court existed before 1779, only it was not during that period a Supreme Court. Thus, it would seem that “the Supreme Court did not exist before 1779” comes out false rather than, as we might think, true. However, a similar problem attends talk about ordinary objects. If I say “the desk in my office did not exist before last Tuesday,” what is it that I am picking out? The desk would seem to be identical to the collection of its parts, and these existed before they were used in the construction of a desk. Thus, “the desk in my office did not exist before last Tuesday” seems to come out false. We could say the desk is not identical to its parts, but because it occupies the same place and there is no property including being a desk that the parts (taken together) lack that the desk has, it is difficult to make good sense of this (not that philosophers have not tried). I suggest instead that in this case, what we really understand by “the desk in my office did not exist before last Tuesday” is “the desk in my office did not exist before last Tuesday as a desk.” Thus, we would understand this to mean: (∃t)[The X: X is(t) a desk in my office]~[∃t´: t´ < t* & t´ < last Tuesday](X exists(t´) and X is(t´) a desk). I suggest similarly that we read “the Supreme Court did not exist before 1779” as “the Supreme Court

132

C ollective Attitudes and A ctions

did not exist as a supreme court before 1779” and read this as “(∃t)[The X: X is(t) a Supreme Court]~[∃t´: t´ < t* & t´ CC > DD > CD Person B: CD > CC > DD > DC Philosophical friends of cooperation who are also friends of standard rational choice theory try to combine both of these ideas by introducing utility transfers, which model cooperative behavior as altruistic motivation or a common ethos (i.e., a common system of rules and norms as well as a specific feeling of belonging).8 And indeed, it seems to be an attractive idea to solve the problem of cooperation by embedding it within the broader frame of altruistic motivation. But in doing so, we are on the wrong track, although every prisoner’s dilemma can be resolved by assuming the necessary transfer rate of utilities to overcome its format. In fact, there is only one way to universally avoid a conflict between individual optimization and cooperation, and that is to give all utilities involved equal weight. Let us call it the “utilitarian solution.”9 However, if we give our own utilities more weight than those of others, we cannot overcome all PD-situations. They will remain PD-situations even after the transfer. If, conversely, we assign more weight to the utilities of others than to our own, some PD-situations will be transformed into a “saint’s dilemma”: The saint’s dilemma is the reverse format of the prisoner’s dilemma. Being overtly altruistic, the saint ends up in a Pareto-inferior collective action if they individually optimize outcomes. Utilitarians might find this result attractive since it seems to show that there is only one way of overcoming the cooperation dilemma, and this is to act on the basis of utilitarian principles. However, I think that this result can be interpreted in both ways, speaking either in favor of utilitarianism or in favor of a plausible (structural) account of cooperation. In order to be able to be cooperative, individuals do not need to develop a strong altruistic and egalitarian motivation (i.e., giving equal weight to every person’s utilities), because they are able to cooperate directly, without changing their preferences. The utilitarian solution avoids genuine cooperation, but there is no cultural or even genetic mechanism that forces human individuals not to cooperate. On the contrary, there may be genetic and cultural

Structural Rationality and Collective Intentions

215

barriers against a utilitarian motivation.10 Collective action in general amounts to more than the results out of individual optimizing strategies. It is not possible to integrate cooperative rationality easily within the conceptual frame of standard decision and game theory. Optimizing outcomes is the criterion of rationality in decision and in game theory, even if there are considerable differences between these two branches of rational choice.11 Applied to individual actions and strategies, the predicate rational is expressed by a function with two decision-theoretic variables, namely, subjective probability and subjective utility. In standard game theory (in contrast to the Bayesian variant of game theory), optimization becomes a function of the individual’s utilities, that is, outcomes alone. However, in addition, a meta-criterion requires that every player (i.e., every participant in an interaction) has the same knowledge regarding the game format and the rationality criterion, that is, the solutions of the game. This supplement to decision theory singles out the Nash equilibrium as the solution of standard game theory.12 If you translate the game-theoretic format of the prisoner’s dilemma into a decision-theoretic one, there can be no doubt that a rational person chooses D, because D is the dominant strategy (i.e., the strategy that has better results whatever the circumstances). Translated into a decision-theoretic situation, i.e., a situation in which the actor is confronted not with other persons’ strategies but with natural events, dominant strategies are the only rational ones (provided they exist). From this observation, one may deduce that choosing D in a PD-situation amounts to treating the other player like a natural event. One might say that if there is no causal relation between the players’ strategies, one should indeed treat the other players like natural events. Think of an analogy to this decision-theoretic translation. If, for some reason, the probabilities of the circumstances depend on your choice of strategies, the dominant strategy might not be rational. The essential point of my argument is that cooperative action is motivated by specific types of reasons. Let us call them cooperative reasons. Like most reasons, they are constituted by an interpersonal relation. I have a cooperative reason for acting if I want that we, the co-operators, contribute each individually to some collective action that we jointly prefer. This jointly is to be understood collectively, not distributively. This does not deny that the co-operators may not know any better collective action. It merely means that we—as the group of co-operators—have the common normative belief that this collective action is acceptable to all of us. To get a better understanding of this form of collective acceptance, it is useful to look at collective choice theory. Its basic concept is an aggregation function that attributes collective preference relations to a collection of individual utility functions. Aggregations of functions attribute to each collection of individual preference orderings13 that form collective preference orderings. They are called social welfare functions.14

C o l l ec t i v e R at i o na l i t y

216

Cooperation takes place if there is (1)  a cooperative attitude and (2)  a commonly approved collective action and (3)  if the strategy combinations are such that there is an n-person prisoner’s dilemma (possibly with several cooperative strategies, i.e., strategy combinations that are Pareto-superior to the result of individual outcome optimization). All rational forms of preference aggregation are Pareto-inclusive; that is to say, if a collective action is Pareto-superior to another, it is collectively preferred. Pareto-inclusiveness is compatible with many different normative accounts of aggregation. For example, the famous difference principle developed by Rawls in his Theory of Justice (1971) is Pareto-inclusive in this way. Even the principle of majority vote is Pareto-inclusive as is the unanimity-principle; even more, the latter allows collective decisions only in case of Pareto-superiority (note that it is not the case that any Pareto-optimal outcome results in unanimity). Table 9.8  B

A

b1

B2

a1

2  / 1

0  / 0

a2

0  / 0

1  / 2

a1 / b1 and a2 / b2 are each Pareto-optimal collective decisions. However, there is no consensus between A and B about which of these two Pareto-optimal collective decisions should be realized. A prefers a1 / b1 and B prefers a2 / b2. Pareto-optimality therefore does not result in unanimity.

Hence, I conclude that cooperation is enhanced by any generally accepted reasonable rule of aggregation because of its Pareto-inclusiveness. To put it differently: it is a normative consensus shared by all individuals that only cooperative collective actions can be jointly approved of; that is, only cooperation is collectively rational. The impossibility results of collective choice theory (the theorems of Arrow, Gibbard, Satterthwaite, and Sen) and the “breakdowns of rationality” in game theory (Howard Nigel) suggest that there is no universal rule to identify collective rationality. More specifically, there is no universal rule to identify the cooperative actions that can be jointly approved of. It is up to institutional arrangements, social relations, cultural traditions, and moral judgements to choose a cooperative action that can be jointly approved of. Equity-, fairness-,

Structural Rationality and Collective Intentions

217

and efficiency-conditions play an important role in selecting these cooperative collective actions.15 The reorientation of game theory, as suggested by Thomas C.  Schelling,16 proved to be quite successful in its application to international relations. It can also be interpreted as a theory of reducing the plurality of possible cooperative collective actions by introducing the criterion of salience. At first sight, salience is not a normative concept. However, in bargaining processes, salience can be used as a remedy to individual optimization and the instabilities and indeterminacies resulting from it. We should agree on a salient solution because many possible cooperative solutions exist, and none of them are more justified than the other. It is easier to reach a consensus if some cooperative solutions seem somehow outstanding. The normative “ought” of cooperation is thus transferred to the salient solution. We say: “If we both want to cooperate, we should prove it by both consenting to the salient solution.” Egalitarian cooperative collective actions fulfill both criteria, that is, that of equity and that of salience. For example, the division of the cake among five hungry children amounts to an infinite set of Pareto-optimal alternatives. Every division of a cake in five parts is Pareto-optimal if all of the children are sufficiently hungry. The equal division is salient even if the children differ in weight, appetite, or age. It is possible to model the situation as a prisoner’s dilemma. Every child may hope that if he screams louder than the others, he will get a larger part of the cake; some of the children may think that they have priority because of some reason or other. But, probably, they all know that there is only one general consensus possible, which is sharing the cake equally. The D-strategy refers to the conditions in which every child demands more than his equal share, and the C-strategy refers to the condition of behaving civilized, and thus being content with one’s equal share. Cooperation is also often made possible by a consensus on second order rules. The participants each have their own first-order rules of how to evaluate feasible collective actions. Some may place special weight on equity, others on efficiency; and some may combine these two principles in a reasonable way. In many cases, it would be impossible to reach a consensus on the basis of first-order rules because divergent opinions exist on exactly how to evaluate and judge different alternatives. However, the participants may agree on second-order rules, which help to find a cooperative collective action that is acceptable for everyone. For instance, simple majority vote is one of these second-order rules. If we cannot reach a consensus about which of the different feasible cooperative collective actions should be performed, we vote and everybody performs her action as a result of this collective action.

218

C o l l ec t i v e R at i o na l i t y

9.3  Collective Intentions In the previous sections, I argued that a specific type of collective intentionality is indispensable for cooperation: (1) Every (potential) participant must have a cooperative attitude—that is, she must be willing to fulfill her part in an act of performing a cooperative collective action. (2) There must be common knowledge regarding (1), that is, every group member must know that every other member intends to fulfill her part of the collective action. (3) There must exist at least one cooperative collective action (i.e., a combination of individual strategies that has a better result for everyone involved than the result of individual optimization) that is acceptable for all (potential) co-operators—hence, (3) is a common normative belief. (4) There must be common knowledge regarding (3). (5) There must be a consensus that yields one of the feasible cooperative collective actions as acceptable for all. However, as shown in section 9.2, no abstract universal criterion exists to solve the uniqueness problem.17 Individual cooperative action requires a certain detachment from the personal point of view; it requires that I treat my potential co-operators as different from natural events. Cooperative reasons address all potential co-operators. There are no solipsistic reasons. My desires, attitudes, and interests do play a role, but desires are always tied to attitudes and interests of others. In collective actions, the actions of other group members—as well as my own actions—are understood as part of a common practice. To act cooperatively would not make sense (and a fortiori would not be rational) if it did not include an intention to contribute to a collective action. This intention is constituted by common normative and empirical knowledge; it includes mutual expectations that are interpersonally coherent. The kind of collective intentionality that is constitutive of cooperation does not necessarily require a community or a group that is bound together by a specific ethos. Think of a society in which a strong individualism prevents the development of any sense of cultural, social, or ethnic community. If the members of this society are asked which community they belong to, they would answer “none.” If communities are understood as being held together by a common ethos, that is, a common system of rules and norms and a specific feeling of belonging, we still can imagine a society in which there are no communities

Structural Rationality and Collective Intentions

219

of this kind but in which cooperation among individuals nevertheless exists. The basis of this cooperation may be trust, interest, and some common normative beliefs regarding autonomy and fairness. For the sake of the argument, let us suppose that all individuals share these same principles of individual autonomy and fairness. This will enable them to cooperate on the basis of a general cooperative attitude (which we have shown can be rational) as well as a common knowledge regarding the persons and interests involved. Also, the selection of a unique cooperative collective action is included, which fulfills the criteria of individual autonomy and fairness and which is selected—if necessary—by a second-order rule like majority vote. Note that this interpretation is only possible on the basis of my account of structural rationality. The reason for this is that cooperation without community, that is, without group ethos, does not introduce any changes in values or utilities. The participants have only one type of evaluation, which is represented by the game format of the interaction. The only additional assumption one needs is that the persons do not optimize outcomes but accept a specific type of interpersonal structural intentions (i.e., cooperative intentions). These intentions can be described by referring to the respective game format. They cannot be described by referring to the outcomes of each participant’s strategic options. Thus, we drive a wedge between choice and preference because strategy combinations cannot be identified with outcomes anymore. Outcomes retain their original status as representing subjective consequences.18 Preference orderings over outcomes are not mirrored by preference orderings over the respective strategy combinations. In the prisoner’s dilemma case, this means that A may prefer CC to DC, that is, A may act cooperatively if A expects B to act cooperatively, although A prefers to . The latter preference may be revealed if we ask A if he prefers (1), that is, receiving four and B receiving one, or if he prefers receiving three and so does B. The account of structural rationality allows A to answer (1), although he acts cooperatively if he expects B to act cooperatively, that is, he chooses C if he expects B to choose C. Since I define structural reasons as guiding rational actions, I do not need to wait for utility changes in order to render cooperative action rational. In other words, we dismiss the narrow rationality criterion of outcome optimization in order to be able to integrate structural reasons, structural intentions, cooperative action, and collective intentionality into the decision-theoretic conceptual frame. In the previous sections, I  have described cooperative action as an exemplary case of collective intentionality without any reference to group ethos or community values.19 In this case, there is genuine collective intentionality if individual optimization of outcomes would lead to a different result than individual rational behavior. Collective intentionality is constituted by common knowledge—a common cooperative attitude and the selection of an acceptable

220

C o l l ec t i v e R at i o na l i t y

cooperative collective strategy as described above. No transformation process is presupposed that changes values or utilities. This exemplary case can be generalized in the sense that there is no specific linkage between the cooperating partners—there is no WE and THEY; no collective identity of whatever kind is involved.20 Cooperation comes up whenever the interaction is of the respective type and the conditions (1)–(5) of the scheme above are fulfilled. Thus, cooperation in the specific way of collective intentionality comes up whenever the circumstances allow for it. The constitution of groups, communities, and collective identities; the development of a group ethos; and the discrimination between WE and THEY are not necessary in order to render cooperation rational. Even if we alter these conditions and allow for different forms of collective intentionality that are based, for example, on different normative judgments in different groups of agents, we can uphold the assumption that values and utilities remain unchanged. Normative judgments that are constitutive for some groups of co-operators but not for others are exclusively deontological. They refer to the rules that make cooperation possible, that is, the rules that pick out the acceptable cooperative collective action from the set of feasible cooperative collective actions but do not result in changing the evaluation of outcomes. In a way, the agents remain as they are. If subjective probabilities and utilities characterize an agent so long as his beliefs and values are sufficiently coherent, then, by cooperating, structural aspects of rationality are added without changing the agent.

References Arrow, K. 1963. Social Choice and Individual Values. 2nd ed. New York: Wiley. Davidson, D. 1973. “Radical Interpretation.” Dialectica 27: 314–28. Douglas, M. 1986. How Institutions Think. Syracuse: Syracuse University Press. Fleurbaey, M., and A. Trannoy. 2003. “The Impossibility of a Paretian Egalitarian.” Social Choice and Welfare 21: 243–63. Gilbert, M. 2000. Sociality and Responsibility:  New Essays in Plural Subject Theory. Lanham, MD: Rowman & Littlefield. Lewis, D. 1969 Convention: A Philosophical Study. Cambridge: Harvard University Press. ———. 1974. “Radical Interpretation.” Synthese 23: 331–44. Margolis, H. 1982. Selfishness, Altruism, and Rationality:  A  Theory of Social Choice. Chicago: University of Chicago Press. Nagel, T. 1970. The Possibility of Altruism. Princeton: Princeton University Press. Nida-Rümelin, J. 1993. Kritik des Konsequentialismus. München: Oldenbourg. ———. 1994. “Rational Choice. Extensions and Revisions.” Ratio 7: 122–44. ———. 1997a. Economic Rationality and Practical Reason. Dordrecht: Kluwer. ———. 1997b. “Why Consequentialism Fails.” In Contemporary Action Theory, vol. II, edited by Ghita Holmström-Hintikka and Raimo Tuomela, 295–308. Dordrecht: Kluwer.

Structural Rationality and Collective Intentions

221

———. 2001. Strukturelle Rationalität. Stuttgart: Reclam. ———. 2000. “Rationality: Coherence and Structure.” In Rationality, Rules, and Structure, edited by J. Nida-Rümelin and W. Spohn, 1–16. Dordrecht: Kluwer. ———. 2005a. Über menschliche Freiheit. Stuttgart: Reclam. ———. 2005b. “Why Rational Deontological Action optimizes Subjective Value.” Protosociology 21: 182–93. ———. 2011. Verantwortung. Stuttgart: Reclam. Pazner, E. A., and Schmeidler, David 1978. “Egalitarian Equivalent Allocations: A New Concept in Economic Equity.” Quarterly Journal of Economics 92: 671−87. Rawls, J. 1971. Theory of Justice. Cambridge: Harvard University Press. Rescher, N. 1975. Unselfishness. The Role of the Vicarious Affects in Moral Philosophy and Social Theory. Pittsburgh: University of Pittsburgh Press. Savage, L. J. 1954. The Foundations of Statistics. New York: Dover Publications. Scheffler, S. 1982. The Rejection of Consequentialism. A Philosophical Investigation of the Considerations Underlying Rival Moral Conceptions. Oxford: Oxford University Press. Schelling, T. C. 1960. The Strategy of Conflict. Cambridge, MA:  Cambridge University Press. Searle, J. 1969. Speech Acts: An Essay in the Philosophy of Language. Cambridge: Cambridge University Press. Sen, A. K. 1977. “Rational Fools: A Critique of the Behavioral Foundations of Economic Theory.” Philosophy and Public Affairs 6: 317–44. ———. 1986. “Social Choice Theory.” In Handbook of Mathematical Economics, vol. 3, edited by K. J. Arrow and M. D. Intrilligator. Amsterdam: North Holland. ———. 1997. “ Maximization and the Act of Choice.” Econometrica 65: 745–79. Shapiro, J. 2006. “ Genome Informatics: The Role of DANN in Cellular Computations.” Biological Theory 1: 288–301. Spohn, W. 2012. “Reversing 30 Years of Discussion:  Why Causal Decision Theorists Should One-Box.” Synthese 187: 95–122. Tuomela, R. 2007. The Philosophy of Sociality: The Shared Point of View. Oxford: Oxford University Press. Varian, H. 1974. “ Equity, Envy, and Efficiency.” Journal of Economic Theory 9: 63–91.

Notes 1. I introduced the term “structural rationality” in my habilitation thesis (Nida-Rümelin (1993)) in order to describe phenomena of rational agency that are not covered by the consequentialist optimization model. Later on, I extended this account to a general nonconsequentialist theory of practical reason. Cf. Nida-Rümelin (2001), Nida-Rümelin (2005a) and Nida-Rümelin (2011). I  mention these publications in order to make the context explicit, but the argument developed in this paper is self-contained. For a short presentation of the account of structural rationality, cf. Nida-Rümelin (1997b), followed by Nida-Rümelin (2000) and Nida-Rümelin (2005b). In the following, I shall refer exclusively to English publications. 2. Cf. Nida-Rümelin (1997a). 3. I borrow this example from Nagel (1970). 4. The Pareto condition requires that alternatives that are preferred by all people must also be preferred by the collective. Correspondingly, we find a Pareto efficient

222

C o l l ec t i v e R at i o na l i t y

allocation of goods among a group of individuals if it makes at least one individual better off while making no one else in the group worse off. 5. Classical texts dealing with these universal traits of language are Lewis (1969; 1974, 23); Searle (1969, ch. 2.7); Davidson (1973). In the chapter “Communication” of Nida-Rümelin (2001), I argue that the constitutive rules of communication are not to be understood as solving coordination games, as Lewis assumes, but as solving cooperation games. 6. Regarding this point, my views differ from the position held by Tuomela in Tuomela (2007, ch. 7). 7. Cf. Savage (1954). 8. Cf. Tuomela (2007, ch. 7), but much earlier, also Rescher (1975). 9. Cf. Margolis (1982) and Scheffler (1982). 10. Cf. Shapiro (2006). 11. Cf. Nida-Rümelin (1994). 12. Cf. Spohn (2012). 13. Preference relations are defined here in the usual way, i.e., they are reflexive, transitive, and connective. 14. This terminology, however, is misleading, because one of the most interesting applications of collective choice theory is the analysis of political decision making in which political preferences are aggregated. But the latter do not usually represent individual utilities.Classical works in social choice theory are Arrow (1963); Sen (1986). 15. Cf. Varian (1974); Pazner and Schmeidler (1978); Fleurbaey and Trannoy (2003). 16. Schelling (1960). 17. Kantian ethics can be read as incorporating the imperative to respect individual autonomy and refrain from individual optimization if there is a cooperative solution. Two possible readings of the categorical imperative underlie this interpretation. First, the generalization of maxims should be logically possible; second, individuals should be able to desire generalizing their maxims as a principle of general behavior. The second interpretation excludes maxims that are destructive for cooperation because no one desires that all citizens of a society act in a way that destroys cooperative behavior. However, the problem with Kantian ethics is that it has no answer to my (5). Hegelian practical philosophy contains a culturalist and historicist solution of (5), but this results in the loss of normativity and universality. Counterfactual rationality tries to keep as much normativity and universality as possible without losing the grounds of our Lebenswelt. 18. Amartya Sen has developed several arguments that speak in favor of a new decision-theoretic account that “drives a wedge between choice and consequences,” but he never said what type of new decision theory, including game theory and collective choice, should result. The account of structural rationality can be read as an answer to Sen’s Challenge. Cf. Sen (1977, 1997) and Nida-Rümelin (2000). 19. Note that this case is not the same as the I-mode type of group action that Tuomela (2007). 20. This stands in contrast not only to collectivist and institutionalist notions like that of Mary Douglas (cf. Douglas (1986) and other publications of hers) but also to moderate accounts like that of Margaret Gilbert (cf. Gilbert (2000)).

Index

Acceptance 64, 67 collective group  37, 46, 47, 48, 50, 61, 62 and goals  50 in contrast to belief  67 joint 65, 64 Acting as a group member  36 Action collective  48, 112, 113, 155 sentence 112, 114 group 116 plural 113, 116 singular 114 ontology of 112 primitive 126 Andersen, H. 164 Anscombe, G. E. M.  16 Arrow, T.  6, 194, 216 Bacharach, M.  36, 190 Bekkering, H. 21 Block, N. 139 Bratman, M. E.  1–2, 4, 14–16, 61, 151 Broome, J.  198–9 Chant, S. R.  107, 122 Clark, A. 98 Chersi, F. 22 Churchland, P. M.  98 Collective goals 4, 34, 35 classification of  37, 55 satisfaction of 36 Collectivity condition  40, 41, 45 and membership  45, 47 and mutual belief condition  50 Commitment

and collective attitudes  2 joint  2, 62, 78, 82, 85 and membership  47 ontological 139 Consensus normative 210, 216 Copp, D. 5 Cornes, R. 53 Corporate will  51 Dennett, D. C.  98 Deontic logic  197 Discursive dilemma  137, 143 Doki, J. 22 Ernst, Z.  107, 123 Fagan, M. B.  6–7, 170 Feldman, F. 203 Fleurbaey, M. 216 Frankfurt, H. 22 Fodor, J.  100–1 Fox, 173–4 Gallese, V. 21 Gallistel, C. R.  110 Game theory  195, 215, 217 General will  4, 34, 35, 51 Gibbard, A.  188, 216 Gilbert, M.  1–2, 7, 13–5, 19, 62, 64, 66, 68, 70–1, 77–9, 81–5, 88, 164, 155, 175–81, 183 Gibson, J. 17 Giere, R. 164 Gibbard, A.  188, 216 Goals private I-mode  41 pro-group I-mode  41

Index

224

Goals (Cont.) we-mode 39, 40 see also collective Gold, N. 106 Gopnik, A. 100 Goldstone, R. L.  99–100, 108 Graham, G. 22 Group decison 147 intention 14, 150 mind  7, 137, 141 Grush, R. 102 Gureckis, T. M.  99–100, 108 Hakli, R. 36 Horgan, T. 22 Horty, J. 197 Huebner, B.  99, 103 Iacoboni, M. 21 Indispensability argument  137, 138 explanatory version of the  137, 138 practical 138, 141 and the discursive dilemma  147 Intention (Intentionality) collective  1, 2, 3, 14, 15, 218, 220 dynamic theory of collective  3, 13, 16, 18 shared  2, 13, 14, 15, 18, 19 Irrationality 188 Jackson, F. 199 Jacob, P. 17 Jeannerod, M. 17 Jeffrey, R. 192 Jordan, J. S.  21 Kawato, M. 18 Kitcher, P.  163–64 Knoblich, G. 21 Knowledge  163, 164, 175 collective 165 common 15 scientific  6, 164, 166 Kolodny, N.  200–1 Kornhauser, L. 142 Kutz, C. 15

Levy, N. 22 List, C.  5–6, 149 Ludwig, K. 4–5 Margolis, H. 214 McNamara, P.  127, 197 Membership and believing p as a body  87, 88 and collective commitments (acceptance)  47, 50, 81, 181, 183 criteria of 174 different sense of  121 and group action  148 and mutual belief  50 and obligation  180, 181 and primitive agent relation  126, 127 and rational act  187, 194, 196, 197 and scientific knowledge  174 and social constituted relation  117 socially constructed  120 and we-intention  2 and quantifier over members  115 and will  51 Miller, K.  2, 4, 36 Nichols, S. 100 Nida-Rümelin, J.  6–8, 209 Nigel, H. 216 Nisbett, R. 100 Obhi, S. S.  21 Obligation and collective acceptance  37 and collective goals  52 conditional 199 and joint commitments  2 normative web of  180 and rationality  196, 197 Ovarian dogma  173, 174, 175, 181 Pacherie, E.  14, 16–8, 22–3, 26 Pazner, E. A.  216 Pargetter, R. 199 Pettit, P.  5, 6, 8, 138–9, 140, 143–4, 147–9, 151, 154, 156 Piccione, M. 202

Index

Plato 193 Public goods  53 Plural subject account  2, 10, 164, 175, 177, 180 Plurality of good reasons  208 Preference reveale 208 Primitive agent  114, 115, 128 Prisoner dilemma  189, 203, 204, 211, 213 Quinn, W. 201 Rawls, J.  35, 53, 216 Rational choice theory  214 Rationality collective  3, 6, 7, 8, 36, 141, 187, 189, 191, 193, 194 conditional requirement  199 consistency 194, 196 individual and collective  5, 6, 145 obligation 196, 197 of group agent  6 structural  7, 207, 208, 213 Rescher, N. 214 Rolin, K. 164 Roth, A. 6–8 Rousseau, J.-J.,  35, 51–2, 193, 194 Rubinstein, A. 202 Rupert, R.  4, 98–99, 101, 104, 107 Russell, B. 189 Sandler, T. 53 Satterthwaite, M. 216 Savage, L. J.  213 Schweikard, D. 3 Scheffler, S. 214

225

Schelling, T. 216 Schmeidler, D. 216 Schmidt, B. 3 Schmitt, F. 4 Searle, J.  1, 3, 14, 64, 139 Sebanz, N. 21 Sellars, W.  51, 52 Sen, A. K.  216 Shall-operator 52 Shapiro, J. 214 Socializing metaphysics  193 Social welfare function  215 Stich, S. P.  100 Sudgen, R. 106 Supraindividualism 62, 88 The Big Four and the Big Five  2, 5 Bratman’s account  2 Gilbert’s account  2 Searl’s account  2 Tuomela’s account  2 Pettit’s account  5 Theory of justice  216 Theory-theory of mind  100 Tienson, J. 22 Tollefsen, D.  3, 14 Tomberlin, J. 198 Trannoy, A. 216 Tuomela, R.  1–2, 14, 35–6, 46, 214 Varian, H. 216 Weirich, P.  6–7, 187, 202–3 Wilson T.  100, 107 Wolpert, D. M.  18 Wray, B.  164, 178–79